❯ Nvidia board reportedly prepares to remove Jensen Huang as a second alleged affair intensifies succession discussions
Directors debate his futureNvidia’s leadership crisis escalated on November 11 in the United States. The Wall Street Journal reported that directors were preparing to remove founder and CEO Jensen Huang; Reuters subsequently learned independently that the board was meeting that day to decide whether to do so. No formal removal or successor appointment had been announced by the morning of November 12 in Beijing, leaving the outcome unresolved.
A second relationship is reportedEarlier that day, The New York Times cited five people who had worked at Nvidia as saying Huang had an extramarital relationship with healthcare vice president Kimberly Powell, a direct report, and that it had been discussed inside the company. Reuters said it could not immediately verify the account, while Nvidia and Powell did not immediately respond to requests for comment. The reported involvement of a subordinate adds a direct question about workplace power and corporate governance to the controversy.
An investigation follows earlier supportThe controversy previously centered on Huang’s relationship with businesswoman Alice Huang. In a November 8 staff memo, he acknowledged infidelity and took responsibility; that day, Nvidia’s board announced an investigation overseen by independent directors with outside independent counsel, with Huang cooperating. Directors then expressed confidence in him and his leadership team and said they had no reason to believe company secrets had been compromised, while committing to examine available information.
An outside search takes shapeOfficeChai’s account of Bloomberg’s reporting says directors have been working with an outside firm to identify possible successors and are weighing an interim CEO during the permanent search. The report cites an unnamed person familiar with the situation; the firm, candidates and potential interim arrangements have not been disclosed. An interim appointment could allow directors to address immediate leadership questions before completing a longer search.
A founder-centered organizationHuang has led Nvidia since co-founding it in 1993, overseeing its development from a graphics-chip maker into a central AI computing supplier. Reuters notes that he owns about 3.6% of its shares, has dozens of direct reports and has no obvious successor. Replacing him would therefore entail handing over technical decisions, management coordination and customer relationships, a transition that extends well beyond naming a new executive.
Markets price in uncertaintyNvidia shares fell about 8% in after-hours trading on November 11 following the removal reports. Investors and analysts interviewed by Reuters did not necessarily expect immediate damage to chip demand, but the developments raised questions about reliance on the founder and effects on the company’s partnerships and investment network. The board’s decision, the investigation’s findings and any transition leadership or division of responsibilities remain unresolved.
▮ SIGNALHuang’s concentration of technical and commercial authority makes the board’s decision a test of both Nvidia’s succession arrangements and its ability to operate independently of its founder.
❯ Anthropic IPO draws orders several times the shares available as bankers consider a valuation of $1.5 trillion or more
Large orders arrive earlyDemand for Anthropic’s planned listing has strengthened soon after the roadshow began. Bloomberg reported on November 11 that indications of interest amounted to several times the available shares, driven primarily by large institutional orders; the exact multiple was not disclosed. Marketing continues, and the final price, allocations and proceeds have not been determined.
The filed offering takes shapeClaude developer Anthropic launched its IPO roadshow on November 10, proposing 825 million Class A shares at $70–$78 each and applying to list on Nasdaq as ANT. About 801.35 million shares would be sold by the company and 23.65 million by existing shareholders, whose proceeds would not go to Anthropic. At the range’s upper end, the base share sale would total approximately $64.35 billion, with an additional underwriter option available.
Bankers consider a higher priceAccording to Yahoo Finance’s account of Financial Times reporting, strong interest has prompted bankers to consider pricing above the filed range, potentially taking the fully diluted valuation to $1.5 trillion or more. That measure includes potential shares such as equity options; the current range’s upper end implies roughly $1.4 trillion. A higher price remains under discussion, and the reported talks do not themselves amend the announced $70–$78 range.
Capital for computing capacityThe listing also includes a planned $26 billion private placement. Bloomberg, citing filings, puts the combined IPO and placement at up to $90.4 billion, a total that includes existing shareholders’ sales. Coverage of the Financial Times report says Anthropic wants additional funds for computing expenditure. Cash available for operations determines how much capacity it can secure for training models and serving Claude, so a higher corporate valuation and money actually received by the business are distinct outcomes.
The pitch reaches into medicineIn the roadshow video released on November 10, Dario Amodei predicted that cancer, Alzheimer’s, heart disease and mental illness could become problems of the past within a decade. Business Insider reported objections from medical professionals who see potential for AI in research but dispute that sweeping timetable. It also noted that Anthropic has begun drug discovery work; those efforts and Amodei’s forecast are not demonstrated treatment outcomes.
Investor meetings continueBloomberg says Amodei and JPMorgan Chase CEO Jamie Dimon are scheduled for an investor livestream on Friday, November 13, while Goldman Sachs and Morgan Stanley lead the offering. Demand gives the banks room to consider adjusting terms, but indications of interest still need to become final allocations, and neither the IPO nor the accompanying financing has closed. Public investors will also assess how Claude’s existing business supports continuing investment and how longer-term medical research might eventually produce commercial results.
