Default Action: directlink
Default Link Follow: nofollow
Default Link Target: newtab
Affiliate Code:
Default Link Color is defined : #555555
How Long Will Burnham Last? Authored by Mark Littlewood via DailySceptic.org, Over the past decade our political system has chewed through and spat out Prime Ministers at a faster rate than Premier League football teams burn through managers. As Andy Burnham enters 10 Downing Street, why would we think there is any reason he will do much better โ in terms of longevity โ than his immediate predecessors? Iโm keeping my eyes peeled for a bookmaker offering odds on Burnham being removed from office before the next election rather than at it. As the Labour Party slowly coalesced around a consensus that Starmer simply had to go there was an oft repeated refrain of, โWe have to get this right first time if we are going to do it, we only have one chance.โ The theory here seems to be that ditching a leader โ however unpopular โ and finding a fresh one is not the sort of thing that the electorate approve of. They have passed their judgement at the ballot box and are usually unimpressed when the powers-that-be decide to second guess that decision. More generally, ditching an incumbent leader against their will gives the impression of chaos and instability. It gives such an impression precisely because chaos and instability are typically what have brought about the decision to make a change. The circumstances around removing a Prime Minister are always different but can be broadly batched into three categories โ they are based on aesthetics, ideology or health. The last two are easier to explain to the public than the first. Margaret Thatcher was removed for ideological reasons โ she was unwilling to retreat on the hugely unpopular poll tax and was also pursuing a European policy that a large chunk of her party disapproved of. On the basis of needing to wholly reset our approach to the EU, David Cameron jumped and Theresa May was pushed. These sorts of switches of leader seem to have a fair chance of working. Indeed, each of Thatcherโs, Cameronโs and Mayโs successors went on to retain office at a General Election. But changing Prime Minister just on the basis that the outgoing leader wasnโt very good and without a clear and explicit shift in an area of policy seems fraught with electoral danger. Callaghan and Brown went on to lose elections having entered office mid-term. Both had different approaches to their predecessor but there wasnโt a single, specific and explicable change in policy. Similarly, the switch from Johnson to Truss and then swiftly onto Sunak didnโt appear to improve the standing of the Conservatives amongst the electorate. In fact, it probably worsened it. Burnham finds himself in the same vague, messy scenario. He didnโt launch a coup against Starmer because he fundamentally disagrees with him on, say, the war in Iran or Net Zero or membership of the ECHR. Instead, itโs simply based on Labourโs abysmal poll ratings, Starmerโs catastrophic approval ratings and the hope that a fresh approach to โcommsโ can somehow revive the partyโs fortunes. Itโs early days (indeed, we are not yet even in the early days of a new administration as I write these words), but the evidence of recent polls suggests little or no Burnham bounce for Labour. The governing party remains stubbornly stuck on only around 20% in the polls. If there was a wave of national enthusiasm โ or even a noticeable trickle โ about Burnham entering Downing Street, one would have expected some sort of uplift in the polls in anticipation of it happening. So far, nothing. Burnhamโs supporters will insist โ or at least hope โ that once we see their man in action as our actual PM the electorate will warm to him. He is certainly a more charismatic politician and a better communicator than Starmer, but Iโm sceptical that this will meaningfully move the dial of public opinion. If our new Prime Minister has a markedly different economic strategy to that of the last two years, Iโve yet to hear it (or at least Iโm yet to understand what the various vague analyses and prescriptions put forward actually amount to). He lamented a โseries of wrong turns in the 1980sโ โ his teenage years โ which, he believes, sent the country on the path to ruin. โPolitical power was centralised and economic power was privatised,โ he said in his coronation speech as Labour leader on Friday. โThe country surrendered control of the essentials, housing, water, energy, transport and left people exposed to higher costs.โ He lumps the Thatcher years and everything that followed, Tory or Labour, as โfour decades of the neoliberalism that began in the 1980s [which] have not been kind to the places that built our partyโ. His solution is โcontrolโ โ more state ownership, more diktats from politicians. All of this suggests to me that we are likely to continue on the pathway of taxes being dialled up, growth being low, debt increasing and public spending ballooning (particularly welfare spending). I suspect if there is any Burnham honeymoon at all, it will be very brief. If this is so, what happens if in a year or so the Labour Parliamentary party is still staring down the barrel of electoral oblivion? Having decided to twist rather than stick when it came to Starmer, might the same apply to Burnham? In such a scenario, it will likely be too late to change policy direction in any dramatic fashion โ the party will be locked in as a government that seeks to tax and spend to the maximum. However, desperate folk in politics are prone to throw the dice. Having set the precedent of throwing the captain overboard because the polling numbers look bleak it can easily become habit-forming. Historically, the Conservatives have had a much greater propensity for regicide than Labour, but thereโs no persuasive reason why the latter couldnโt adopt the same psychopathy. This is why Iโm looking forward to a book opening on Andy Burnhamโs departure date, even if it seems a bit mean-spirited to do so right at the start of his reign. If I can find 3/1 or longer that he will be out by the end of 2028, I will certainly be snapping up those odds. Tyler Durden Mon, 07/20/2026 - 02:00
'Convergence': Secret Labs Already Deployed Lifelike Female Androids While Public Watches Optimus Toy Demos Authored by Madge Waggy, Technology has an unusual habit of revealing itself twice. The first version appears quietly, almost unnoticed, inside research papers, patent filings, laboratory demonstrations and corporate presentations that attract little attention beyond specialists. The second version arrives years later, polished into a consumer product that suddenly convinces everyone the breakthrough happened overnight. History repeatedly favors this illusion. By the time society begins discussing a technological revolution, thousands of engineers have already spent years solving problems the public never realized existed. That gap between discovery and public awareness has become one of the defining characteristics of the twenty-first century, and nowhere is it more apparent than in the race to build machines that no longer resemble machines at all. During the last three years alone, humanoid robotics has advanced at a pace few analysts considered realistic even a decade ago. Tesla continues developing Optimus as a general-purpose humanoid assistant, Figure AI has demonstrated robots capable of performing increasingly complex industrial tasks while collaborating with advanced language models, Boston Dynamics has introduced an entirely electric generation of Atlas after retiring one of the most recognizable research robots ever built, and companies such as Sanctuary AI, Agility Robotics and Apptronik are openly competing to place human-shaped machines inside factories, warehouses and commercial environments. Simultaneously, breakthroughs in computer vision, reinforcement learning, synthetic materials and multimodal artificial intelligence have dramatically reduced the distance between experimental prototypes and practical deployment. None of these developments belong to science fiction anymore. They are unfolding in plain sight, documented through academic publications, investor briefings and live demonstrations watched by millions around the world. Yet an uncomfortable pattern has begun emerging alongside this extraordinary progress. Every major technological leap seems to generate a parallel conversation taking place far from universities and conference halls, unfolding instead across anonymous forums, encrypted chat groups and obscure corners of the internet where speculation often grows faster than evidence. Some theories disappear within hours because they collapse under the weight of obvious contradictions. Others survive for years despite the complete absence of proof, not because they successfully explain reality but because they ask questions that remain surprisingly difficult to dismiss. One recurring idea has become particularly persistent: if companies publicly demonstrate machines capable of walking, reasoning and manipulating objects with human-like dexterity today, what might exist inside research facilities whose work will not become public until five or even ten years from now? No credible evidence has ever answered that question. The absence of evidence, however, has never prevented the internet from trying. Among the countless stories that circulate through speculative technology communities, one in particular continues resurfacing despite never being independently verified. It describes an invitation-only exhibition allegedly organized inside an anonymous research complex where visitors were shown humanoid entities unlike anything previously demonstrated in public. According to archived discussions that periodically reappear online, the event itself was almost deliberately ordinary. There were no theatrical presentations, no dramatic product launches and no executives standing beneath oversized screens announcing the future of artificial intelligence. The building resembled a research facility more than a convention center, its corridors illuminated with clinical precision and stripped of nearly every identifying feature that could reveal who financed or operated the installation. Whatever attracted attention that day was not the architecture. It was what stood behind the glass. The descriptions vary in countless small details while remaining strangely consistent about the central image. Long rows of transparent containment chambers reportedly filled an immense exhibition hall, each enclosing a female humanoid whose appearance challenged the very definition of robotics. Visitors allegedly expected exposed actuators, composite frames or polished metallic joints similar to those displayed by contemporary engineering prototypes. Instead, they found figures possessing skin that reflected light with subtle biological imperfections, naturally distributed hair, nearly imperceptible facial asymmetry and eyes that appeared disturbingly capable of maintaining silent attention. Some accounts insisted they were sophisticated androids. Others argued they were advanced synthetic organisms. A few dismissed the entire story as an elaborate digital art project designed to exploit growing public fascination with artificial intelligence. None of these interpretations has ever been supported by verifiable evidence, yet none has managed to erase the story either. Perhaps that persistence says less about the photographs than it does about the era in which they allegedly appeared. Only a few years ago, generating a convincing human face required painstaking digital artistry. Today, artificial intelligence routinely produces images, voices and video sequences capable of deceiving experienced observers under ordinary viewing conditions. Simultaneously, roboticists are steadily solving challenges once considered decades away from practical implementation. Artificial muscle systems continue improving. Synthetic skin capable of sensing pressure and temperature is advancing through laboratories across Europe, Asia and North America. Neural networks increasingly interpret visual information with remarkable precision, while large language models have transformed natural conversation into something machines perform with astonishing fluency. Each breakthrough, taken individually, appears understandable. Together they create an unsettling possibility that public imagination often races ahead to explore long before reality catches up. The modern conspiracy landscape thrives inside precisely that narrow space between demonstrated capability and undocumented possibility. It rarely invents technology from nothing. Instead, it observes genuine scientific progress, extends every trend several years into the future and asks whether society is already witnessing carefully selected fragments of a much larger picture. Throughout history, classified aviation projects, cryptographic systems and surveillance technologies have all existed years before their official acknowledgment. That historical precedent fuels endless speculation whenever another rapidly advancing field begins transforming the world. Humanoid robotics, perhaps more than any emerging technology today, naturally invites the same questionsโnot because hidden laboratories have been proven to exist, but because technological acceleration has repeatedly surprised even the experts attempting to measure it. THE EXHIBIT THAT ARRIVED BEFORE THE FUTURE Technology rarely introduces itself with a dramatic announcement. More often, it arrives disguised as another research paper, another patent application, another demonstration watched by a few thousand engineers before disappearing beneath the next day's headlines. Looking backward, every technological revolution appears inevitable. Looking forward, it almost always resembles coincidence. That contradiction has become increasingly difficult to ignore over the past several years as artificial intelligence and humanoid robotics have advanced at a pace that even optimistic forecasts struggled to anticipate. What seemed extraordinary in 2020 became commercially viable by 2024, and by 2025 discussions that once belonged exclusively to science fiction had quietly entered boardrooms, government agencies, and manufacturing plants around the world. Unlike previous waves of automation, today's race is no longer centered around machines built exclusively for factories. The objective has shifted toward creating systems capable of operating naturally inside environments originally designed for people. That distinction changes everything. Companies such as Tesla continue refiningย Optimus, Figure AI has demonstrated humanoid robots performing industrial tasks alongside advanced language models, Boston Dynamics has transitioned to an entirely electric version ofย Atlas, while Sanctuary AI, Apptronik, and Agility Robotics are openly competing to deploy human-shaped machines into warehouses, logistics centers, hospitals, and commercial environments. None of these developments are hidden behind classified documents. They are documented through public demonstrations, engineering conferences, investor presentations, and peer-reviewed research. The world is not waiting for humanoid robots to arrive. They have already arrived; the only remaining question concernsย how quickly they become ordinary. ย What makes this timeline remarkable is not any single breakthrough, but the compression between them. Technologies that previously evolved over decades are now progressing within months. Computing power continues expanding, neural networks become more efficient with every iteration, synthetic materials increasingly mimic biological tissue, and robotic dexterity improves through reinforcement learning systems trained inside simulated environments before ever touching the physical world. Several research groups are now experimenting withย electronic skinย capable of detecting pressure, temperature and texture, while others are developing artificial muscle fibers designed to reproduce the flexibility of biological movement. Independently, each achievement appears incremental. Viewed together, they begin resembling pieces of a much larger puzzle whose final image remains frustratingly incomplete. Throughout history, transformative technologies have almost always existed in limited forms before becoming public knowledge. Stealth aircraft remained classified long before their official acknowledgment. Modern cryptographic systems evolved inside government programs years before entering consumer electronics. Satellite reconnaissance, advanced computing and even the internet itself all followed similar trajectories, transitioning gradually from restricted environments into everyday life. None of this proves that humanoid robotics follows the same path, yet it explains why discussions surrounding undisclosed research facilities continue appearing wherever technological acceleration outpaces public understanding. Among the countless stories circulating across futurist forums, one recurring narrative refuses to disappear despite the complete absence of verifiable evidence. According to these accounts, anonymous images allegedly depicting a private exhibition began surfacing shortly after several major robotics announcements captured international attention. The photographs themselves were unremarkable at first glance. There were no dramatic explosions of light, no cinematic laboratories filled with exposed machinery and certainly no science-fiction aesthetic designed to impress an audience. Instead, they portrayed an environment whose greatest source of discomfort came from its overwhelming normality. Clean architectural lines, carefully controlled lighting, visitors quietly observing transparent containment chambers, and inside those chambers, female humanoids whose appearance blurred the distinction between biological life and engineered design so effectively that many viewers struggled to identify where one ended and the other supposedly began. What fascinated online communities was not the imagery itself but the possibility it represented. If the photographs were fabricated, they reflected an extraordinary understanding of contemporary robotics and human anatomy. If they depicted an elaborate artistic installation, the creators had achieved precisely the emotional response they intended. Yet ifโpurely as a matter of speculationโthey represented a glimpse into technology more advanced than anything publicly demonstrated, they suggested something profoundly unsettling about the pace at which artificial intelligence and embodied robotics might already be evolving beyond public awareness. WHEN MACHINES STOP LOOKING LIKE MACHINES Perhaps the most significant shift occurring today has little to do with processing power or mechanical engineering. It concernsย perception. Early industrial robots never attempted to resemble humans because efficiency mattered more than familiarity. Modern humanoid research pursues the opposite objective. Engineers increasingly recognize that machines designed to work alongside people benefit from recognizable gestures, natural movement and intuitive communication. Artificial intelligence no longer exists solely inside software; it is gradually acquiring a physical presence capable of navigating environments originally built for biological life. That transition introduces questions extending far beyond engineering specifications. It touches psychology, ethics, economics and identity itself. Researchers studying theย uncanny valleyย have documented for decades that human beings respond differently to machines once they become almostโbut not entirelyโindistinguishable from living people. Small imperfections that might otherwise pass unnoticed suddenly become psychologically significant. A smile held a fraction too long. Eyes that maintain uninterrupted focus without natural micro-adjustments. Facial muscles moving with perfect synchronization rather than subtle asymmetry. These details rarely appear frightening in isolation, yet together they produce an instinctive discomfort difficult to explain rationally. As synthetic materials improve and artificial intelligence becomes increasingly sophisticated, the boundary responsible for that sensation continues narrowing. Whether humanity ultimately embraces lifelike humanoids or resists them remains impossible to predict. What can be stated with confidence is that the underlying technology continues advancing regardless of public opinion. Investment in embodied AI has accelerated dramatically, governments have begun drafting regulatory frameworks for autonomous systems, and major technology companies increasingly describe robotics as one of the next defining industries of the coming decade. Against that backdrop, it becomes easier to understand why fictional stories about hidden laboratories, invitation-only exhibitions and technologies existing several years ahead of public demonstrations continue capturing the imagination of millions.