AI (artificial intelligence): analysis of 16 November 2026

Back to the Chronology · Current news in this section

Photo from Is bad news for Jensen Huang also bad news for the AI industry?, TechCrunch AI

Analysis

Nvidia's leadership crisis deepens as Jensen Huang's resignation collides with a China chip-diversion scandal, a petition for his reinstatement and rumours of a comeback, while Anthropic heads toward a record IPO amid frontier-model and compute uncertainty.

The AI sector's dominant hardware supplier is in turmoil: Jensen Huang has resigned as Nvidia CEO after reports of multiple affairs, including one with an executive at a firm suspected of funnelling chips to China, Colette Kress is interim CEO, and the board says it has not formally engaged successor candidates even as a 115,000-signature petition and reported comeback manoeuvres add pressure 414192122. At the same time the frontier labs are pushing hard: OpenAI's GPT-6 Astra is credited with helping produce a lattice definition of the Standard Model and with making a standalone robotics team redundant, while Google's Gemini 4 Carbon and Anthropic's Fable 5.2 have displaced OpenAI from the top of the MathArena benchmark 12935. Anthropic's forthcoming IPO, described as potentially the largest in history, is shadowed by Dario Amodei's admission of 'huge uncertainty' over compute commitments of nearly $901 billion and by a departing researcher's warning that algorithmic gains are accelerating 2536. In the background, China's Meituan has released a 1.6-trillion-parameter agent model trained entirely without Nvidia GPUs, and Australia's central bank reports AI is already hitting young entry-level workers 1618.

Key developments

  • Nvidia succession in flux. Jensen Huang resigned as Nvidia CEO after 33 years following reports of multiple affairs, including one with a suspected Chinese spy; CFO Colette Kress is interim CEO and the board says it has not formally engaged any successor, though it reportedly approached Arm's Rene Haas [21][32][37]. A petition signed by Elon Musk and other executives calling for Huang's reinstatement passed 115,000 signatures, and The Verge reports unconfirmed rumours that Huang is seeking a path back, which would complicate any orderly transition [4][12][22]. Commentators note that 'so much of this economy is riding on Nvidia', so the instability has implications well beyond one company [3]. 34122122
  • China chip-diversion allegations. The WSJ reports that Megaspeed, owned by Huang's former lover Alice Huang, brokered nearly $12 billion of Nvidia chip orders for Chinese companies in 2024-25, and short seller Culper Research estimates Nvidia's true China revenue share exceeds 20 percent, above official figures; the short-seller estimate is an interested-party claim and remains unconfirmed [14]. Separately, Nvidia's security chief quietly left in August, and a former intelligence officer argues this kind of leak runs through everyday conversations that corporate security rarely monitors [19]. The combination of export-control exposure and governance failure is likely to attract regulatory as well as investor scrutiny. 1419
  • Anthropic's record IPO under a cloud of uncertainty. Days before what could be the largest IPO in history, Dario Amodei called buying compute 'the deepest business dilemma' facing Anthropic, whose commitments total nearly $901 billion with no clarity on whether revenue will justify them; the IPO is largely intended to fund those costs [25]. Former researcher Jacob Coxon says a roadshow slide showing Anthropic can retrain Mythos with 'a fraction of the compute originally needed' reveals accelerating algorithmic progress and is renewing his call for a pause on frontier development [36]. Those two accounts cut in opposite directions for investors: cheaper training could ease the compute dilemma, or it could intensify a race that keeps spending hi 2536
  • Frontier model leapfrogging. Google's Gemini 4 Carbon and Anthropic's Fable 5.2 have knocked OpenAI's GPT-6.1 Sol off the top of MathArena, though Gemini costs 15 times more and ignored time limits in testing [29]. Anthropic has made Claude Opus 5.5 the default on paid plans and told users to drop 'think hard' prompting in favour of a Medium effort setting, while Fable 5.2 generates images by writing code that simulates brushstrokes rather than computing pixels [10][31]. OpenAI, meanwhile, is leaning on GPT-6 Astra's scientific and robotic generality, launching a science blog and dissolving its standalone robotics team because general-purpose models can reportedly control robots directly [1][2][35]. 12102931
  • China builds without Nvidia. Meituan released the weights of LongCat-2.5, a 1.6-trillion-parameter agent model trained entirely on Chinese hardware and deployed on 194 Huawei Ascend chips, scoring 88.4 on Terminal Bench, just behind Claude Opus 5 [18]. TechCrunch Mobility separately describes a 'Nvidia backup plan' for China's self-driving industry, and analysts argue Huang's exit opens the door for AMD and Google [7][37]. If chip-diversion allegations trigger tighter enforcement, demonstrations of competitive non-Nvidia training will matter more. 71837
  • Agents move into payments and enterprise operations. OpenAI introduced a ChatGPT Wallet for agent-led payments requiring biometric approval of every transaction, and Google connected Wallet to Gemini in the US for spending and budget queries with optional Plaid bank data [33][42]. AT&T says nearly 500 agentic workflows are in production and plans to triple that next year, while NiCE's Cognigy pitch is that running agents, not building them, is the hard part [24][45]. The pattern is a shift from assistants to autonomous actors handling money and business processes, raising the stakes for the prompt-injection vulnerabilities being discussed on Hacker News [11][23]. 1123243342