▮ SIGNALInstitutional demand strengthens Anthropic’s ability to finance computing expansion, while its medical ambitions extend public expectations well beyond revenue from its models.
❯ Google launches AlphaGenome 2 to predict the effects of DNA variants in specific cells
A closer view of cellsGoogle DeepMind released AlphaGenome 2 on November 11 to predict how DNA variants affect particular cell types and states. Trained on more than 100 million individual cell profiles across 68 tissues, it helps researchers ask why the same genetic change affects one kind of immune cell much more than another.
Beyond tissue averagesThe first AlphaGenome predicted molecular activity across broader tissues, which can contain many different cell types. The new model takes a target biological condition alongside the DNA sequence and can use small samples of newly measured cells to predict unfamiliar states without retraining. This may reveal effects that are hidden when researchers average across an entire tissue.
Improvements in research testsDeepMind reports that, on the TenK10K dataset covering 28 primary immune-cell types, AlphaGenome 2 identified 30% of causal gene-activity variants versus 16% for its predecessor adapted to single-cell data. It also computationally reproduced a known inflammatory bowel disease mechanism involving increased ETS2 activity in gut macrophages. Such predictions can narrow experimental priorities, while biological mechanisms still require laboratory validation.
Research access opensThe API and tutorials are available for noncommercial research, with commercial research in private preview. The model has not been validated or approved for clinical or diagnostic use. Its immediate role is to help researchers prioritize variants, genes and cells for investigation, rather than supply ready-to-use treatments.
▮ SIGNALIdentifying the cells in which a genetic variant matters could reduce unproductive early experiments and make subsequent drug research more targeted.
❯ Alibaba opens Qwen4-Max preview through its office, coding and chat products
Early access to a new generationAlibaba announced on November 11 that Qwen4 is coming and opened the Qwen4-Max preview. QwenWork, Qoder, the Qwen app and QwenStudio have integrated it, with developer access through Alibaba’s Token Plan. The Max preview is available now, while the complete rollout schedule for the broader Qwen4 family remains undisclosed.
Understanding and actingAlibaba Cloud’s documentation describes a flagship multimodal model with text generation, image and video understanding, and reasoning enabled by default. It supports tools, web search and structured output for coding, office work, frontend design and complex agent tasks. Applications can combine reading screenshots or documents with subsequent tool actions, subject to the tools and permissions they provide.
A defined preview windowAPI access currently requires Token Plan rather than pay-as-you-go billing. The preview runs from November 11 at 10:00 through November 25 at 24:00 Beijing time, after which this preview model will be retired. Alibaba says it will remain unchanged during that period, giving developers a fixed version to evaluate; arrangements for its successor remain to be announced.
Compatibility eases evaluationQwen4-Max-Preview supports multiple OpenAI and Anthropic interface protocols, allowing developers to configure it in existing clients and applications. Reasoning and final answers share an output budget, so a limit designed only for the answer can prematurely end longer tasks. Launching across office, coding and chat products also gives Alibaba access to feedback from varied real-world workflows.
▮ SIGNALQwen4’s arrival inside established work products exposes its practical task performance earlier than a standalone chat demonstration would.
❯ ByteDance reportedly prepares a Muse-like assistant as Zhang Yiming sets a 2029–2030 frontier research goal
A launch within weeksBloomberg reported on November 11 that ByteDance plans to release an assistant similar to Meta’s Muse in the coming weeks. At a recent AI all-hands meeting, Zhang Yiming reportedly outlined the plan and a goal of competing with Anthropic and OpenAI at the research frontier by 2029 or 2030. ByteDance did not immediately comment, and the product’s name and exact release date remain undisclosed.
Voice and everyday assistanceMuse, launched by Meta in September, combines voice interaction, an animated avatar and task assistance, with distribution through hardware including smart glasses. Bloomberg uses it to describe the role envisaged for ByteDance’s product, while noting that accompanying hardware plans are unclear. The direction emphasizes continuing help with daily activities beyond individual question-and-answer exchanges.
Bringing research togetherByteDance already has Doubao, its Seed foundation models and Seedance video models, giving the proposed assistant several capabilities it could bring into one experience. Bloomberg describes a Seed team of roughly 1,000 researchers and engineers led by Wu Yonghui, working on language, video and images as well as robotics and other hardware-related AI. Turning those capabilities into an assistant also requires memory, tool use and coordination across tasks.
Two development schedulesZhang reportedly emphasized long-term investment and original research, rather than relying excessively on distillation from other models’ outputs. The 2029–2030 ambition concerns frontier research, while the assistant is planned much sooner. Its initial features and ability to carry out tasks consistently remain to be shown at launch.
▮ SIGNALA unified assistant would let ByteDance test its separate model efforts in one user setting, where coordination between tasks will help determine whether people build lasting habits around it.