ย BEYOND THE GLASS Whether the exhibition ever existed eventually became the least interesting question. The discussions that followed took on a life of their own, gradually shifting away from anonymous photographs and toward something far more unsettling. Engineers began debating theoretical manufacturing limits. Neuroscientists questioned whether synthetic cognition would eventually require emotions rather than simply simulating them. Military analysts speculated about autonomous decision-making systems, while ethicists found themselves asking an entirely different question: if a machine could imitate every observable characteristic of human behavior, what objective measurement would still separate the creator from the creation? Within the universe surrounding the alleged exhibition, researchers supposedly referred to the project using an unusual expression that later appeared across several archived discussion boards:ย "Convergence."ย The name carried no technical explanation, only a philosophical one. According to the mythology that slowly developed online, the objective had never been to construct machines capable of replacing human labor. That milestone had already become commercially achievable through ordinary automation. The real ambition was allegedly something considerably more difficultโto construct artificial beings capable of existing inside society without ever being recognized as artificial in the first place. The concept itself had already escaped. The documents discussed by online communities described laboratories unlike conventional research facilities. Engineers supposedly worked alongside psychologists, behavioral scientists, linguists, neurologists and artists, each responsible for solving a different aspect of the same impossible equation. Mechanical movement alone could never convince an observer they were looking at genuine life. Human beings unconsciously detect thousands of microscopic behavioral patterns every hour without realizing it. Eye contact lasts for predictable intervals. Facial muscles contract unevenly. Breathing changes with emotion. Tiny pauses interrupt ordinary conversation. Even silence possesses rhythm.ย An anonymous character allegedly summarized the problem in a single sentence that became strangely popular among the hidden communities based on these controversial topics: "People don't recognize humanity because of perfection. They recognize it because perfection never survives long enough to become human." THE FINAL PROTOTYPE WAS NEVER A MACHINE As years passed, technological progress accelerated alongside real industry developments. Every genuine breakthrough in artificial intelligence strengthened public conviction regarding the rapid pace of innovation. Public demonstrations showed humanoid robots becoming smoother, quieter, and more capable each year. Language models learned to hold increasingly natural conversations. Synthetic voices lost their unmistakable mechanical cadence. Artificial vision systems improved beyond what many experts believed possible only a few years earlier. Observers interpreted every public announcement as confirmation that society was witnessing technologies whose foundational frameworks had already been quietly developed over the preceding years. That progression reflected a psychological phenomenon repeatedly observed throughout technological history. Human imagination rarely invents futures from nothing. Instead, it extends visible trends until they become predictable realities. Steam engines became locomotives before becoming intercontinental railways. Primitive computers became smartphones through thousands of incremental improvements rather than one miraculous discovery. In much the same way, the Convergence Program mapped a future assembled not through impossible inventions, but through the gradual integration of disciplines already advancing today. According to unpublished data, the exhibition itself was never intended to impress investors or government officials. It served another purpose entirely. Visitors unknowingly became participants in a behavioral experiment. Cameras hidden throughout the exhibition allegedly measured eye movement, hesitation, emotional response, interpersonal distance and unconscious reactions as guests walked past each observation chamber. Every moment of curiosity became data. Every expression of uncertainty refined the next generation of synthetic behavior. The humanoids behind the glass were not being evaluated. The observers were. That single twist transformed the entire legend from an ordinary tale into something psychologically far more disturbing. It suggested that humanity had misunderstood the experiment from the beginning.ย This ongoing focus on rapid technological evolution reflects a broader recognition that modern civilization is crossing a significant historical threshold with profound long-term implications. Artificial intelligence is actively transforming key sectors, including education, medicine, finance, manufacturing, and scientific research. Concurrently, humanoid robotics continues to advance at an accelerated pace, while brain-computer interfaces are successfully progressing through human clinical trials. Furthermore, synthetic materials are increasingly replicating the functional properties of biological tissue. While each of these advancements represents an isolated milestone, their combined integration shapes a highly complex and rapidly changing socio-technological landscape. Tyler Durden Sun, 07/19/2026 - 23:20
A Deep Dive Inside Kimi K3, And All Other Chinese AI Models: The Definitive China LLM Primer We were lucky enough to read the tea leaves ahead of the "frontier", so to speak, and conclude that various "open" models out of China would soon be the biggest talking point - not to mention the market's fulcrum catalyst. See for example: June 19:ย Answering The "Trillion Dollar Question": Are China's AI Models A Better Value Than US Models June 22:ย Deep Seek 2.0? China's GLM 5.2 Model Takes AI World By Storm With Stunning Mix Of Capabilities & Price More recently, last week we summarized virtually all of China's AI Models in "The Definitive LLM Primer" (July 11), which we republished below for our readers' convenience, yet in the fast-paced world of AI, even that article is now woefully out of data due to the recent arrival of Moonshot's open-sourced Kimi K3 (July 16), which has not only taken the AI world by storm, but promptly sparked a momentum meltdown amid growing fears that Chinese LLMs are catching up too fast to frontier US models (something we warned about one month ago here). The reason for the jarring market reaction is that Kimi K3 is viewed as being on par, if not better, than most leading US frontier models. Big news: Kimi-K3 by @Kimi_Moonshot is now #1 in the Frontend Code Arena with 1679 pts, surpassing Claude Fable 5. This is a 17-place jump from Kimi-k2.6 (#18 -> #1). In Frontend, Kimi-K3 ranked #1 in 6 of 7 domains: Brand & Marketing, Reference-Based Design, Data & Analytics,โฆ https://t.co/YDN3BufGkC pic.twitter.com/Oa6teaQnWp โ Arena.ai (@arena) July 16, 2026 And while there is rampant debate whether this performance is accurate or gamed to beat specific benchmarks, the bigger issue is that China is now clearly developing stunning(ly cheap) open-sourced models which are on par with much more expensive US models, which in itself renders the entire ROI calculus behind trillions in AI capex spending null and void, because if China can achieve 98% what the US has done with a fraction of the capex... China is now leading the AI arms race after spending a tenth what the US did https://t.co/CkPbXzffwi pic.twitter.com/sJOwU0KZhp โ zerohedge (@zerohedge) July 17, 2026 ... then all those trillions already allocated to AI capex are nothing more than sunk costs, as end markets will quickly pivot to what is cheapest, since in most cases it is also on par with what is best.ย So what happened for those who went on vacation early last week and are stunning how everything has changed? Well, as Ronald Keung, the Goldman strategist who wrote the original Chinese LLM primer said late on Friday, we have gone "from cost efficiency (DeepSeek), rise in model intelligence (GLM) to new frontier/pricing power at Kimi K3." Below we excerpt from his latest note (available to pro subs): On July 17, the Kimi K3 open-weight model was released with 2.8 trillion parameters, and has set a new frontier in coding and certain agentic capabilities globally, as per Arena.ai coding rank and Artificial Analysis Intelligence score. Accordingly, Goldman notes K3โs pricing was set at a new high amongst Chinese models, US$2.3 per 1M tokens (blended) vs. Qwen3.7 Max of US$1.4/Zhipuโs GLM5.2 of US$0.9/MiniMaxโs M3 of US$0.22/DeepSeekโs V4 Pro of US$0.18, being still below global SOTA models. As highlighted in our recent China LLM primer, Chinaโs AI open-source/open-weight models are reaching a critical point of intelligence performance for global proliferation. Amongst signposts into 2H 2026, we have anticipated that competition could intensify in the high-end coding segment via coding data flywheel and scaling (to larger models, up to 2-5 trillions parameter sizes). The share price reaction of Knowledge Atlas/Zhipu (where Goldman recently initiated at Neutral, -28% on July 17) and MiniMax (Buy-rated, -16% on July 17) have been due to concerns around Chinese AI model competition and who the potential long-term winners will be, given the competitive landscape/sustainability of model company leadership remains highly dynamic. Here, Goldman repeats that Independent AI model companies stand out in its Competitive Positioning framework (see full report for more). The bank also notes the positive read from Chinese President Xiโs comments at the opening ceremony of World Artificial Intelligence Conference (WAIC) in Shanghai held last Friday, drawing parallels of AIโs societal impact to the invention of electricity and steam engines, and in offering Chinaโs technology infrastructure to developing nations with its continued open approach, yet cautioned on the need for human oversight/controls and risk mitigation. What to watch out for from here?ย Harness/agentic applications: We expect China AI model companies to increasingly focus on positioning their harness/agentic applications as key entry points, especially in coding (e.g. Zhipuโs ZCode, Tencentโs Workbuddy, and Alibabaโs Qoder which are aggregate platforms that support the full suite of AI models) as model companies attempt to close the loop in capturing more real-life coding and agentic data for scaling. Co-work and industry expert agentic products could be the next priorities. Multiple large parameter high-end coding model launches and potentiallyย stricter access to the most advanced Chinese AI models outside of China: After Kimi K3 launch, we anticipate for further new Chinese AI model launches over 2H26 (Zhipu GLM, Alibaba Qwen, MiniMax M3 Pro and more) with significantly larger total parameter sizes of 2-5 trillion across Chinese AI model players. Coding/programming segment competition will remain intense. The addition of multi-modal/visual understanding will also be the next upgrades amongst Chinese AI foundation models. Continued suppressed API pricing for the lower end agentic-focused segment:ย Goldman expects API pricing and therefore gross margins to remain under pressure at the lower-end pricing segment (around US$0.1-0.2 per 1M tokens) into the second half, as Chinaโs AI model players have significant cash buffers post-fund raising to subsidize competitive pricing. As a result, we see financial strength as one of the three most important metrics (alongside pricing power and cost efficiencies) in assessing a Chinese AI model company. Multi-modal/video-generation models to see further ARR ramp-up from global adoption: Goldman expects continued healthy industry pricing and gross margins within video generation (unlike foundation text) where key players ByteDanceโs SeeDance (at reportedly 70% gross margins, latest US$2bn ARR run-rate), Kuaishouโs (Buy) Kling and MiniMaxโs (Buy) Hailuo/upcoming H3 models to enjoy healthy growth over 2H 2026 amid new functionality breakthroughs (combination of video-generation with LLM) and tight computing resources where demand significantly outpaces capacity. Expect increasing domestic ASICs supply, any stricter access to Chinaโsย future most frontier models for overseas markets, Western marketsโ policies on China models, and access to high-end computing in model training as key swing factors/risks to our China AI model token/revenue growth trajectory. Goldman's key ideas/stock picks Within Cloud & Data Centers (the bank's top preferred sub-sector within China Internet),ย GS continues to highlight key ideas (Alibaba, GDS, VNET, and Kingsoft Cloud) on the back of higher AI hyperscaler capex spending into 2H 2026. Within AI models, Keung highlights MiniMax on upside skewed risk-reward. Key swingย factors for MiniMax to improve its overall competitive positioning will hinge on its pricing power following its recent completed fund-raising that has strengthened financial strength. Expect its H3 video generation model performance (imminent launch in a more favorable industry landscape vs. text models) and time-to-marketย for its next M3 updates (focusing on further coding intelligence level uplift from post-training/reinforcement learning, we estimate M3 update over July-Aug, and a larger parameter size M3 Pro model later in 2H 2026) will be the next key drivers. Yet what may be the clearest signal yet just how competitive with the US Chinese LLMs have become comes from none other than OpenAI's "head of strategic futures" (presumable that title refers to whoever can demand protectionism the loudest) who essentially is begging for protectionism against open Chinese LLMs... Some observations on Kimi: 1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it alsoโฆ โ Dean W. Ball (@deanwball) July 17, 2026 ... and the scathing retorts from Trump's (former?) AI Tsar David Sacks, who responds that "K3 just fixed 15 critical security bugs that Codex and Fable refused because of โcyber guardrails.โ Thereโs no reason to limit American models on tasks that Chinese models handle without issue. Weโre only making ourselves less competitive." And then again, here to point out that Chinese LLMs are already more useful to some programmers than the best of the (very expensive) best Anthropic and OpenAI have the offer, to wit: "Hereโs another example: Hugging Face tried using American frontier models to analyze an AI-powered cyber attack. But the guardrails blocked requests containing real exploit payloads so they switched to GLM 5.2 running locally. The guardrails actually impaired defensive security." More in the full Goldman China LLM update focusing on Kimi K3, available to pro subs.ย Meanwhile, for those who missed our original report on Chinese LLMsย from July 11, we republish it below in its entirety.ย * * *ย Three weeks ago, we attempted a lengthy answer of the "trillion dollar question" namely are Chinese AI models a better value than US modelsย and, using extensive research from UBS, concluded that at almost 95% of the capability (and rising) and just 10% of the cost, the answer was a resounding yes. Source: UBSSource: UBS Fast forward to today when Goldman analyst Ronald Keung also addressed the $64 trillion elephant in the room, and publishedย a 50-page China AI models LLM primer (available to pro subs), in which he agrees with our conclusion, namely that "China's AI open-source/open-weight models are reaching a critical point of intelligence performance vs. global proprietary models, with a significant ramp up in domestic enterprise & global SME adoption that will enable a positive data flywheel of further model improvement."ย In the report, Goldman evaluates: How these models achieve such performance at low costs/tight computing resources; Why they pursue an open-source/open-weight approach and how they monetize; What the key addressable markets are, as global enterprises shift from 'token-maxxing' to ROI-first, where Goldman highlights two favored ARR quadrants for Chinese models; and Who the potential long-term winners will be under the bank's Competitive Positioning framework. In keeping with the recent newsflow, Ronald notes that ongoing politicization of AI, namely any stricter access to China's future 'most frontier' models for overseas markets, Western markets' policies on China models, and access to high-end computing in model training, are three key swing factors/risks to the bank's China AI model token/revenue growth trajectory.