Short positions in U.S. equities have surged to a record high, as the 'wall of fear' facing the bull market grows ever taller.
As risks associated with artificial intelligence intensify, short-selling activity in U.S. equities has reached a record high.
World's third most powerful, with 'fable-level immersion'! Kimi K3 with 2.8 trillion parameters sets a new open-source record
In evaluations, Kimi K3's overall intelligence level is close to that of the world's leading closed-source models. According to GDPval-AA v2 data on knowledge work benchmarks, Kimi K3 scored higher than Claude Opus 4.8 Max, ranking just behind Claude Fable 5 Max and GPT-5.6 Sol Max. Prior to the model's official release under an open-weights license, reports indicated that Moonshot AI internally regarded it as a key product targeting Anthropic's latest flagship model, Claude Opus 4.8.
Bank of America CEO joins chorus of concern, warning that Mythos is threatening the security of the financial system
Bank of America CEO Moynihan warned that Anthropic's AI model Mythos is significantly increasing the workload and response pressure on financial institutions' defense systems. Bank of America has received limited access for defensive testing and to share information with peers. Previously, JPMorgan CEO Dimon issued a warning about Mythos, comparing the broad public release of the system to 'giving ballistic missiles to individuals.'
ASML Holding ramps up production, Taiwan Semiconductor increases investment, and Anthropic achieves a profitability leap: the 'AI compute glut' narrative faces pushback—has the AI bull market entered the era of inference-driven compounding?
From massive orders for photolithography machines to Anthropic's soaring profits, data centers are moving beyond the era of pure cash burn, as AI computing infrastructure enters an epoch of cash flow revaluation. Exponential View estimates that generative AI generated approximately $110 billion in revenue over the past 12 months, with a latest monthly annualized revenue run rate of around $175 billion; SemiAnalysis forecasts that Anthropic will achieve over $1 billion in profit in Q3 2026.
Japan Aims to Build a 'National Robotics Team,' Plans to Procure 27,500 NVIDIA Rubin AI Chips as the AI Computing Super-Cycle Shifts from Cloud to 'Physical AI'
Japan plans to purchase NVIDIA's next-generation Rubin chips to build its own foundational artificial intelligence model for robotics. Noetra will oversee the project, construct a large-scale data center, and release the AI model by March of next year, followed by regular updates.
Anthropic is racing toward an IPO, potentially listing as early as October, which could allow it to beat OpenAI to the public markets.
Anthropic is accelerating its IPO process, planning to hold intensive investor meetings in the coming weeks and could list as early as October this year. The company’s valuation has now risen to $965 billion, with Morgan Stanley, Goldman Sachs, and JPMorgan serving as joint lead underwriters. Strong revenue growth from its enterprise AI business provides solid support for the listing, though regulatory uncertainty remains the key variable affecting its subsequent performance.
For the first time in history, are people cheaper than software? If AI isn't used effectively, it can be far more expensive than human labor!
The AI boom is fueling a striking paradox: a technology originally intended to replace human labor is not only avoiding layoffs but has also driven software costs above personnel costs for the first time. Ninety-nine percent of AI revenue remains confined to programming, while other sectors are mired in a 'management failure' quagmire. Overnight, enterprises have 'hired' millions of AI employees—yet no one knows how to manage them. History is repeating itself: the next trillion-dollar opportunity belongs to those who truly know how to keep the trains running on time.
ASML Holding has factored the AI computing boom into its 2028 capacity expansion plan! Wall Street bets that AMD (AMD.US), holding both CPU and GPU capabilities, will dominate the AI inference era.
Orders for lithography machines are the 'ultimate lie detector' of AI computing demand: ASML Holding's robust guidance of €45 billion has significantly reinforced the bullish AI narrative, while Goldman Sachs raised AMD's price target to $640, collectively underscoring that the AI compute boom remains intact, with capital shifting from NVIDIA's solo performance toward a broad-based expansion across the entire AI infrastructure stack.
OpenAI's first hardware product has been revealed, with plans to launch a smart speaker next year, though litigation from Apple could disrupt the release timeline.
Reports indicate that OpenAI’s screenless smart speaker will integrate ChatGPT capabilities, supporting smart home control, media playback, and message handling. The device features a next-generation GPT-Live voice mode and includes mechanical components capable of producing dynamic movements, deepening personalized experiences over time. It is scheduled for announcement this year and official launch in 2027, but Apple has filed a lawsuit alleging theft of trade secrets, casting uncertainty over the release timeline.
Express News | OpenAI plans to launch a smart home speaker as its first hardware product.
BofA Fund Manager Survey: AI Bubble Now the Market's Top Concern!
Bank of America’s July Global Fund Manager Survey shows that 'AI bubble' has surged to become the top tail risk for the first time, cited by 45% of respondents (up from 28% last month), surpassing 'second-wave inflation.' A record-high 82% of respondents identified 'long global semiconductors' as the most crowded trade in the current market. The Bull & Bear indicator reached an extremely bullish reading of 9.4, yet most investors have not taken short positions in AI-related assets, revealing a clear divergence between sentiment and positioning.