❯ ByteDance AI chip chief Shi Yunfeng reportedly enters departure handover as chip adaptation work continues
A leadership handoverLatePost reported on November 11 that Shi Yunfeng, who leads ByteDance’s AI chip and DPU teams, has submitted his resignation and is handing over responsibilities. He joined in 2018 after roles at MediaTek, Marvell and HXT, and reports to ByteDance vice president and Volcano Engine head Yang Zhenyuan. No successor or handover completion date was disclosed.
A specialized chip strategyEstablished in 2019, the team moved from AI Lab into the Data organization before relevant teams began reporting directly to Yang this year. It follows an application-specific chip approach similar to Google’s TPU. LatePost says chips completed fabrication this year but performed below expectations, while work on a subsequent generation continues.
Software work remains essentialAda for Doubao and SeedChip for Seedance are undergoing operator adaptation, making model computations run correctly and efficiently on the hardware. Specialized chips cannot simply use the complete CUDA software ecosystem, so models, frameworks and underlying software need adaptation after fabrication. LatePost reports that the chips have not been deployed widely inside ByteDance, with migration and debugging requiring participation from product teams.
External technology remains part of the planByteDance is also investing and partnering externally, including with ReRAM memory specialist Xinyuan Semiconductor. LatePost says a proposed integration of another invested chip project into the Picoheart team has not closed. The personnel change does not itself end internal chip development: software adaptation, next-generation designs and outside technology integration remain at different stages.
▮ SIGNALGetting ByteDance’s chips into real workloads requires joint investment by hardware, model and software teams, making organizational coordination part of the deployment challenge.
❯ CoreWeave seeks at least $4 billion in investment-grade borrowing backed by customer contracts
A potential deal next weekAccording to Bloomberg, CoreWeave is preparing at least $4 billion of investment-grade borrowing, led by Morgan Stanley and MUFG, potentially syndicated in stages starting next week. Management previously discussed a 7%–7.5% yield, roughly half the trading yield on its existing high-yield bonds. The financing is not final, and size, pricing and timing could change.
Construction precedes customer receiptsCoreWeave supplies cloud resources for training and running AI models, buying chips and deploying data centers before recovering investment through customer usage. Its third-quarter results show revenue of $3.661 billion, net interest expense of $846 million and a $709 million net loss. Full-year capital expenditure guidance of $37 billion–$39 billion makes borrowing costs consequential for expansion returns.
Assets and contracts support creditInvestment-grade treatment concerns the proposed financing structure, rather than a comprehensive upgrade of CoreWeave itself. The report says the company has obtained such ratings for some arrangements backed by chips and customer contracts, an approach it plans to continue. Lenders can assess particular equipment and contracted cash flows rather than rely solely on the corporate balance sheet.
Funding must match deliveryCoreWeave reported $132.5 billion of revenue backlog at September’s end, raised annual revenue guidance to $13 billion–$13.3 billion and expects about $9 billion of customer prepayments and deposits over the following three months. Backlog conversion still depends on delivery and service availability. If completed on anticipated terms, the loan would help fund additional infrastructure at a lower cost before all related customer revenue arrives.
▮ SIGNALTurning customer contracts into cheaper financing directly affects CoreWeave’s ability to fulfill orders, making strong customer credit an expansion resource in its own right.
❯ Applied Intuition discusses raising at least $1.5 billion at a $30 billion pre-money valuation
Terms remain under discussionInvesting.com, citing Bloomberg, reports that Applied Intuition is negotiating at least $1.5 billion in new funding, led by General Catalyst with a planned $250 million commitment. The proposed $30 billion valuation is before the new investment. The round has not been announced as closed, and final size and terms may change.
Infrastructure for vehicle developersFounded in 2017, Applied Intuition sells simulation tools, a vehicle operating system and autonomy software to help customers develop, test and deploy intelligent vehicles. Simulation lets engineers repeat road and operating scenarios virtually, reducing reliance on physical tests after every change. In its previous financing announcement, the company said 18 of the world’s top 20 automakers use its products, with applications also spanning trucking, construction, mining and agriculture.
A proposed doubling of valuationIn June 2025, the company announced a $600 million Series F financing and tender offer at a $15 billion valuation, co-led by BlackRock-managed funds and Kleiner Perkins. The current proposed pre-money valuation is roughly double that level, and General Catalyst is already an investor. Development tools, vehicle operating software and autonomy products together underpin the company’s expanding deployments.
Growth expectations enter the priceBloomberg’s sources say the company expects more than $560 million in fiscal 2026 revenue, compared with roughly $474 million the prior year, and forecasts more than $1 billion in 2027. These figures include forward-looking targets rather than guaranteed growth. If financing closes at the proposed valuation, customer expansion and revenue performance will need to support a substantially higher entry price for investors.
▮ SIGNALCombining simulation with vehicle operating software increases potential spending per customer, while a higher valuation raises the growth the business needs to deliver.