ย Additionally, the Goldman strategist introduces his Chinese AI model Competitive Positioning framework based on pricing power, cost advantage and financial strength, overlaying with token scale and market share progressions, and identify Knowledge Atlas (Zhipu, initiation) and DeepSeek (private) as the strongest positioned in foundation models, andย ย Bytedance (private) in multi-modal. Before we get into the weeds, excerpted from the gull Goldman report, lets start with a visual summary of China's AI Model and Hyperscaler Ecosystem... ... andย a Breakdown of China's open-source models unit economics today and path to profitability Which brings us to the core overarching theme of Chinese AI development, from cheap to smart, or as Goldman's Ronald Keungย puts it:ย From DeepSeekโs moment last year (on cost efficiency) to Zhipuโs GLM moment this year (on model intelligence). Chinaโs AI open-source/open-weight models are reaching a critical point of intelligence performance vs. global proprietary models, with a significant ramp up in domestic enterprise adoption and a proliferation of global consumer and SME demand. In this report, Goldman evaluates: How these models achieve such performance at low costs/tight computing resources; Why they pursue an open source/open weight approach and how they monetize; What the key addressable markets are, where the bank highlights two favored ARR quadrants and risk factors; and Who the potential long-term winners will be under Goldman's Competitive Positioning framework. Chinese models are reaching a critical โgood enoughโ stage for agentic tasks/specific coding scenarios, and rising fragmentation in Chinaโs AI model landscape (but the strong will get stronger). While pricing power remains strong for models with frontier performance/multi-modal, the lower-end segment is in a price war; nevertheless, Agentic AI is driving explosive demand for these value-for-money models at the lower-end. Access to computing will be a swing factor, where US/China regulations, balance sheet and inference efficiency are key. Accordingly, the bank introduces its Chinese AI model Competitive Positioning framework based on pricing power, cost advantage and financial strength, overlaying with token scale and market share progressions, and identify several companies its views as the strongest positioned in both foundation models and in multi-modal. How do Chinese models achieve competitive performance at low costs/tight computing resources? By leveraging smaller-sized parameter models for equivalent benchmark performance, and Mixture-of-Expert and new architectural innovations. Chinese AI models are bifurcating into a two-tiered market where performance and time to market are key to pricing power, thereby leading to two โARR maximizingโ quadrants based on token adoption and pricing. A positive flywheel is taking effect for top Chinese AI models driven by increasing actual real world coding adoption, and reducing their reliance on model distillation practices. Why are Chinese models pursuing an open source/open weight approach, and ways to monetize? Open source allows for greater flexibility in model training/deployment, and allows for the widest adoption and an open community. Open-source modelsโ disclosed ARRs are likely understating total deployment and revenue potentials, and Goldman expects more shifts to open weight (with Community License, i.e., commercial terms if for commercial use) among Chinese AI models down the road. What are the key addressable markets, domestically and internationally, and key risks? The bank highlights two favored ARR quadrants and estimate China/China AI models market to see token growth of 25X by 2030E; the coding landscapeย to consolidate while the agentic/low-end segment could remain fragmented. International (going global) should be a key upside, with the potential for higher pricing and global proliferation, especially in non-US markets as global enterprises increasingly pivot from a token-maxxing to a ROI-first model that prioritizes clear task boundaries, number of agents per day, back-end process automation and actual output over pure computational token volume. Key risk factors are market access/anti-distillation and regulations, including any stricter access to Chinaโs future most frontier models for overseas markets, access to highest-end leased computing equipment (used for training), market access to western markets, further restriction lists/entity list designations (but this could be positive for Chinaโs path to further AI self-sufficiency across software/CPU/ASICs), and competition from SLM/threat from AI architectures. Who are best positioned to be the long term winners within Chinaโs AI model companies? Goldman expects players with the largest ARR scale with a gross margin advantage + financial strength to be the long-term winners. The bank highlights independent AI model companies mostly stand out in its Competitive Positioning framework in pricing power + cost advantage (in aggregate represent over US$200bn in implied valuations, based on latest market cap/funding rounds), where Zhipu and DeepSeek are the most strongly positioned in text based foundation models, while ByteDance leads in multi-modal capabilities. Signposts for 2H 2026 Harness/agentic applications: China AI model companies will increasingly focus on positioning their harness/agentic applications as key entry points, especially in coding (e.g. Zhipuโs ZCode, Tencentโs Workbuddy, and Alibabaโs Qoder which are aggregate platforms that support the full suite of AI models) as model companies attempt to close the loop in capturing more real-life coding and agentic data for scaling. Co-work and industry expert agentic products could be the next priorities. Enterprises will increasingly focus on overall cost per task instead of headline pricing per token, with an openness to using different models (multiple-models approach). Multiple large parameter high-end coding model launches and potentially stricter access to the most advancedย Chinese AI models outside of China: Multiple new Chinese AI model launches are expected over 2H26 with significantly larger total parameter sizes of 2-5 trillion across Chinese AI model players. Coding/programming segment competition will intensify as Chinese models try to challenge Zhipu GLMโs leadership via training on high-quality real-life coding data (where available) and scaling to larger parameter model sizes. There may be a potential shift from an open-source to an open-weight approach for best-performing models (i.e. from free-for-all use cases to requiring revenue sharing/a take rate for commercial use). The addition of multi-modal/visual understanding will be the next upgrades amongst Chinese AI foundation models (e.g. for GLM, DeepSeek), while MiniMax M3 already excels in these given the multi-modal focus of MiniMax from the start. That said, press reports on potential future restrictions on overseas access to Chinaโs most advanced AI models (both closed and open source if at the frontier level) as cutting-edge artificial intelligence is increasingly being seen as a critical national asset. Continued suppressed API pricing for the lower end agentic-focused segment: While DeepSeek recently announced an increase in peak-hour pricing from mid-July, API pricing and therefore gross margins should remain under pressure at the lower-end pricing segment (around US$0.1-0.2 per 1M tokens) into the second half, as Chinaโs AI model players have significant cash buffers post-fund raising to subsidize competitive pricing at zero/negative gross margins in the near term. As a result, financial strength will be one of the three most important metrics (alongside pricing power and cost efficiencies) in assessing a Chinese AI model company, where cash on hand, net cash as % of assets and valuation multiples will be the critical financial strength metrics for long-term success. This said, Goldman is Buy-rated on MiniMax as the company stands out on cost efficiency/cost advantage metrics under a Competitive Positioning framework. With its M3 model well positioned in the favored ARR maximizing quadrant (attractive pricing + high token volumes), alongside its discounted valuation at 13X P/2026E year-end ARR (vs. China/global peers which command multiples several times higher at similar ARR stage), risk-reward is skewed to the upside (Goldmanb reiterates its Buy rating of the stock). Key swing factors for MiniMax to improve its overall competitive positioning will hinge on its pricing power and financial strength. Its H3 video generation model performance (imminent launch in a more favorable industry landscape vs. text models) and time-to-market for its next M3 updates (focusing on further coding intelligence level uplift from post-training/reinforcement learning, estimated over July-Aug, and a larger parameter size M3 model later in 2H 2026) will be the next key drivers. Multi-modal/video-generation models to see further ARR ramp-up from global adoption: Goldman expects continued healthy industry pricing and gross margins within video generation (unlike foundation text) where key players ByteDanceโs SeeDance, Kuaishouโs (Buy) Kling and MiniMaxโs Hailuo/upcoming H3 models to enjoy healthy growth over 2H 2026 amid new functionality breakthroughs (combination of video-generation with LLM) and tight computing resources where demand significantly outpaces capacity. According to China news reports like LatePost and 36Kr, ByteDanceโs Seedance gross margins have been at a healthy 70% at its latest US$2bn+ ARR run-rate. The bank's strategists also expect increasing domestic ASICs supply, tighter access to overseas computing resources and potential marketย access limitations (could mirror TikTokโs trajectory where rapid expansion in western markets was followed by more regulations/focus on ensuring data security where computing will have to be conducted within local jurisdictions). Goldman's assessment of AI model companies: ARR scale x gross margin advantage + financial strength Largest ARR scale (Token scale x pricing power) Gross margin advantage (Training & Inference efficiency, Technology) Financial strength (Balance sheet, Access to computing) Accordingly, the bank lays out its Competitive Positioning framework for AI model players based on quantifiable metrics, 1) Pricing Power (amongst which, based on Time-to-market of model launch, Arena score based on actual usage cases and pricingย level), 2) Cost advantage (based on token volume scale, throughput/cache hit rate, parameter sizes/activation ratio and our estimate of inference gross margins), and 3) Financial strength (based on cash on-hand, net cash as % of assets, and valuation multiples). Next, an overview of competitive analysis for key LLM labsโ positioningย The bank identifies two favorable ARR quadrants (in maximizing ARR) which is a combination of maximizing token volumes and/or pricing level Comparing Chinaโs key LLM players Decoding Chinaโs AI model tokens & our forecasts on token/revenue share How do Chinese models achieve competitive performance at low costs/tight computing resources Smaller-sized parameters models for equivalent benchmark performances, Mixture-of-Expert and new architectural innovations As readers of our Chinese LLM primer may recall, compared with a year ago, Chinese modelsโ coding and agentic capabilities have reached a critical level in being able to complete more coding and autonomous agentic tasks with higher success rates as context windows have been expanded to 1 million tokens. The smaller parameter model sizes of Chinese models (spanning from 200bn to 1.6T parameters, at 2-10% of leading SOTA models, due to constrained access to high-end computing), and highly efficient architectures (MoE, Sparse Attention, OCR etc., at 3-5% activated parameters only vs. total parameter sizes) all contribute to the much lower training and inference costs for Chinese models vs. leading US models. Goldman attributes the recent step improvement of Chinese models in coding according to Arena.ai to data curation, distillation techniques and reinforcement learning post training, despite their relatively small parameter model sizes (1.6T for DeepSeek V4 Pro, 0.7T for Zhipuโs GLM5.2 and 0.4T for MiniMaxโs M3). On June 27, DeepSeek introduced DSpark, a speculative decoding framework that makes existing DeepSeek-V4 models serve faster. DSpark has already been deployed in DeepSeek-V4 Flash / Pro online serving, improving per-user DeepSeek-V4 generation 60-85% faster on V4-Flash and 57-78% faster on V4 Pro without changing the modelโs weights or output quality. Chinese AI models bifurcating into two-tiered market (where performance and time to market are key to pricing power), with two โARR maximisingโ quadrants based on token adoption and pricing Goldman is seeing pricing power for the highest performing Chinese AI models, e.g. Zhipuโs GLM5.2 model and Alibabaโs Qwen3.7 Max models at around US$1 per blended 1M tokens, at 5X that of low-end Chinese AI models. The reported tighter US processes in allowing most SOTA model access has also opened new arenas for Chinaโs top performing coding models for enterprises and SMEs. Smaller parameter and activated ratios allow Chinaโs top performing models to be priced at US$1, 10-25% vs. US SOTA models at US$4-8 per blended 1M tokens, while generating double digit 10-20% gross margins (GSe) that are lower than global SOTA models due to relatively lower pricing power.ย In the lower-end segment, agentic focused models are priced at US$0.06-0.2 per blended 1M tokens, which are enabling these models to tap into new global TAMs for price sensitive SMEs and one-man companies. MiniMax generates 60-70% revenues from overseas.ย Note that DeepSeek announced that its V4 official version is set for launch in mid-July, alongside the introduction of peak/off-peak API pricing to better allocate resources and improve service stability. V4 Pro/Flash non-peak pricing remains unchanged, while peak hour (9am-12pm/2pm-6pm China time) will be charged at 2X non-peak rates, implying blendedย pricing of US$0.35/US$0.12 per 1M tokens due to strong Chinese AI model demand that is increasingly causing significant compute tightness in work/productivity scenarios. Positive flywheel is taking effect for top Chinese AI models from increasing actual real world coding adoption, with less reliance on model distillation ahead Compared with learning and distillation tactics from global SOTA models in the past, top Chinese models like GLM5 are reaching a critical stage of adoption by Chinaโs major enterprises and global users. As per LatePost, AI-generated code has increased to as high as 90% at some China mega-cap companies, up from 20-30% in 2H25, which will enable a positive data flywheel effect of further improvements via reinforcement learning with actual user data (both successful and unsuccessful coding cases) and post training. These have underpinned the step improvements in GLM5.2 from GLM5.1 in just over the course of a few months in 2026, and expect to see further step improvements to Chinese AI models over the next 6-12 months. In coding/agentic tasks, Chinese players are reaching global top-tier positionsย Compared with global SOTA leading models of several tens of trillions, China open source models are mostly around or below 1 trillion parameter in size, and adopt a MoE structure, with low activated to total parameter ratios for higher inference efficiency. Blended LLM token price (SDLLMTK) has been declining since early June, potentially driven by rising adoption of cost effective China models Historical and projected capex for major US & China cloud service providers Capex to operating cash flow ratio is still healthy for China hyperscalers Case study on Meituanโs LongCat 2.0: A milestone for Chinaโs domestic AI infrastructure Released on June 30, 2026, Meituanโs LongCat 2.0 marks a major milestone as Chinaโs first official 1.6 trillion-parameter open-source Mixture-of-Experts (MoE) model trained and deployed entirely on a 50,000-card domestic compute cluster. Purpose-built for agentic coding and complex software engineering workflows, the model features a native 1-million-token context window enabled by LongCat Sparse Attention (LSA) and dynamically activates an average of 48 billion parameters per token to optimize inference costs. Implications for a more self-sustainable China AI model outlook that is less reliant on foreign high-end chips for model training: The successful end-to-end pre-training and inference of a trillion-parameter class model on Chinese silicon (reportedly utilizing Huawei Atlas-950 SuperPods) fundamentally drives a more sustainable Chinaโs AI model development outlook, in our view. While previous Chinese flagship models, such as DeepSeek V4-pro, mentioned domestic chips for inference, LongCat 2.0โs ability to overcome critical memory bottlenecksย and distributed stability challenges during the compute-heavy pre-training phase proves the viability of a wider and more localized hardware stack for AI model training in the future. Why are Chinese models pursuing an open source/open weight approach, and ways to monetize? Open source allows for higher flexibility in model training/deployment, and allows for the widest adoption and an open communityย Alibabaโs Qwen model family has long pursued an open source approach (before adopting a closed source for its largest and highest performing Qwen-Max models for better monetization), while other key Chinese AI model players have mostly pursued an open source/open weight approach including DeepSeek, Zhipuโs GLM and MiniMax M3 series models with the exception of ByteDanceโs full closed proprietary approach for its Seed model. The open source approach allows for the highest flexibility in terms of the locations of where models are trained (and thus where the models can be deployed both inside and outside of Mainland China after training). An open source approach also allows for the highest adoption amongst the AI community with full transparency of the model parameters/architecture for trust, and an open community in driving more user feedback and thus model iterations/improvements. The open source approach vs. worldโs leading closed/proprietary approach also provides an alternative choice for worldwide AI users when the best performing closed models have stricter user access and higher costs from their premium pricing, especially at a time when โtoken-maxxingโ has become a key cost consideration for many corporates. Open source modelsโ disclosed ARRs are likely understating total deployment and revenue potentials While open source model companies offer their own coding plans and a chargeable open platform API channel (where the model companies conduct their own model inference), the majority of open source models also allow individuals/third-party hyperscalers/neocloud providers to deploy the models without a charge even for commercial use (e.g. Alibaba Cloudโs Bailian MaaS platform can house GLM5.2 open source model without needing to pay a fee/take rate to Zhipu). As a result, while Zhipuโs last stated ARR target for year-end 2026 is at US$1bn, the actual deployment of GLM5.2 model worldwide is and will be multiple-fold higher vs. Zhipuโs own API channel token volumes and revenue. There is also increasing post training of Chinese AI models that are re-branded by global companies (e.g. Composer 2 etc.) where the Chinese AI models do not necessarily receive any revenues