AI Computing Bottleneck Shifts Upstream! Behind Micron's Investment in GlobalWafers: The Memory Super-Cycle Moves from 'Chip Shortage' to 'Wafer Rush'
The supply of silicon wafers—the core raw material required for expanding chip manufacturing capacity—could ultimately become another 'critical AI computing hardware' bottleneck for the AI computing industry chain and the global technology sector at large.
Express News | Ministry of Foreign Affairs: **** will attend the opening ceremony of the 2026 World Artificial Intelligence Conference and the High-Level Meeting on Global Governance of Artificial Intelligence and deliver a keynote speech.
GPT-5.6 fully launched! New ChatGPT agent released simultaneously; Altman stated that 'many adjustments' were made following discussions with the U.S. government.
OpenAI has fully launched three new GPT-5.6 models, which outperform Anthropic's Fable 5 in benchmark tests; the flagship Sol model costs one-fourth as much as Fable 5, while Terra and Luna cost only one-sixteenth as much. On the same day, OpenAI introduced ChatGPT Work agents, capable of working continuously across applications for multiple hours. Altman did not specify the exact nature of recent adjustments and also stated that it remains uncertain whether the company will go public this year. GPT-5.6 achieves a 54% improvement in token efficiency for agent programming, and open-source models from China are becoming 'very good.'
Former Federal Reserve Chair Ben Bernanke joins Anthropic's governance board, focusing on AI's impact on the economy and employment.
① Anthropic announced that former Federal Reserve Chair Ben Bernanke has joined its oversight body, the 'Long-Term Benefit Trust,' which holds the authority to appoint and remove certain members of the company’s board of directors; ② Bernanke, who led the Federal Reserve through the 2008 financial crisis and was awarded the Nobel Memorial Prize in Economic Sciences in 2022, will provide expert guidance to Anthropic on economic and labor-related issues associated with artificial intelligence.
Betting on Voice as the AI Gateway! OpenAI Launches New Model GPT-Live, Ushering ChatGPT into the Era of Real-Time Conversations
GPT-Live is available on web, iOS, and Android platforms, with API integration planned for the future. It offers real-time translation, dictation, and adjustable speech speed, and can seamlessly invoke advanced models such as GPT-5.5 in the background to perform complex reasoning and conduct internet searches. Video and screen-sharing voice capabilities are not supported at launch. Two versions—Live-1 and Live-1 mini—are initially targeted at paying and free users, respectively, and both demonstrated significantly superior performance compared to Advanced Voice Mode in blind one-on-one tests lasting 5–10 minutes. OpenAI has also strengthened voice safety mechanisms, enabling the system to proactively terminate conversations in high-risk scenarios and notify parents. The company reports that over 150 million people already use voice features weekly to interact with ChatGPT.
Regulatory Easing for GPT-5.6: OpenAI Authorized for Full Deployment, Marking the End of the Interim Approval Regime for Frontier AI
OpenAI's flagship model GPT-5.6 has received formal approval from the U.S. government, transitioning from a restricted release limited to 20 partners to full commercialization, becoming the first case to complete the entire review process under the U.S. frontier AI oversight framework. The flagship model Sol will launch this Thursday, with API pricing approximately half that of Anthropic’s competing offering and superior performance. This move marks a new phase in U.S. AI regulation, with OpenAI securing a first-mover advantage in commercialization through proactive engagement and making a strong push into the enterprise API market.
Structural divergence behind the AI boom intensifies: 59% of S&P tech stocks have already entered bear market territory
The AI sector is undergoing its most intense internal adjustment since the COVID-19 pandemic. According to Goldman Sachs, momentum trades related to AI have retraced 27% from their late-June peak, and 59% of S&P technology stocks have already entered bear market territory. Market capital is shifting from semiconductors to cloud computing, accelerating the transition in AI investment logic.
Goldman Sachs: AI investment focus shifting toward physical industries; capital expenditures could reach $7.6 trillion over the next six years
① AI expansion is extending beyond computing power and data centers into real-world applications such as manufacturing, energy, logistics, defense, and life sciences; ② AI supply chain constraints are shifting from a shortage of chips to shortages of electricity, equipment, and labor; ③ Declining model inference pricing is accelerating software revaluation, with industrial digital twins, warehouse automation, and unmanned systems among the first to realize commercial value, while financing mechanisms for physical AI remain underdeveloped.
AI trading sentiment shifts! Morgan Stanley weighs in: momentum behind chip stocks is waning, with capital expected to rotate into hyperscale cloud providers.
Wilson noted that as investors shift capital toward previously underperforming sectors—including AI hyperscale cloud computing providers—the upward momentum in semiconductor stocks is waning.