given the open source spirit. Expect more shifts to open weight (with Community License) approach among Chinese AI models down the road While Zhipuโs open source GLM modelโs MIT license allows for free for all use cases (regardless of revenue), MiniMax M series models have pursed a restricted license (where the industry terms it as an open weight with Community License model), which requires MiniMaxโs agreement and commercial terms (e.g. revenue sharing/a take rate) on commercial use. This will be the likely next path for other open source models in the Chinese AI model industry, in order for eventual gross profits of inference tokens to cover training costs and for each AI model company to achieve a sustainable path to returns. What are the key addressable markets, domestically and internationally, and key risks? According to Goldman estimates, China AI modelsโ aggregate API+subscription revenue to increase from Rmb35bn in 2026E to Rmb879bn by 2030E from rising model intelligence, in particular with recent models like GLM5.2 reaching a critical point for global adoption and attractive pricing. The bank's revenue pool estimates for China AI models imply total Chinese AI model daily token consumption of 350T in 2026E to increase to 4,600T by 2030E. Goldman estimates domestic market to see token growth of 25X by 2030E; coding landscape to consolidate while agentic/low-end segment could remain fragmented. At 140tn daily token volumes for the country in March 2026 and several hundred trillions by June 2026 as per the National Bureau of Statistics, open source/open weight models have roughly a 30% token share (vs. 70% token market share by ByteDance alone, which is closed source and mainly driven by its Doubao app enterprise and individual user base, as the #1 used AI chatbot in China). Similar to the US, the coding segment (at a premium pricing level) will continue to be dominated by SOTA best performing models. Meanwhile, the lower-end segment focused on agentic AI will remain fragmented with multiple players due to the financial strength of AI model companies/mega-caps which would sustain the lower-end segment price war for longer. International (going global) to be the key upside; with potential for higher pricing and global proliferation, especially in non-US markets Goldman's US research team estimates agentic AI will drive 24X growth in token consumption by 2030 (from 2026) to 120 quadrillion tokens per month (or 4 quadrillion tokens per day, from their estimate of 170 trillion daily tokens today), with the biggest driver at 55X from enterprise agents and 12X from consumer agents. The global (ex. China) landscape has seen significant token share gains from Chinese AI models, as rising model intelligence and attractive token costs have driven higher adoption across 24/7 Hermes/Claw/Productivity agents, and shifting global SME mindset on using Chinese models in managing token costs given Chinese models have reached a โgood enoughโ stage in terms of intelligence/performance. Pivoting from โtoken-maxxingโ to ROI-focused metrics, e.g. Daily Active Agents/Agentic Work Units The AI token proliferation is undergoing a paradigm shift from โtoken-maxxingโ (an initial focus from late 2025 to early 2026 where enterprises equated high AI token consumption directly with organization productivity) towards an โROI-firstโ model that prioritizes clear task boundaries and output over raw computational volume. โToken-maxxingโ has been attributed to corporate inefficiencies and cost overruns: Data from a Jellyfish AI Engineering trends study indicated heavy AI users at enterprises consumed 10X more tokens but only had a 2X increase in output. Meanwhile, multiple US Internet companies have commented back in April 2026 that either their engineering teams utilized a full yearโs AI budget in just four months using agentic products (promoting stricter monthly caps per tool) or changed their internal token utilization leaderboards that previously wrongly incentivized staff to launch inefficient/low value autonomous agent tasks. Enterprise framework is pivoting towards an ROI-first model that prioritizes clear task boundaries, number of agents per day, backend process automation and actual output over pure computational token volume. Besides a marked increase in adoption of Chinese AI models, global enterprises have been downgrading default models to cheaper flash models for more typical tasks (while reserving SOTA models for only the top most value creating tasks like coding/programming). There is a transition from just token tracking to metrics like Daily Active Agents (DAA) and Agentic Work Units (AWU), and the overall cost per task will become more relevant than price per token metrics. Open source/open weight approach allows for the option for U.S. hyperscalers to host Chinese models, operated within U.S. cloud ecosystem Alphabet and Amazonโs respective cloud services Gemini Enterprise Agent Platform (via. its Model Garden) and AWS Bedrock already offer a broad selection of Chinese AI models including DeepSeek, MiniMax, Moonshot, GLM and Qwen which are fully managed by the US hyperscalers. Besides the model layer, into applications, worthy of note is Microsoft (covered by Gabriela Borges) CEOโs recent remarks at a Wall Street Journal interview (link) where he noted Microsoft is considering hosting versions of DeepSeek on Copilot as an optional, cost-effective model which could give its customers access to cheaper choices alongside US proprietary models. Microsoft indicated that if it hosts DeepSeek, the model would operate within its cloud ecosystem, ensuring customer data stays inside Azure. Noting key risks around any potential tighter geopolitical policies given Chinese AI models inroads into western markets. Key risks to the โgoing globalโ opportunity will hinge on end market access (especially in western countries, and with a focus on where computing is done/data is stored), access to high-end computing for AI model training (that could impact iteration pace and cost structure of Chinese models), and restrictions on Chinese AI model companies could impact supplier relationships/access to US companies and/or access to US capital.ย Who are best positioned to be the long term winners? Introducing our Competitive Positioning framework Players with the largest ARR scale with a gross margin advantage + financial strength are expected to be the long-term winners. ARR scale x gross margin advantage + financial strength Largest ARR scale (Token scale x pricing power) Gross margin advantage (Training & Inference efficiency, Technology) Financial strength (Balance sheet, Access to computing) Accordingly, Goldman introduces a Chinese AI model Competitive Positioning framework based on pricing power, cost advantage and finance strength, overlaying with token scale and market share progressions, and identify Knowledge Atlas (Zhipu initiation) and DeepSeek (private) as the most strongly positioned in foundation models, and Bytedance (private) in multi-modal capabilities Foundation models Goldman assesses each key playerโs competitive positioning from its flagship foundation model, scored across 3 aspects: pricing power, cost advantage and financial strength (of the company), with each aspect further built upon granular quantitative metrics. Pricing Power Time to market:ย Goldman assesses this in terms of how quickly and effectively a player delivers frontier competitive models. From comparing the launch date of the companyโs flagship models with its prior generation & other models at similar performance level, Goldman assesses whether the release delivers a meaningful capability step-up over its past generation, and whether it narrows the gap to the current global SOTA models. Arena score (overall text): Model intelligence is viewed as the primary determinant of pricing power, since models capable of handling higher-value tasks can deliver clearer ROI and therefore sustain their pricing premium. Here, the LMArenaโs score is used instead of any static benchmark because it reflects a large scale of blind user reviews, making it a more objective read on the real-world capability. Pricing (blended, US$ per 1M tokens): The bankย refer to the realized headline price (across input/output/cached input) for flagship models, where sustained higher pricing with iteration signals stronger pricing power, whereas pricing cuts may indicate a more volume-prioritized strategy. Cost advantage: the structural cost-to-serve (i.e. inference costs) that sets the floor for price and margin The below metrics are referred to as proxies for inference efficiency: Throughput (tokens/second): Measured by the number of tokens a single GPU generates per second. The more tokens a GPU outputs per second, the more fixed hourly compute cost is spread out. Therefore, throughput is a strong indicator of inference efficiency, which depends on model architecture (sparsity & attention design) and serving efficiency (batching & utilization). Cache hit rate (%): The share of input tokens served from cache rather than recomputed. In conversations and agentic workflows, much of the input repeats across calls, so models can read those tokens from cache memory instead ofย recomputing from new inputs. A higher rate cuts compute, and since cached tokens are near-free to serve despite being billed at a discount (10X-100X cheaper than input cost), it is also margin accretive. Parameter size/activation ratio: The share of total parameters activated per token (available for open-source MoE models), where fewer active parameters mean fewer FLOPs per token and therefore lower inference cost, at any given level of performance.ย Inference GPM (where disclosed, or GS estimates based on total parameter size/activated parameters): The realized gross margin on model API, where ~90% of COGS is inference costs, as a direct indication on cost efficiency Financial strength: capacity to keep funding frontier R&D and training before a profitability turnaround Cash on hand and net cash/debt as % of assets: Goldman uses total cash on hand as a measure of balance sheet strength, with net cash as % of assets to normalize across players of different scale. Mega-caps (compared with individual players) are seen as having greater resources to accumulate compute, fund multi-modal exploration and push faster product distribution. Valuation multiples: For independents, P/ARR 2026E multiples are used as they are not yet profitable and P/ARR is a more comparable measure of business scale and future monetization potential, and for mega-caps the P/E 2026E is used. Appendix China's Key Players at a Glance:ย Mega-caps China's Key Players at a Glance: Key Independent Players Much more, including the full assessment of multi-generational models, as well as upside and downside risks, in the full Goldman note available to pro subscribers. Tyler Durden Sun, 07/19/2026 - 22:45
'Going Dark': The High-Tech Tricks Behind Illegal Chinese Fishing In The Pacific Authored by AAP via The Epoch Times, The interception of a Chinese vessel accused of illegal fishing highlights the technical lengths many use to remain undetected across millions of square kilometres of open ocean. Tuvalu Police operating on a Sea Shepherd patrol boat boarded and arrested a Chinese longliner, the Lu Rong Yuan Yu 138, on July 9 for allegedly illegally operating within the small South Pacific nation's waters. In a supplied image, fish found on board Chinese longlining vessel, the Lu Rong Yuan Yu 138, at the time it was detained for alleged illegal fishing in NA, Tuvalu on July 9, 2026 PR Image/Supplied by Sea Shepard The vessel was allegedly running with its navigation light turned off and transmitting false data showing it to be in different area - an increasingly common practice known as "spoofing." After boarding the ship, officers found it did not hold a valid permit to fish the area, detaining the vessel and escorting it to the Port of Funafuti for further investigation. Sea Shepherd Australia Chair Peter Hammarstedt said illegal fishers were increasingly "going dark" by switching off their automatic identification systems, or using them to broadcast false information. "It's very common practice, especially in the Pacific Ocean, for Chinese fishing vessels to transmit false positions," Hammarstedt told AAP. "If you go onto some vessel tracking software websites, you'll see some Chinese fishing vessels appear on land, like in the Mongolian desert, for example. "Obviously the fishing vessel isn't there, but is operating elsewhere and simply inputting false data to mislead investigators." In the case of the Lu Rong Yuan Yu 138, it was allegedly transmitting a location that put it squarely on the equator at 0 degrees, raising the suspicions of investigators. Sea Shepherd has partnered with Tuvalu Police for the past two years, since the nation's only operational patrol boat was damaged in a cyclone. The small island nation monitors a marine domain covering roughly 750,000 square kilometres, with limited assets to keep an eye on such a large area, meaning most illegal fishing operations go undetected. "The fishing vessels that do this kind of behaviour operate under the expectation that they're just not going to get caught," Mr Hammarstedt said. Australia and Pacific nations have recently ramped up efforts to clamp down on illegal fishing under the Pacific Islands Forum Fisheries Agency Fishing Authorisation. "The Tuvaluans live by the sea, with the sea, and have done so for hundreds and hundreds of years," Hammarstedt said. "When I see illegal fishing on an industrial scale happening in the Tuvaluan waters, I see this as theft from the Tuvaluans." Tyler Durden Sun, 07/19/2026 - 22:10
India's Nuclear Push Further Strains The Massive Uranium Supply Gap Prime Minister Modi secured another major uranium supply line. During his visit to Australia this month, the two sides finalized the administrative arrangements needed to move Australian uranium to India under the previous 2014 nuclear cooperation agreement. The fuel must stay under IAEA safeguards for exclusively peaceful use.ย Australia holds roughly 28% of global uranium reserves. New Delhi has eyed those reserves for years, and now the yellowcake spigot can open. This follows the March deal with Canada's Cameco. That contract covers nearly 22 million pounds of U3O8 from 2027 through 2035 at market-related pricing, worth around C$2.6 billion.ย India already buys from Russia and Uzbekistan, and holds some domestic production. The message is consistent: a country targeting 100 GW of nuclear capacity by 2047 cannot rely on thin indigenous resources alone.ย Current operable capacity sits near 8 GW. The gap requires sustained annual additions measured in gigawatts, not megawatts. Cumulative net deficits keep expanding as reactor builds in China and Russia outpace primary supply response. Goldman models have only recently started taking into account the significant expansion in the SMR space as well, resulting in some relatively alarming supply gaps in the years ahead. The Cameco agreement already reflected this reality. CEO Tim Gitzel noted that sovereign buyers are locking up volumes in a window where available supply grows more uncertain. Kazakhstan has also moved to build its own strategic reserve. Couple these data points with Goldman's estimate of the ongoing lack of long-term uranium supply contracting by the existing global fleetโฆ We've been pointing to the uranium upside trade for several years now. We laid out a longer form of our thesis for the anticipated continued rise of the price of uranium just a few months ago. The idea is incredibly straightforward. There's not enough mines, and the demand is exploding faster than people can keep track of. Tyler Durden Sun, 07/19/2026 - 21:35
478 Individuals Charged In Border-Related Enforcements In Texas, Arizona Authored by Naveen Athrappully via The Epoch Times, A total of 478 individuals have been charged across Texas and Arizona over the past week in cases involving border and immigration-related crimes. A United States Border Patrol agent enters through the United States border wall outside of San Diego, on Jan. 20, 2026. John Fredricks/The Epoch Times In Texas, 199 people, including 175 illegal immigrants, were charged in 194 cases between July 10 and 16, according to a July 17 statement from the Department of Justice (DOJ). Authorities filed 146 criminal complaints in the Southern District of Texas for felony reentry after prior removal. Illegal entry charges were filed against 17 individuals. According to the DOJ, most of the arrested illegal immigrants have prior felonies, such as violent crimes, immigration crimes, and narcotics. Twenty-eight people were charged with human smuggling, including 12 illegal immigrants. Charges related to firearms have also been leveled against some of the apprehended people. One of the arrested individuals has a prior conviction for online solicitation of a minor. In Arizona, 279 individuals were charged during the July 11-17 week following immigration enforcement operations, according to another DOJ statement on July 17. Authorities filed 56 cases in the District of Arizona in which illegal immigrants were accused of reentering the United States unlawfully. Officials charged 206 illegal immigrants with illegally entering the country. Cases were filed against 17 individuals for allegedly smuggling foreign nationals into Arizona. The 478 arrested individuals were apprehended under Operation Take Back America, established by the then-deputy attorney general Todd Blanche through a March 2025 memorandum. The operation seeks to repel the entry of illegal immigrants into the United States, achieve the total elimination of cartels and transnational criminal organizations, and set up Homeland Security task forces to dismantle cross-border human smuggling and trafficking networks. In California, 111 border-related cases were filed this week, the DOJ said in a July 17 statement. Charges filed at the Southern District of California include reentering the United States after deportation, importing controlled substances, and bringing in illegal immigrants for financial gain. Among the arrested are a Mexican national with three prior felony immigration convictions, and a U.S. citizen who tried to traffic 131 pounds of cocaine when applying for entering the United States. ICE Enforcement The Trump administration is facing opposition from certain jurisdictions over immigration-related arrests. In April, the Philadelphia City Council voted unanimously to pass the ICE OUT legislation package, which placed restrictions on Immigration and Customs Enforcement (ICE) activities in the city, according to a council statement on April 15. The legislation bans the city from collaborating or sharing data with ICE, restricts ICE activity in city-owned spaces, and makes it illegal for ICE and other law enforcement agents to hide their identities with face masks and unmarked vehicles. The committee's vote on the ICE OUT package "affirms that we will stand with our immigrant communities and refuse to turn our backs on those who call this city home," Councilmember Rue Landau said in a statement at the time. A federal judge temporarily blocked the ban on face masks and unmarked vehicles on July 2 after the federal government requested a preliminary injunction. The prohibition was set to come into effect on July 7. Judge Chad Kenney of the U.S. District Court for the Eastern District of Pennsylvania, who issued the ruling, said the city's law "attempted to sidestep the Constitution's clear mandate." Meanwhile, ICE said in a July 1 statement that the Secure America Act, signed into law by President Donald Trump in June, provides the agency with $350 million in additional funding that would enable the arrests of more criminal illegal immigrants hiding in sanctuary jurisdictions. Acting ICE director David J. Venturella said in the statement that sanctuary jurisdictions refuse to honor ICE detainers or coordinate to safely transfer illegal immigrants from local custody to federal authorities. "As a result, criminal illegal aliens who could have been transferred directly to ICE are instead released into our communities, forcing our officers to locate and arrest them in neighborhoods, businesses, and other public locations," Venturella said. "The funding from this legislation will help ICE increase its capacity to monitor releases and arrest removable criminal aliens when sanctuary jurisdictions refuse to cooperate with federal immigration authorities." Tyler Durden Sun, 07/19/2026 - 21:00
The Data-Center Revolt Goes National: Tea Party Veteran Leads 142 Rallies Across 42 States The backlash against the AI data-center build-out - which we've been tracking since it was a smattering of county fights across 28 states - staged its first coordinated day of action on Saturday: 142 protests across 42 states, from Wasilla, Alaska to Naples, Florida, organized by Humans First, the nonprofit that Tea Party veteran Amy Kremer co-founded and chairs. The crowds spanned both sides of the aisle...uniting a MAGA stalwart in a 'faith, family, freedom' T-shirt in New Jersey, a first-time activist in Texas, and a left-leaning organizer in California's Imperial Valley. The fight looked like Kenilworth, New Jersey. Residents of the 8,500-person borough gathered outside the municipal court at mid-morning with drums, plastic horns, and sidewalk chalk to protest the $1.8 billion CoreWeave AI data center their planning board approved in May 2025 on the former Merck campus - a project that has since drawn more than 12,000 petition signatures against it, several thousand more names than the town has people. The woman in the "faith, family, freedom" T-shirt marched beside neighbors holding "Build community, not data centers" signs. When heavy rain arrived later in the day, they pulled on ponchos, shared umbrellas, and kept marching. One sign, caught by Business Insider's photographer on the scene: "You think this is pressure? Wait 'til there's no water pressure." Texas, the country's hottest data-center market, hosted the most rallies - 18 - with Georgia at 11, California at eight, and Pennsylvania, Florida, and Indiana at seven apiece. In Imperial Valley, where a proposed facility could pull 260 million gallons a year from the Colorado River, Ivan DelSol, 54, told Reuters that around 50 people turned out in 100-degree heat. "It's dystopian that you would use this much fresh water for AI," he said. Organizers released no headcounts; turnout ran below expectations in some rural areas and in Atlanta, where about a dozen showed - most of them, a volunteer there said, driving in from the smaller Georgia towns where the biggest data centers are going up. Kremer is a founding figure of the Tea Party movement who went on to found Women for Trump, and an organizer of the January 6, 2021 rally that preceded the Capitol riot (she neither planned nor took part in the riot itself). She has spent months calling data centers the defining fight of her lifetime, warning the technology could threaten humanity itself, and she is open about running the old playbook: grassroots pressure, town by town, aimed at both parties. Today was incredible! 142 protests. 42 states. One message: ONE NATION UNITED AGAINST DATA CENTERS! Communities across America showed up today for a National Day of Protest Against AI Data Centers to say that We The People deserve a voice before massive data centers areโฆ pic.twitter.com/TwE7oRjR6V โ Kylie Jane Kremer (@KylieJaneKremer) July 18, 2026 Her crowds bear that out. One of Saturday's Texas rallies, in Tyler, was organized by Eva Cardona, a 31-year-old self-described political nomad and first-time activist who told Reuters she wanted something more hands-on than posting on Facebook; about a dozen people came. And in Imperial Valley, the man who helped lead the rally leans left. The polling explains why a coalition that broad holds. A June Reuters/Ipsos survey found only 14 percent of Americans would support a data center in their own community. Gallup polling fielded in March found 71 percent oppose building an AI data center in their area - 48 percent strongly - a worse number than a local nuclear plant gets. And Morgan Stanley told clients in a July 14 note that support for local data-center bans runs strongest among Republican, higher-income, and urban voters, while Morning Consult's national tracker crossed a line of its own in May: "stop building" (about 45 percent) overtook "keep building while expanding energy supply" (about 38 percent) for the first time since last October. Several democratic candidates, including candidate for Texas Governor Gina Hinajosa, stopped in Lubbock to discuss campaign topics including data centers. A data center protest was also held to collect signatures to pause building hyperscale data centers in the Lubbock area. pic.twitter.com/wh3mdGNcu4 โ KAMC News (@KAMCNews) July 19, 2026 For all the movement's reputation, Kremer's demands stop well short of a shutdown. She opposes a national moratorium and statewide moratoriums alike, telling Business Insider that each community should choose what gets built inside it, and that too many of those choices are made behind closed doors. Humans First's platform runs to transparent approval processes, environmental review before permits are granted, union construction jobs, and binding developer commitments of the kind lawyers call community benefits agreements. She has aimed as much fire at her own side, accusing Republicans of giving Big Tech a free pass and predicting the industry will cozy up to Democrats the moment the majority flips. The fix, she argues, belongs to Congress. Amy Kremer is the cofounder of Women for Trump and Women for America First. Now she's taking on AI data centers. Jacquelyn Martin/AP The organization is a narrower thing than the crowds it convened. Humans First announced in April that its non-conservative team members would spin off into a separate group, with Kremer promising "a topflight team of conservatives" to fight Big AI and its lobbyists. The banner over Saturday's rallies, in other words, was a conservative one - which makes the mix of people who marched beneath it all the more striking. Official Backlash The rallies capped a fast-moving week. On Tuesday, New York Governor Kathy Hochul signed an executive order imposing the nation's first statewide moratorium on new hyperscale data centers - an immediate pause of up to a year on state environmental permits for projects of 50 megawatts or more while regulators draft standards covering energy demand, water use, and air quality. Her office promised localities community-benefit guidance within 60 days, and Hochul will pursue repeal of the state's sales-tax exemptions for massive data centers. A tougher bill passed by the legislature, with a 20-megawatt threshold, remains unsigned on her desk; her office has called it complicated, and Hochul said the state wants to be "the first to get it right." New York was not alone. Virginia's new tax on data-center electricity - 1.1 cents per kilowatt-hour - took effect July 1. Pennsylvania's House passed a ban on non-disclosure agreements in data-center deals by a 171-31 vote, and separately voted 197-5 to repeal the industry's sales-tax exemption, a break worth roughly $517 million a year by 2030. Arizona's governor signed a three-year moratorium on new data-center tax breaks in June. Legislators have filed more than 300 data-center bills this year; local pauses have passed at the county, city, and tribal level in at least 15 states. Meanwhile In China Nothing comparable is happening on the other side of the Pacific. Two days before the marches, Beijing-based Moonshot AI released Kimi K3, a 2.8-trillion-parameter system billed as the largest open-weight model ever built and claimed to perform level with America's best frontier models. Bloomberg has reported that Beijing plans to spend roughly $295 billion over five years on a nationwide network of AI computing hubs - and none of it will face a zoning board. Under the state's "Eastern Data, Western Computing" program, the buildout is steered into the arid, sparsely populated west; provincial governments compete to attract data centers with tax holidays, cheap land, and compute vouchers, and the state absorbs up to half of operators' energy costs, so the strain never shows up on a household bill. No Chinese county has passed a moratorium, because no Chinese county gets a vote. The financial toll is no longer hypothetical. Third-party trackers cited by Morgan Stanley in its July 14 note put the value of cancelled or delayed projects at roughly $156 billion in 2025 and another $130 billion in the first quarter of 2026 alone - about $286 billion in all - set against the bank's own estimate of $877 billion in AI capital spending this year. The underlying quarterly count comes from Data Center Watch, a tracker run by 10a Labs, an intelligence firm whose client list includes AI companies, which logged at least 75 projects blocked or delayed from January through March - matching in one quarter the number of projects derailed in all of 2025 - as active opposition groups more than doubled from 396 to 833 and spread to 49 states. Saturday itself stayed peaceful - chalk, chants, drums, umbrellas - though the wider fight has had harder edges: developers of the Piedmont transmission line into Northern Virginia's data-center corridor asked a federal court last summer for U.S. Marshals to escort survey crews after landowners threatened workers, an episode we covered at the time. The industry answered forcefully. The Data Center Coalition warned that New York's moratorium tells investors the state is "closed for business" and will push jobs and tax revenue to neighboring states. Seven major AI and cloud companies - Amazon, Google, Meta, Microsoft, OpenAI, Oracle, and xAI - point to the Ratepayer Protection Pledge they signed at the White House in March, a voluntary commitment to cover the grid costs their facilities create. And White House AI czar David Sacks went after Hochul's case point by point on the All-In podcast, calling data centers "the scapegoat for all of the angst that people have about AI." His answer to the utility-bill complaint: let developers build their own generation behind the meter instead of competing with households for grid power. David Sacks: Everything Kathy Hochul Said About Datacenters Was a False Accusation@DavidSacks: โThese data centers become the scapegoat for all of the angst that people have about AI. Everything she's saying there is a false accusation on datacenters. So let's just go one byโฆ pic.twitter.com/gAxesw3fRD โ The All-In Podcast (@theallinpod) July 18, 2026 Some industry allies go further, pointing to OpenAI's June disclosure that it banned a China-linked network using ChatGPT to mass-produce comics and comments blaming data centers for rising power bills. OpenAI itself found the operation gained almost no authentic traction - and the crowds in Kenilworth and Imperial County were unambiguously homegrown. Kremer spent Saturday evening thanking volunteers and looking past the weekend. "America is not for sale, and our communities are not collateral," she wrote on X. She expects data centers on the ballot in November, and again in 2028. The midterms are less than four months away. Tyler Durden Sun, 07/19/2026 - 21:00
China's Moonshot Eyes IPO After Sparking Another "DeepSeek Moment" For US AI Stocks Chinese startup Moonshot AI sparked another DeepSeek-like moment for US AI stocks and hyperscalers last week. David Sacks, the former White House AI and cryptocurrency czar, warned that this is "very concerning." Moonshot AI debuted Kimi K3, which ranks number three on the Artificial Analysis Intelligence Index, trailing only OpenAI's GPT-5.6 Sol and Anthropic's Fable 5. This only shows the continued trend of open-weight models toward the AI frontier. Kimi K3 has emerged as a top challenger across reasoning and coding workloads, highlighting the rapid pace of innovation and the increasing competitive pressure on models in terms of capabilities and downward trend in pricing. This only sparked weakness in sentiment toward AI US stocks and hyperscalers last week due to their massive capex binges to build out data centers. Related: Goldman Warns Massive Hyperscaler Bond Issuance Means "Risk Skewed To Downside" It serves as a reminder for companies amid the 'tokenmaxxing' woes that there are lower-cost alternatives for end users besides expensive frontier Anthropic and OpenAI. Now, Moonshot AI is leveraging last week's momentum and preparing for an initial public offering in Hong Kong within six months. Kimi K3 has strengthened investor confidence, positioning the startup to enter public markets. Bloomberg reports that Moonshot is in the process of concluding a funding round that values the three-year-old startup at $30 billion. Annual revenues topped $300 million in June, up from $200 million in April, supported by chatbot subscriptions and enterprise services. The startup has held talks with CICC and Goldman Sachs about the offering and is dismantling its red-chip structure to facilitate an overseas listing. Meanwhile...ย Weโve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax. These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models. โ Anthropic (@AnthropicAI) February 23, 2026 Here is the latest institutional commentary on Moonshot AI: Jefferies analyst Roger Song Moonnshot AI triggered another Deepseek-like moment + semi/ memory lost momentum, both potentially benefiting biopharma rotation. While still waiting for FDA leadership renewal, we started to see more potential policy reversals (while maintaining good ones, nice to see CRL release back for now) along with swift industry reactions. Open clinical trial prediction betting - will that be a distraction/ noise or another actionable datapoint for real investment? We shall all see. Wolfe Research analyst Alex Zukin China's Moonshot Unveils Kimi K3, Autonomously Designed Functional Chip in 48 Hours - Moonshot released Kimi K3, a 2.8 trillion parameter Mixture-of-Experts model with a 1-million-token context window and native text, image, and video support, positioned as the first open 3T-class frontier model with full weights releasing July 27, 2026 and API pricing at $3/$15 per million input/output tokens. The model autonomously designed a functional 4mmยฒ chip in a single 48-hour run with zero human intervention, handling design, optimization, verification, and simulation, achieving over 8,700 tokens/second decoding throughput, while also building a custom GPU compiler called MiniTriton from scratch that reportedly matches parts of Nvidia's official Triton compiler. Kimi K3 scored 88.3 on Terminal-Bench 2.1, trailing but competitive with OpenAI's GPT-5.6 Sol, while lagging flagship closed models like Anthropic's Claude Fable 5 in broader evaluations, with the model available today via the Kimi app, Kimi Code, and OpenRouter Canaccord Genuity analyst Martin Roberge Meanwhile, sentiment toward AI stocks and hyperscalers' capex binge keeps weakening and today's release of a new AI model by Chinese company Moonshot, capable of competing with AI models from OpenAI and Anthropic, serves as a reminder that there are lower-cost alternatives for end users. The race among AI startups to go public has China-based DeepSeek targeting a 2027 IPO, while US rival Anthropic could list as early as late 2026 and OpenAI increasingly appears headed for a 2027 debut. Both US firms have confidentially submitted IPO paperwork. Tyler Durden Sun, 07/19/2026 - 20:25
Taiwan's President Ratchets Anti-China Rhetoric In Party Speech: Must Resist 'Red Terror' Of Beijing President Lai Ching-te said on Sunday that Taiwan must renew efforts to protect its democracy and never become part of China. He urged members of his Democratic Progressive Party (DPP) to oppose the "red terror" coming โfrom Beijing. The ruling DPP champions Taiwan's separate โidentity from China, and is working to combat whatย Lai calledย China's "legal warfare". via Bloomberg "I also expect comrades within โthe party to stand on the front lines, unite as one, and jointly oppose the โthreat posed by China's 'red terror' โto Taiwanese society," Lai added, speaking in Taiwanese, rather than the main language of government, Mandarin. "We must work together โto protect our democratic and free way of life, and absolutely never allow 'democratic Taiwan' to turn back and become 'China's Taiwan'," he said. "Regardless of ethnic group, regardless of who came earlier or later, anyone who identifies with Taiwan is a master of โthe country. Taiwan's future must be decided jointly by the 23 million people of Taiwan," he said. One establishment US publication has held up a prime example of Beijing's "legal warfare" against an autonomous Taiwan in the following: On March 12, Chinaโs legislatureย adoptedย the Law on Promoting Ethnic Unity and Progress (Chinese;ย English translation), a sweeping new statute that codifies Beijingโs approach toward Chinaโs 56 officially recognized ethnic groups.ย Substantively, the law enshrines a decades-long shift towards aggressive assimilationist policies.ย Structurally, it reflects a deepening merger of Party ideology and state law that is becoming increasingly prevalent under Xi Jinping. This new law is the culmination of a policy trajectory that has been building for over a decade, dating back to theย 2014 Central Ethnic Work Conference.ย Under Xi, Beijing is steering away from the post-1949 legal framework of nominal ethnic autonomy (albeit under tight Party control) imported from the Soviet Union.ย In its place, officials have steadily been pivoting towards what scholars have termed โsecond-generation ethnic policiesโโan aggressive assimilationist approach that emphasizes a common Chinese national identity over accommodation of ethnic differences. Provincial and municipal authorities across China have enacted a wave of local โethnic unity and progressโ regulations in recent years, such as those inย Xinjiangย (2015) orย Inner Mongoliaย (2021). The new national legislation elevates this approach to the level of a national statute governing all of China. The new lawโs core concept is captured in the termย zhulaoย โ to โforgeโ or โcastโ metal โ and its instruction that โforging the communal consciousness of the Chinese nationโ is core to the Partyโs ethnic policies.ย As James Leibold hasย pointed out, this phrasing reflects a hardening of Beijingโs political line under Xi Jinping โ explicitly written into the Partyโsย Charter at the 19thย Party Congress in 2017 โ aimed at โmeltingโ subnational and ethnic identities into a shared collective one. Meanwhile, on a global stage Beijing has continued to present itself as the only peace guarantor and as aย force for stabilityย and is seekingย 'Taiwan's willing participation' - at a moment the Middle East is on fireย largely as a result of American policy and quickness to result to force and surprise attacks. President Donald Trump suggests the United States might not come to Taiwanโs aid if China attacks, arguing that Taiwan โtookโ Americaโs advanced semiconductor business and should therefore pay more for its own protection. โ @atrupar (July 2023)pic.twitter.com/WNMUZe6IUf โ The Intellectualist (@highbrow_nobrow) July 19, 2026 And yet, President Trump has of late publicly touted his personal relationship with Chinese President Xi Jinping as "amazing". Planned weapons deal with Taiwan have been indefinitely put on hold as Washington tries to repair relations with Beijing. Tyler Durden Sun, 07/19/2026 - 19:50
Judge Strikes Down Race-Based Provision In Biden-Era Internet Access Grant Program Authored by Aldgra Fredly via The Epoch Times, A federal judge ruled on July 15 that a race-based provision of the Digital Equity Act, signed by President Joe Biden in 2021 to close digital gaps, was unconstitutional. A judge's gavel rests on top of a desk in a courtroom in Miami, Fla., on Feb. 3, 2009. Joe Raedle/Getty Images The Digital Equity Act was part of Biden's Infrastructure Investment and Jobs Act, which appropriated $2.75 billion to the National Telecommunications and Information Administration (NTIA) to establish grant programs to expand high-speed internet access for minority groups and communities in rural areas. After taking office for a second term last year, President Donald Trump halted the competitive grant program authorized under the Digital Equity Act, saying it was unconstitutional because it allocated federal funding based on race. The National Digital Inclusion Alliance, a recipient of the competitive grant program, later filed a lawsuit in October 2025 seeking to reinstate the program. In a 35-page order, U.S. District Judge John Bates ruled that the Digital Equity Act's provision authorizing the use of race in awarding federal funds was unconstitutional, citing the Supreme Court's 2023 ruling that struck down race-based preferences in higher education admissions. Bates said that while the Digital Equity Act aims to address the digital divide among minority groups and other covered populations, the Supreme Court precedent showed that remedying general social disparities alone does not justify the use of race in government action. "Addressing that gap is a laudable goal, but the Supreme Court has admonished that ameliorating general societal inequalities - as opposed to specific instances of past discrimination - 'does not constitute a compelling interest that justifies race-based state action,'" the judge stated. "Otherwise, Congress could deploy racial classifications when confronted with any situation of an uneven resource distribution." Bates said the grant program could be reinstated without the race-based provision, and the government had committed to restoring it upon a judicial determination that the provision was unconstitutional. Trump welcomed the ruling in a Truth Social post, calling it a "big win" for the American people. "The so-called 'Digital Equity Act,' a Biden DEI law, was ruled exactly what I said it was last year - A RACIST and UNCONSTITUTIONAL giveaway that never should have become Law," he wrote. The decision to end the Digital Equity Act comes amid the Trump administration's efforts to eliminate diversity, equity, and inclusion (DEI) programs from federal agencies and government initiatives. Trump stated in a Jan. 20, 2025, executive order that the previous administration had forced "illegal and immoral discrimination programs" across virtually "all aspects of the federal government" through DEI initiatives. Christopher Mitchell, director of the Community Broadband Networks Initiative at the Institute for Local Self-Reliance, credited the National Digital Inclusion Alliance with helping to secure the program's restoration. "Yesterday's ruling on the Digital Equity Competitive Grant Program is, on balance, a victory," Mitchell said in a statement. "The only real question now is how quickly NTIA moves to actually implement it." The National Digital Inclusion Alliance did not return a request for comment by publication time. Tyler Durden Sun, 07/19/2026 - 18:40
NBA Boss Calls Caitlin Clark A "Political Football" Over Racism Controversy The WNBA has been the butt of jokes for decades.ย The women's league is a perpetual money pit, with annual losses of around $40 million per year, all subsidized by the men's league.ย The level of play is often tedious, the pace is much slower and the athleticism is lacking compared to the men's game.ย It's the reason why the WNBA has far less viewers, makes no profit and why the women athletes are paid far less.ย It's not sexism ad many of the players claim, it's just basic math. One would think that any player that doubles (in some cases triples) the viewership of league games by her mere presence would be widely celebrated by all the people involved including the commission, but sadly, this has not been the case with Caitlin Clark.ย Clark's expertise and superior ability on the court has brought new life to a floundering sport; the problem is, she's relatively new to the league and, she's white.ย This combination triggered many of the minority players, leading to public back-biting on sports talk shows and flagrant fouls on the court.ย The audience and social media started to take notice.ย In fact, one could argue that it was all the clips of jealous WNBA athletes attacking Clark on the court that made her more popular than ever.ย Not to mention, it seems as if most of the fouls are committed by black players.ย The latest flagrant foul involved Phoenixย Mercury forward Alyssa Thomas, who appeared to knee Clark in the groin area and then press her closed fist into Clark's throat while she was on the ground.ย No foul was called.ย ๐ฅHold upโฆA closer look at the dirty play by Alyssa Thomas last night w/ a new angle โ Wait for the freeze + zoom on each. - Knee to the groin - Pinches her underarm - Fist to the throat - Attempt to stomp on her (she moved) Why do @WNBA players hate Caitlin Clark so much? ๐คฏ pic.twitter.com/83JZSChdEu โ HeroOfTheDay (@Hero_OfThe_Day) June 25, 2026 Clark suffered a back injury from the incident and has to sit out multiple games.ย This is a regular occurrence with Clark; she stands near the top of the list for most fouls and flagrant fouls in the WNBA.ย In many cases involving flagrant fouls, referees do not make the call, leading to anger and social media backlash among fans.ย Some of the dirtiest tricks in basketball are on display regularly in WNBA games, and the attacks on Clark look practiced and organized. ย ย It's not always black players that go for the cheap shots, but there is a clear pattern emerging.ย It could be envy over Clark's far larger fan following despite being a newer player (female cattiness), but suspicions are hanging in the air that her treatment is largely race motivated.ย By far, Clark's most vocal critics in sports media are black. The WNBA and the NBA are not happy with the political undertones of the Clark controversy.ย NBA Commissioner Adam Silver recently called Clark a "political football" in interviews as he attempted to dismiss the WNBA's numerous problems with officiating fouls.ย He seems to outright dismiss any concerns about race-based targeting of Clark. Speaking as part of a panel at an event in New York, Silver said the debates surrounding Clark had become about broader political and cultural issues in the United States rather than basketball alone.ย He specifically referenced the foul by Alyssa Thomas that led to Clark's injuries.ย ย "That particular incident is not about whether a foul should have been called at the time of the game or whether that was ultimately a flagrant non-review....I've come to know Caitlin really well. She's an incredible player and also an incredible person.ย "And she wants to focus on being the best player she can. And she's become a bit of a political football in this country, and I think it's incredibly unfair to her."ย ย ย Speaking of unfair, Clark was snubbed by the WNBA in a 30th Anniversary commemorative poster (she was strangely absent from the roster), and she was snubbed by the 2024 Olympic "Dream Team" despite being a top star and #1 draft pick.ย Emmanuel Acho, a former NFL player and sports analyst (who is also coincidentally black), asserted on the Speakeasy podcast in June that Clark has "become a distraction" and that the WNBA would be better off without her.ย He argued that Clark got audiences to watch the games, and now the WNBA can throw her away.ย ๐จ๐จWORST TAKE OF ALL-TIME๐จ๐จ Ex-#NFL LB Emmanuel Acho says the WNBA would be better off without Caitlin Clark because she is a distraction. โThe WNBA would be better without Caitlin Clark, because she is a bigger distraction than she is an additiveโ pic.twitter.com/MVwHoWRS8u โ MLFootball (@MLFootball) June 29, 2026 They hate Clark and her popularity with a passion, but they like the idea of stealing her audience for themselves.ย It's rather pathetic, but we're talking about the WNBA.ย They have lived in the gutter for so long they can't imagine climbing out of it.ย Of course, Acho is delusional if he thinks all the "eyes" Clark brings to the WNBA won't leave with her if she exits the league.ย ย The bottom line is this: Clark puts butts in seats.ย No other WNBA player comes close.ย Yet, the league and many of its commentators treat her like a pariah because they can't make their crow's nest of minority players behave.ย They are the cause of the controversy.ย They made Clark into a "political football" by remaining apathetic and dismissive.ย ย Tyler Durden Sun, 07/19/2026 - 17:30
Trump Endorses Graham's Sister For Full Term, Shattering Expectations Of Short Stay Following the sudden July 11 death of Senator Lindsey Graham, South Carolina Gov. Henry McMaster triggered eye-rolling across America when he appointed Graham's sisterย Darline Nordone to take over the seat. At the time, the appointmentย of the unqualified sibling was widely understood as a mere interim honor, with Nordone serving as a placeholder pending the outcome of a special primary election to choose a Republican candidate for November's previously-scheduled election for the seat. Now, President Trump has asked Nordone to run in that primary race and pursue the full 6-year term that will start in January.ย "During her visit, I asked Darline, for the Good of our Nation, to run for the U.S. Senate in the Special Republican Primary on Tuesday, August 11, 2026," Trump wrote on Truth Social. "I hope Darline does this, in that there would be nobody better to honor the legacy of her beloved brother, Lindsey. Darline...comes from an absolutely incredible family [and] has been a WINNER all of her life...RUN, DARLINE, RUN!"ย While Nordone herself hasn't announced that intention, there are reports that she's "considering" it, and it's highly unlikely Trump would have bothered with a lengthy public endorsement unless Nordone was on the verge of jumping in the abbreviated primary race that already has less than a month left in it.ย ย At the press conference announcing her interim appointment, Nordone's language reinforced universal expectations of a short stay in the Senate: โIt is such a privilege to get to finish some of [my brother's] important work, and I promise to work hard over the next several months to support the president and carry forward the efforts of my brother on behalf of the citizens of South Carolina and the United States.โ Nordone has never held an elected office. She's a commissioner on the South Carolina Commission for the Blind, and previously held a communications role in theย South Carolina Vocational Rehabilitation Department. Trump said he personally pitched Nordone on going after the seat on a permanent basis when she visited him in the Oval Office on Thursday.ย So far, three people have officially announced they're running for the seat: businessman Mark Lynch, who was trounced by Graham in the June GOP primary; Ralph Norman, a 73-year-old US House member; and Duke Buckner, a lawyer who has previously run for the state's heavily-Democratic 6th Congressional District seat. Shortly after Graham died, Norman asked Trump for an endorsement. The president was said to have replied, "Give me a week." In the meantime, Norman has been endorsed by Florida Sen. Rick Scott and Utah Sen. Mike Lee.ย Following Lord Lindseyโs passing, Darline, Duchess of Charleston, ascended the High Chamber: crowned Baroness of the Palmetto Seat. Her brother's fiefdom and noble lineage are preserved. Long live the House of Graham and DEI! Aristocrats celebrate in Alexandria and Tel-Aviv. https://t.co/y5JGzSwz5v โ Glenn Greenwald (@ggreenwald) July 14, 2026 The candidate list had been expected to grow,ย but Trump's preemptive endorsement of Nordone could have a chilling effect.ย Rep. Nancy Mace, who's poised to hand over her House seat in January -- having foregone reelection for a failed bid for governor --ย teased at a run on social media within hours of Graham's death. On Friday, Axios reported she's still actively considering it. Rep. Russell Fry and Lt. Gov. Pamela Evette are among others whose names have circulated. An Emerson College poll found that a Trump endorsement would likely sway 41% of South Carolina GOP voters, but an equal percent claimed it wouldn't have an effect. If no candidate scores not just a plurality but a majority of the votes on Aug. 11, the top two vote-getters will advance to a runoff election on Aug. 25.ย Graham was a chief champion of the West's proxy war against Russia in Ukraine, and a zealous collaborator with Israel in promoting American warfare against Iran. To the extent that Nordone would attempt to pick up those torches and run with them, she's sure to garner the backing of the military-industrial complex and the pro-Israel lobby. That said, it's far from clear how effective she would be, given her utter lack of experience in Washington politics. Tyler Durden Sun, 07/19/2026 - 16:55
Judge Dismisses DOJ's New Mexico Voter Data Request Authored by Kimberly Hayek via The Epoch Times, A federal judge on July 14 rejected the Justice Department's attempt to obtain the private personal data of every registered voter in New Mexico, adding to the overall number of cases the DOJ has lost in which it has sought similar information. The U.S. Department of Justice in Washington on Aug. 7, 2025. Madalina Kilroy/The Epoch Times U.S. District Judge Judith C. Herrera dismissed the latest voter roll lawsuit, ruling that the department's demand letter lacked facts, proof of a pattern of violations, and explanations of necessity. The court found that "the DOJ's Demand Letter fails because it altogether lacks an identifiable 'basis.' Nowhere does the DOJ articulate any factual suggestion that New Mexico has violated the NVRA [National Voter Registration Act] or HAVA [Help America Vote Act], indicate the State has a pattern or practice of noncompliance with the same, nor does it explain how the unredacted PII is necessary to evaluate compliance with the NVRA and HAVA." The DOJ did not immediately return a request for comment. New Mexico Secretary of State Maggie Toulouse Oliver approved of the court's decision. "I am pleased with the court's decision to dismiss this case. Federal and state legal guardrails on social security numbers and dates of birth exist for the identity protection of every voter in our state," she said. She doubled down on the security practices deployed under her leadership. "I absolutely will not risk any disclosure of voters' private data, as it could carry very real and severe consequences for the personal lives of New Mexicans participating in our democratic process," she said. The Justice Department has now been barred in 14 similar cases nationwide, failing to obtain unredacted voter files via these lawsuits from states that declined its requests, according to Oliver. A federal judge on June 29 threw out the Justice Department's lawsuit against New Hampshire. The state refused to disclose its unredacted statewide voter registration lists. In a 26-page ruling, U.S. District Judge Joseph Laplante stated that those state-generated voter records fall outside those documents the federal government can demand under Title III of the Civil Rights Act of 1960. In April, judges dismissed similar cases in Arizona and Massachusetts. In February, a judge rejected a DOJ request for similar information in Michigan. In January, a federal judge in Oregon ruled along similar lines. U.S. District Judge Mustafa Kasubhai dismissed the Justice Department's suit requesting Oregon's list of registered voters. During a videoconference hearing, he granted the defendants' motion to dismiss the case. The department alleged that Oregon's refusal violated the Help America Vote Act of 2002 and the Civil Rights Act of 1960. The Justice Department filed lawsuits against six states in December 2025. Aldgra Fredley contributed to this report. Tyler Durden Sun, 07/19/2026 - 15:10
Mamdani's New York: West Side Homeless Camp Stretching 12 Blocks Now Allegedly Stealing City Power You've likely seen it if you've the nearby Hell's Kitchen entrance to the Lincoln Tunnel... What began as a growing homeless encampment on Manhattan's West Side has now escalated to alleged utility theft, according to the NY Post. According to reports, one resident of the sprawling encampment was seen Tuesday tapping into city electrical lines to power a makeshift shelter near West 34th Street between 11th and 12th Avenue. The camp now stretches roughly a dozen blocks, from 34th Street to 46th Street near the Intrepid Museum, and has become an increasingly visible flashpoint for residents, commuters, and tourists. The NY Post writes that bus drivers and local workers say the encampment has continued to expand over the past several weeks, with tents multiplying and sidewalks becoming increasingly obstructed. Photo: NY Post "It's been almost a month now and it gets a little bigger every time I come back," one FlixBus driver told the New York Post. He said garbage piled along the sidewalk forces passengers waiting for buses into the street, while some people living in the encampment have reportedly approached travelers asking for money. He added that complaints to city officials have produced little visible response. Police officers were seen Tuesday ordering the individual accused of siphoning electricity to disconnect from the utility line. Elsewhere along the encampment, officers reportedly seized extension cords and power strips from another makeshift shelter. Despite those actions, there appears to be little expectation that the enforcement will significantly reduce the camp's size. Mayor Zohran Mamdani has defended the city's broader approach, saying the priority is moving homeless individuals into shelters and ultimately permanent housing rather than simply dispersing encampments from one neighborhood to another. Asked specifically about the West Side encampment, he said the city would review the situation. Despite growing attention, reports indicate that no sanitation crews or homeless outreach teams had been seen at the site in the days following news coverage of the encampment. The camp itself has reportedly continued to grow, with additional tents appearing over the weekend. Reports also claim that the area has attracted people struggling with addiction, along with sex workers, adding to concerns from nearby businesses and residents who say conditions continue to deteriorate. Tyler Durden Sun, 07/19/2026 - 14:35
Overloaded: From Trinkets To Compute And Market Structure By Peter Tchir of Academy Securities Last weekend we published that the market was sitting on Multiple Inflection Points. We were concerned that many of the inflection points would resolve themselves negatively for the market: Iran, which we didnโt even give the full โinflectionโ point treatment to, has deteriorated. While Brent finished the week at almost $90, up from just above $70, it seemed to be a โside storyโ at this stage. However, with the news of the U.S. service members killed on Friday in Jordan, the question becomes how this alters the U.S. strategy to pressure Iran to stop its attacks on shipping in the Strait and return to the table to continue negotiations. AI Spend was the most important inflection point and that seems to be resolving itself rather negatively with the Philly Semi Index down 10% on the week! More on this later. Russia/Ukraine. General (ret.) Spider Marks, Rachel Washburn, and I spent a lot of time talking about this conflict. I continue to be optimistic, maybe even a bit more optimistic than the consensus Geopolitical Intelligence Group view, but thatโs my take. Japanese Yen. The infamous carry trade did little last week, but strength in the Yen remains a risk for the broader market. Crypto and DATCos. With the volatility in other markets last week, the stability in this space was noticeable. The jury is still out on which way this inflection point will resolve itself, but the case that it is forming a solid base is growing. Inflation. As one of the last people looking for not just cuts before hikes, but also cuts as early as September, Tuesdayโs CPI numbers helped a lot! Some would argue that โCoreโ remains high, but it is artificially inflated by the โmysteriousโ way we choose to represent shelter inflation. (Yes, Iโm hoping to get some call, out of the blue, to help re-evaluate what data decisions should be based on). Unfortunately, the resumption of increased hostilities in the Middle East and the limited amount of oil left in the Strategic Petroleum Reserve donโt help my take on inflation continuing to decline. Though any material slowing in the AI Spend would push the Fed (on both inflation and jobs) to consider cutting. Last weekend (according to WIRP) the market was pricing in 1.5 hikes by the end of the December meeting. It is now at 1.26 hikes. I think that expectation will continue to come down. Will consider this week a โdecentโ win as expectations for hikes declined even as oil prices ripped higher. The risk that these โinflection pointsโ are at risk of overloading markets remains high! SPCH SPCF LOFF SPCU SPAL SPCL SPCM This may set a โnew lowโ in terms of gobbledygook for a heading in the T-Report. Iโm hoping you are wondering what the heck triggered such an insane looking heading! Those are the ticker symbols for 7 ETFs that provide 2X the daily return of SPCX! Iโve listed them in order from largest to smallest. In total they have โonlyโ about $420 million in AUM (it was probably higher before most of these ETFs saw a decline of over 50% since their highs). SPCL tracked a larger โindexโ pre-IPO but converted to SPCX only with the IPO, presumably to be โfirst to marketโ on the SPCX-leveraged IPO. In all fairness, there are at least 4 ETFs allowing investors to short (on a leveraged basis) SPCX. Iโm kind of reminded of some vague saying about rabbits. You start with 2 (presumably male and female) and wind up with a LOT of rabbits very quickly! Does one IPO really need to inspire at least 10 single stock ETFs? Apparently (I was too lazy to pull up the tickers) there are already a half dozen or so single stock ETFs that track the newly launched SKHY ADR. It is already a complex process to price an IPO. That process can be made more difficult when only a portion of the float is sold initially. I have zero clue how any of these single stock ETFs help in terms of allocation of capital, or price discovery! If anything, they tend to amplify moves, as the leverage creates forced end of day buying or selling! That is the opposite of helping price discovery or establishing orderly markets. Iโm assuming any 4 letter combination of CHAT, OPEN, LLM, etc. has been purchased/registered in the โtickerโ world. CHTU (Chat Up) or CHTD (Chat Down) seem obvious ones to own, to sell to an ETF manager (though, again, it is likely that they are already taken). Kind of reminiscent of when people were buying up domain names hoping to sell them. I probably ranted too long to make the point that leveraged ETFs (single stock and index) tend to amplify moves in both directions! SOXL We might as well transition from market structure to the AI spend, with SOXL. 3X leverage on the NYSE Semi Index, with $19 billion of AUM, tends to amplify moves. One โcharacteristicโ of these market-structure impacting ETFs (in my opinion, and one Iโm certainly guilty of) is that: When a stock or sector that XYZ recommends goes higher, all the credit is attributed to the idea. It is all ixnay on the market structure when the market structure is helping support a move. When a stock or sector that XYZ recommends goes lower, XYZ (whoever that may be) is quick to pull out the โtechnicalsโ or โmarket structureโ card, as the culprit. Leveraged ETFs (single stock and index) and Zero-Day to Expiration Options (0DTE and other short-term options) all tend to amplify moves. It is the nature of the beast. If there is potential for stop losses to be triggered, the amplification effect is even more powerful. Is it time to buy the dip? If the parabolic move higher in the various semi-conductor indices owes some thanks to market structure (and I think it does), that market structure may have shifted from a tailwind to a headwind (or worse, a tailwind in the other direction). From Cheap Trinkets to Cheap Compute? I find myself admonishing myself and cautioning clients to put China circa 2005 out of their mind. The argument is basically that yes, China used to make โcheapโ trinkets. That much of what they made was of lower quality and fell apart. No one really picked up an item that had a โMade in Chinaโ label and thought it was fine engineering! In general, that just isnโt true in 2026. China makes high-quality components. China makes high-quality products. Intellectual Property ownership has shifted. Through a variety of methods (ranging from companies willingly exposing themselves to Chinese companies to access via less savory means), China has closed the gap on IP. In some areas, China is ahead on IP. I admit, I checked out a BYD showroom when I was in Munich last week (please try not to judge how I occupy my free time ). The vehicles looked kind of cool and I already had decent expectations, unlike I would have had a few years ago. Is Cheap Compute the โNewโ China 2005 Story? Iโm not sure why I say โ2005,โ but I do (I hope it gets the point across that we are talking about what China was like a couple of decades ago, when they really started to dominate global trade). My take is: China wedged their foot in the door by making some things at cheaper prices than anyone else could. China was willing to accept things, like pollution and horrible labor conditions, to do that. We ceded more and more manufacturing to China, and they got better and better at it. China Inc. ensured that the Chinese government and Chinese corporations were in โsync,โ making it easier to coordinate their push into global markets. Unfair advantages at home, flooding markets to lower prices abroad, etc., were tools in their arsenal to dominate global manufacturing. Over time, they made the shift from Made in China to Made by China: Selling your brands makes more money for your country and companies than just making someone elseโs brands. Globally, companies now compete with China on a much more equal footing, in terms of product quality, while China has the ability to work as China Inc. (typically an advantage). That has played a role in our construction of the ProSec theme, which we provided a mid-year update on. Increasingly, Iโm concerned that we may be at a โ2005โ moment in compute? General (ret.) Groen discussed this and much more with me on a call this week. I will paraphrase much of what was discussed: Are Chinese LLMs (and AI in general) as good as U.S. versions? In general, no. Yes, it made big news on Friday, when a Chinese model did very well on some specific benchmark tests. For all we know, it was created/trained specifically for some benchmarks, which means it will do very well on those tests, but not necessarily perform well in the โwild.โ There are allegations and questions about how Chinese models attempt to train themselves on U.S. models, greatly reducing the cost and timing of training. Bad if true, but can it be stopped? Iโm told the politically correct term for this sort of โtrainingโ is to โdistillโ their models. Is China selling compute cheaper than U.S. compute is being sold for? Yes, potentially 5 to 10 times cheaper. This is where I keep circling back to my โ2005โ date. Were dinky cars made in China as good as elsewhere? Ummm, I guess if you take out lead paint concerns and other things, sure, but the reality was we all kind of โknewโ that there were issues with them. But you could buy a pack of 5 for the price of 1 made elsewhere, and the decision got more complex (apologies to Spider, who wanted me to use snow globes, but I decided to stick with dinky cars). I do NOT have TikTok installed. Iโve only been on TEMU twice in my life. I havenโt even been tempted to try one of the Chinese models (and those of you who know me know Iโm easily tempted). Yet, kind of like the dinky cars, you can see the appeal. The Half-Life of AI model domination seems to be shrinking. Probably an overly complicated (on my part) way of saying that AI is advancing rapidly. Every time you look at the benchmarks, there are models that have risen to the top of the charts that I barely knew existed (and Iโm paying some attention). Is AI able to generate new iterations of AI even faster? If you were, say, 3 generations behind, can you get to 1 generation behind extremely quickly? Seems plausible, especially if you are willing (or able) to โcut cornersโ on training? Electricity. Does the U.S. have the electricity generation capacity to feed the demand for AI (and the general public)? Do we have the ability to get that power from where it is generated to where it needs to be (the grid)? How do we compare to China on that front? While I donโt have the details, it seems on the surface that China has more capability for โplug and playโ on the AI front than the U.S. does. Electricity, energy production, and transfer are near the TOP OF THE ProSec list for a reason. NIMBY. We have discussed the AI Revolution as much as anyone (I think). We have been arguing that Data Center and AI construction (and electricity and to a lesser extent water) would be a major political issue in 2028 or sooner! No idea that it would be an issue that is defining some primaries already! New York State seems to be imposing a 1-year moratorium. I havenโt checked how that will work, but the fact that it is a talking point tells us something. The AI industry needs to do better on community outreach. On top of everything else, there are National Security concerns at play. Iโm not sure how the industry, or the national security apparatus, changes the direction in the U.S., but they need to increase their efforts. There are plenty of areas building and pitching for more data centers and AI, so we are a long way from being out of the game, but we need to do a lot to not only protect the lead but also add to the lead (maybe the English coach could have applied that logic to the last 30 minutes as well). Iโm pretty sure that there is no equivalent of NIMBY in China. There are some things that one culture has that another culture doesnโt have, but at least there is some understanding of why you have that thing, or think that way. Iโd be willing to bet the vast majority of people in China would just stare at you blankly, bemused by the concept of NIMBY getting in the way. Bottom Line Many of the โinflectionโ points have demonstrated a clear direction to which way they are headed, but with everything going on, expect more downside for the markets. DeepSeek was a moment. Treating cheap compute like manufacturing was treated circa 2005 is NOT a โmoment.โ Iโm trying to avoid getting โsucked into the hype of the moment,โ and wish Iโd written the section on cheap compute Wednesday morning, before the recent news hit (weโd look a bit more proactive, rather than reactive), but I didnโt. The AI spend is at risk on multiple fronts, and while I expect earnings to be important, it now seems clear that even strong earnings, with very visible, very strong guidance for years to come, might not be enough! Last weekend, my perception was that earnings could propel sectors, especially the โcomputeโ sector, higher, but Iโm less convinced of that now. The story on โcompute creditโ seems to have deteriorated, even though we think it is overdone. It does seem like we might need a โdebt dietโ moment, where some company takes steps to make creditors happy, and finds that their stock responds positively to that action. While the Middle East is not helpful for lower inflation, any slowdown in the AI spend would be (though it would be awful for the economy). While Iโm loath to end with a chart, today we are going to end with a chart. While the Nasdaq 100 has been in a range for the past few weeks and has started moving lower, the S&P 500 equal weight index has been grinding higher and is extremely close to its all-time highs. In some ways it seems โcrazyโ to think that the S&P 500 equal weight and the Nasdaq 100 should have similar returns (we saw the post-Liberation Day rally in the Nasdaq 100), but finally the two indices closed the gap (noticeably the separation closed when both indices were moving lower). Who knows, by Monday, the President may have sent something on Truth Social to change all of this (he controls the Iran narrative, and he did push back on the New York data center moratorium). If he does, we should find out at the same time as everyone else, because I donโt think โfast accessโ on Truth Social has been implemented yet (Trump Media Subscription Plan). This should be an interesting week. Buy the dip, or get bearish? Of all the things listed, the concept of โcheapโ compute from China concerns me the most! Tyler Durden Sun, 07/19/2026 - 14:00
Tropical Threat Emerges In Gulf Of America With US Refineries In Potential Crosshairs The National Hurricane Center has raised the probability of tropical development for Invest 91L to 80% over the next two to seven days. The broad area of low pressure off Florida's west coast in the Gulf of America should be tracked into the new week, as early Sunday model guidance suggests it could eventually threaten offshore oil and gas rigs and coastal refineries. NHC's latest Invest 91L update: ย A area of low pressure located over the northeastern Gulf of America continues to become better defined and the associated shower and thunderstorm activity is gradually increasing. Continued gradual development is expected, and a tropical depression is likely to form later today or on Monday as the system moves slowly northward or northwestward. Interests along the northern Gulf coast from Florida, Alabama, Mississippi, and Louisiana should monitor the progress of this system, as tropical storm watches or warnings could be required for portions of the area later today. In addition, this system is expected to bring heavy rains to portions of the northern Gulf coast during the next several days. An Air Force Reserve reconnaissance aircraft is scheduled to investigate the low pressure area later today. Formation odds: Formation chance through 48 hours...high...80 percent. Formation chance through 7 days...high...80 percent. Model Track: Refinery Map: Regardless of Invest 91L's development, the broad area of low pressure will traverse northwest, away from Southwest Florida and towards the New Orleans area. The Atlantic Hurricane season is beginning to move into an active phase. Notably, El Niรฑo generally suppresses Atlantic hurricane activity by increasing upper-level westerly winds and vertical wind shear across the Caribbean and tropics, which disrupt tropical systems before they organize or intensify. More on developing El Niรฑo here. Tyler Durden Sun, 07/19/2026 - 13:25
Taylor Farms Announces Recall Of Lettuce Shipped To 27 States Over Cyclospora Risk Authored by Troy Myers via The Epoch Times, U.S. lettuce supplier Taylor Farms expanded on July 17 its voluntary recall of iceberg lettuce from central Mexico due to concerns over its link to a cyclospora outbreak affecting Americans across the country. Lettuce products that were possibly contaminated with the diarrhea-causing parasite were sent to 27 states, according to the company. The Centers for Disease Control and Prevention said July 18 the outbreak was linked to iceberg lettuce across five states and more than 1,000 people in the United States have been sickened by it. โConsumers who have purchased the recalled iceberg lettuce should discard it immediately and not consume it,โ California-based Taylor Farms said in a statement. โFull refunds are also available at the location of purchase.โ States that were sent potentially contaminated lettuce products include Alabama, Arkansas, Connecticut, Florida, Georgia, Iowa, Illinois, Indiana, Kansas, Kentucky, Louisiana, Maine, Maryland, Michigan, Missouri, Mississippi, North Carolina, New Hampshire, New Jersey, Ohio, Oklahoma, Pennsylvania, South Carolina, Tennessee, Texas, Virginia, and Wisconsin. The company added that it is actively removing affected products and ceased sourcing lettuce from the implicated lot in central Mexico. Taylor Farms reiterated that it is continuing to cooperate with the Food and Drug Administration, the CDC, and other authorities. Taylor Farms released a full list of its potentially affected products, including various shredded lettuce and salad mixes, accompanied with their lot codes and use-by dates. According to an earlier statement July 17, the company said it was voluntarily removing from U.S. markets all iceberg lettuce sourced from central Mexico. That decision was based on information the FDA provided to Taylor Farms regarding its products, the statement said. โAs a family owned and operated company, we are deeply concerned for those who became ill, their families, and the many Americans whose trust in the safety of their fresh produce has been shaken,โ Taylor Farms said. โThat trust is something weโve worked for decades to earn, and we are committed to doing everything in our power to restore that confidence.โ U.S. health officials earlier this week announced it had traced โa single supplier of iceberg lettuce from Mexico used by Taco Bell locations where sick people ate before becoming ill.โ The popular fast-food chain addressed the issue in a statement on X, writing that it had removed all Taylor Farm products from its locations. โWe want you to hear it from us,โ the company said. โSo go ahead and enjoy your Taco Bell today.โ U.S. food distributor Sysco had also pulled all its Taylor Farms iceberg lettuce sourced from central Mexico. Cyclosporiasis is caused by the microscopic parasite cyclospora that stems from produce and water contaminated with human feces. As of the latest update available from the CDC, the outbreak has resulted in 1,644 cases and 94 hospitalizations. The agency noted that the true number of cases could be much higher, as many people are able to recover without medical care and are not tested for cyclospora. Tyler Durden Sun, 07/19/2026 - 12:50
Zelensky Boasts Of Hits On 3 Russian 'Shadow Fleet' Tankers After Kiev Military Depot Leveled Russia overnight and into early Sunday morning unleashed what Ukrainian officials have described as its largest barrage of ballistic missiles focused on Kiev of the war. Among the key targets was a large complex owned and operated by UKRTAC, a Ukrainian producer of military gear and personal protective equipment. Its manufacturing facilities and warehouses in the capital were destroyed in the assault. Ukrainian presidency's office UKRTAC,ย confirmedย on social media that facilities and production equipment were entirely wiped out due to a direct hit, but said there were no casualties at the plant. But elsewhere in the capital there were significant casualties, with at least one person killed and 16 others wounded.ย President Volodymyr Zelensky said Sunday it was one of the single biggest missile barrages of the war: A series of powerful explosions thundered across the darkened city in an attack that involved 41 missiles of various types and ravaged buildings across several districts. Also, Ukrainian foreign minister Andrii Sybiha has described the raid involved "around four dozen" ballistic missiles. However, Zelensky has said Ukrainian forces continue to strike back, including a new attack onย three oil depots and a fuel facility inย Russia's southern Stavropolย Sunday morning. Zelensky addedย three Russian 'shadow fleet' tankers were also struck in the Black Sea. "Today, Ukraineโs long-range sanctions reached designated targets that support and finance Russiaโs aggression. SSU units struck three oil depots in the Stavropol region at once, while units of our Armed Forces hit another fuel facility in the same region," he said. "Direct hits were recorded on three Russian shadow fleet tankers in the Black Sea." We continue to respond to Russian strikes in a fully justified and accurate manner. Today, Ukraineโs long-range sanctions reached designated targets that support and finance Russiaโs aggression. SSU units struck three oil depots in the Stavropol region at once, while units of ourโฆ pic.twitter.com/3YWEDFBknE โ Volodymyr Zelenskyy / ะะพะปะพะดะธะผะธั ะะตะปะตะฝััะบะธะน (@ZelenskyyUa) July 19, 2026 The Kremlin over the weekend has boasted of new ground advances along the front lines, but the war has long been focused in the air of late. As for the Ukrainian capital, emergency crews have been scrambling on an almost nightly basis of late. Concerning the new attack, "The Kyiv government said firefighters were responding to blazes in five different districts after the attack,ย one of the biggest in recent weeks, hit residential buildings, office and industrial sites, a dormitory and vehicles," The Independent describes. Tyler Durden Sun, 07/19/2026 - 12:15
The Day Democrats Return To Wall Street ย Submitted by QTR's Fringe Finance If thereโs one thing the market has become spectacularly good at, itโs extrapolating todayโs political environment out to infinityโฆand โpricing inโ things (like the end of the Iran War, Trump Tax cuts, future rate cuts, etc.) over and over again until one day of bad news turns into a year of โmarket optimismโ about things getting better, prompting the S&P to rally to new all-time highs 17 days in a row despite the fact that nothing has changed on the ground. And today, the market is acting as though Washingtonโs current regulatory philosophy, which appears to beย pardon everyone who has ever committed obvious white collar fraudย and take a โhands offโ approach to new enforcement is simply...permanent. But it isnโt. I think one of the biggest risks investors arenโt even remotely considering is what happens if Democrats take back the House this fall, or even more importantly, win the White House in 2028? Thereโs a real chance this could lead to a wholesale repricing of the excesses that have flourished under an administration that has made it abundantly clear it would rather participate in markets than police them. I still think Trump is a better option than anyone on the left side of the aisle when it comes to the Presidency. But ironically, Trump may have handed Democrats the perfect campaign issue. Forget whatever slogan focus groups come up with next. The message practically writes itself: corruption, self-dealing, insider enrichment, pardons for politically connected allies, regulatory capture, and a government that looked the other way while fortunes have been minted in increasingly questionable ways. Republicans, including myself, cried wolf over Nancy Pelosiโs trading, Hunter Bidenโs take from Burisma and President Bidenโs dealings with China. But this administrationโs cash grab, flanking all areas of the marketย from cryptoย toย hi-speed algorithmic trading, puts Biden to shame. If Democrats regain power, I think there is virtually no chance theyย donโtย campaign on restoring โaccountabilityโ to Wall Street, crypto and corporate America. Sadly, theyย will point to the effects of monetary policy widening the wealth gapย thatย their party also supportsย as an impetus for change. And unlike every election where politicians promise to lower drug prices by seventeen cents, accountability on Wall Street is actually something they can begin doing on Day One simply by changing whoโs running the agencies. Markets arenโt remotely priced for that. Todayโs investors have become accustomed to a free-for-all style environment where enforcement often feels optional and the line between financial innovation and outright promotional nonsense has become increasingly difficult to distinguish. Insider trading on prediction markets, one crypto scam after the next, trillion dollar IPOs for companies that arenโt profitable, NBA players rigging games, members of congress insider trading. Itโs theย wild fucking westย out there. Thatโs part of the reasonย Iโm curtailing being an active trader. Itโs all rigged. Crypto has been one of the biggest beneficiaries. The industry has spent years fighting regulators, only to find itself operating in perhaps the friendliest environment it has ever enjoyed. Thatย can change astonishingly fast. A Democratic administration would almost certainly take a much harder look at exchanges, stablecoins, token offerings, disclosures, anti-money laundering rules, and whether many digital assets belong under securities laws. (Read:ย The Crypto Risk No One Is Discussing) The important part isnโt whether every proposal succeeds. Markets donโt wait for legislation. They price expectations. Somewhere along the way, Wall Street seems to have decided that โvisionโ is a perfectly acceptable substitute for profits, promotional CEOs deserve celebrity status, retail investors exist primarily as liquidity providers, and every wildly optimistic forecast deserves a trillion-dollar valuation until proven otherwise. Questionable accounting? Bullish. Aggressive projections? Bullish. Financial engineering? Bullish. Serialย dilution? Bullish. Itโs become difficult to tell whether weโre allocating capital or handing out participation trophies for coming up with nonsense stories. Every new S-1 reads like sci-fi pulp fiction L. Ron Hubbard wrote for a penny a page back in 1964. ๐ฅย 50% OFF FOR LIFE:ย Using this coupon entitles you to 50% off an annual subscription toย Fringe Financeย for life:ย Get 50% off forever And a Democratic administration doesnโt need to outlaw any of this. It simply needs regulators that remember their job description. Suddenly the SEC starts asking questions again. The DOJ develops an interest in corporate misconduct. Congressional hearings reappear. Enforcement actions increase. Companies spend more time paying lawyers and less time posting rocket emojis. That alone changes behavior. Speculative markets donโt like adults entering the room. Could Democrats also push for higher corporate taxes, tougher antitrust enforcement, tighter rules around buybacks or capital gains changes? Of course. Maybe half those ideas never become law, but the possibility alone forces investors to rethink the multiples theyโre willing to pay for companies whose business models seem to depend on regulators never opening the filing cabinet. Whatโs remarkable is how few people seem prepared for any of this. The consensus appears to be that todayโs regulatory climate is simply the new normal forever. Thatโs usually how markets get blindsided. Nobody spends years saying, โCareful, the political pendulum might swing.โ Instead, everyone gets caught leaning the same direction until suddenly theyโre not. The oleโ โwhen the music stopsโ adage comes to mindโฆ And listen, quite frankly, I donโt think Democrats know much about finance. Theyโve historically shown theyโre more than capable of misunderstanding markets, overregulating industries and proposing policies that create plenty of unintended consequences. But Iโll also admit something that probably wonโt make me popular: it wouldnโt be the worst thing in the world if a few very obvious frauds were actually held accountable again. If we truly lived in a pure free market where everyone understood the risks, did their own homework, and accepted the consequences of losing money, Iโd be perfectly happy letting buyers beware and pulling all regulation. But thatโs not the market we have. We have companies that wind up inside passive index funds. Retirement accounts own them automatically. Pension funds own them automatically.ย Millions of ordinary investors buy them without ever making an active decision because theyโre simply embedded inside ETFs and benchmark indexes. When obvious promotional garbage gets institutionalized like that, it stops being just another speculative bet between consenting adultsโฆpeople are getting screwed without even realizing theyโre participating. Thatโs the part that has me, somewhat reluctantly, yearning for just a little more regulation than weโve got today. -- QTRโs Disclaimer:ย Please read my full legal disclaimerย on my About page here.ย This post represents my opinions only.ย In addition, please understand I am an idiot and often get things wrong and lose money. I may own or transact in any names mentioned in this piece at any time without warning. Contributor posts and aggregated posts have been hand selected by me, have not been fact checked and are the opinions of their authors. They are either submitted to QTR by their author, reprinted under aย Creative Commons licenseย with my best effort to uphold what the license asks, or with the permission of the author. This is not a recommendation to buy or sell any stocks or securities, just my opinions. I often lose money on positions I trade/invest in. I may add any name mentioned in this article and sell any name mentioned in this piece at any time, without further warning. None of this is a solicitation to buy or sell securities. I may or may not own names I write about and are watching. Sometimes Iโm bullish without owning things, sometimes Iโm bearish and do own things. Just assume my positions could be exactly the opposite of what you think they are just in case. If Iโm long I could quickly be short and vice versa. I wonโt update my positions. As of May 20, 2026 I am attempting to no longer actively trade (read my story here). Myย investing/saving is mostly done by recurring contributions mostly to sector ETFs and a few select equities, trusted third parties who oversee my accounts, and advisors. Such advisors or funds, through individual equities, options, index funds, mutual funds, ETFs, or other securities, may have positions in, exposure to, or holdings of names mentioned herein that I know nothing about. Basically, via index funds, ETFs and individual equities it is possible I could own, have exposure to, or not own anything at any point. As of the same date, May 20, 2026, in anย attempt to lead a healthier lifestyle, Iโve also excluded myself from fantasy sports, sports betting, online and in-person casinos and prediction markets. And all positions can change immediately as soon as I publish this, with or without notice and at any point I can be long, short or neutral on any position. You are on your own. Do not make decisions based on my blog. I exist on the fringe. If you see numbers and calculations of any sort, assume they are wrong and double check them. I failed Algebra in 8th grade and topped off my high school math accolades by getting a D- in remedial Calculus my senior year, before becoming an English major in college so I could bullshit my way through things easier. The publisher does not guarantee the accuracy or completeness of the information provided in this page. These are not the opinions of any of my employers, partners, or associates. I did my best to be honest about my disclosures but canโt guarantee I am right; I write these posts after a couple beers sometimes. I edit after my posts are published because Iโm impatient and lazy, so if you see a typo, check back in a half hour. Also, I just straight up get shit wrong a lot. I mention it twice because itโs that important. Tyler Durden Sun, 07/19/2026 - 11:40
Social Media Influencers Andrew And Tristan Tate Arrested In Miami Viaย Headline USA, Influencer brothers Andrew and Tristan Tate, whose social media empire promoting wealth, male dominance and misogyny has made them among the worldโs most polarizing internet personalities, were arrested Saturday in Miami as British authorities sought their extradition on rape and sex trafficking charges. The brothers were taken into custody by the U.S. Marshals Service on a sealed warrant, placing the United States at the center of an international legal saga that has stretched from Romania to Britain. STATEMENT OF JOSEPH D. MCBRIDE, ATTORNEY FOR ANDREW AND TRISTAN TATE The world knows Andrew and Tristan Tate are innocent. Their enemies know it best of all. That is exactly why they have been attacked. Last night, the United Kingdom, a corrupt and fallen nation that jails itsโฆ pic.twitter.com/XoMVmPxtgA โ Joe McBride (@McBrideLawNYC) July 19, 2026 Britain is seeking their extradition on rape and trafficking charges British prosecutors announced Saturday that they were seeking the brothersโ extradition on charges alleging they raped and trafficked women between 2010 and 2017. The dual U.S. and British citizens moved to Romania in 2016. They were arrested there in 2022, accused of participating in schemes to lure women for sexual exploitation. They denied those allegations and the Romanian case hasnโt gone forward because of legal and procedural problems. Last year, they were allowed to leave Romania and flew to Florida on a private jet. The brothers are expected to appear in Miamiโs federal court early next week, according to a person familiar with the matter who spoke to on the condition of anonymity to discuss sensitive law enforcement operations. The pending charges in the United Kingdom accused the brothers of abusing women in an area north of London, where they grew up. Their lawyers had said they denied the allegations. Joseph McBride, an attorney representing the Tate brothers, said in a phone interview Saturday evening that he has not been able to speak with his clients but called the new charges out of the U.K. โfilth and slanderโ intended to derail defamation lawsuits filed by the brothers in the U.S. โTheyโre pulling out all the stops to make sure these guys never get their day in court,โ McBride said. โWe are confident that once a competent judge sees the facts, and once the Department of Justice confronts this egregious abuse of its own authority, Andrew and Tristan Tate will walk free. America does not do Britainโs political dirty work.โ pic.twitter.com/qSAn0cywI2 โ Joe McBride (@McBrideLawNYC) July 19, 2026 Tate has been banned from social media platforms for โhate speechโ Andrew Tate, 39, first reached a mainstream audience as a contestant on the U.K. reality television show โBig Brotherโ in 2016. He was removed from the show when a video surfaced that appeared to show Tate assaulting a woman. He and his brother Tristan Tate, 38, are vocal supporters of U.S. President Donald Trump. Andrew Tate has amassed over 10 million followers on X but has been banned from platforms like YouTube, TikTok and Instagram for violating โhate speechโ guidelines. The Tate brothers have consistently denied allegations of abuse and human trafficking, claiming that violent and misogynistic statements have been taken out of context or were intended as jokes. In a statement Saturday, the U.K.โs Crown Prosecution Services said that in addition to the charges publicly announced against the brothers in 2025, involving alleged crimes against three women, it was bringing a total of 38 new charges related to โfour further victims.โ Both brothers are accused of rape and human trafficking. Andrew Tate faces an additional charge of profiting from prostitution, and 19 charges โfor offences relating to indecent images of a child and extreme pornography,โ according to U.K. authorities. Tyler Durden Sun, 07/19/2026 - 10:30