Themes

  • Concentration risk around Nvidia. Several items frame the Huang saga not as corporate gossip but as systemic exposure, from TechCrunch asking whether bad news for Huang is bad news for the industry to speculation about rivals' openings and China's hardware alternatives [3][18][37]. The governance and security lapses now surfacing suggest the risk was not only commercial dependence but also weak controls at the chokepoint of the AI supply chain [14][19]. 314181937
  • AI's labour effects arrive at the entry level. Australia's central bank finds job postings and employment for 15- to 24-year-olds in software development and customer service have fallen more sharply than for older workers since ChatGPT launched, while AT&T is tripling agentic workflows and contact centres remain the primary target for automation [16][24][45]. Even open-source software feels the strain, with Linux maintainers quitting under a flood of AI-generated bug reports that are easy to produce but hard to fix [44]. 16244445
  • Capability acceleration versus cost. OpenAI credits GPT-6 Astra with helping produce a detailed proof toward a lattice Standard Model and with absorbing robotics work, while Coxon says Anthropic's algorithms are improving at an accelerating rate [1][35][36]. Yet the benchmark leader costs 15 times its OpenAI rival and Anthropic's compute commitments remain unproven against revenue, so the economics are not keeping pace with the capabilities [25][29]. 125293536

What to watch

  • Whether Nvidia's board names a permanent CEO or Huang engineers a return, and how the WSJ chip-diversion reporting and Culper's unverified China revenue estimate are addressed by the company or regulators [4][14][21][22]. 4142122
  • Anthropic's IPO pricing and reception given Amodei's stated uncertainty over nearly $901 billion in commitments and Coxon's claims about compute efficiency gains [25][36]. 2536
  • Independent evaluation of Meituan's LongCat-2.5 on Huawei Ascend hardware, which would test how real China's path away from Nvidia has become [18][37]. 1837
  • Apple's reported Siri smart glasses at the next WWDC, with Bloomberg's Gurman saying they could ship by year-end, after Siri AI's repeated delays landed it on TechCrunch's AI graveyard list [9][27]. 927
Stories this analysis is based on (28)
  1. A step toward defining the Standard Model on a lattice OpenAI News
  2. Our new blog for science OpenAI News
  3. Is bad news for Jensen Huang also bad news for the AI industry? TechCrunch AI
  4. Jensen Huang might be plotting a return and more of the latest Nvidia drama The Verge
  5. TechCrunch Mobility: The Nvidia backup plan for China’s self-driving industry TechCrunch
  6. Apple’s Siri AI smart glasses are reportedly launching next year, here’s what to expect 9to5Mac
  7. Anthropic's Claude Fable 5.2 doesn't generate images, it paints them The Decoder
  8. Prompt Injection Attacks in Large Language Model Systems Hacker News
  9. An overview of Jensen Huang’s messy days The Verge
  10. Firm linked to Jensen Huang's ex-lover reportedly brokered nearly $12 billion in Nvidia chip orders for China The Decoder
  11. Australia's central bank finds AI hitting young workers first, especially in software and customer service The Decoder
  12. Meituan releases a 1.6-trillion-parameter agent model that never touched an Nvidia GPU The Decoder
  13. Where was Nvidia’s security team? TechCrunch AI
  14. Who should run Nvidia? Our picks for Jensen Huang’s successor, whether the board likes it or not GeekWire
  15. Elon Musk, other tech execs sign petition calling on Nvidia to reinstate Jensen Huang TechCrunch
  16. Show HN: Deterministic prompt injection detector using formal logic Hacker News
  17. AT&T uses agentic AI to reinvent business processes from the ground up SiliconANGLE AI
  18. Anthropic CEO Amodei admits "there's a huge uncertainty in what's going to happen next year" days before a record IPO The Decoder
  19. The AI graveyard: a running list of projects and startups that didn’t make it TechCrunch
  20. Gemini 4 Carbon takes the MathArena crown but costs 15 times more than GPT-6.1 Sol The Decoder
  21. Anthropic says outdated prompting habits are slowing down Claude Opus 5.5 The Decoder
  22. Google adds Wallet to Gemini for finance insights TestingCatalog
  23. OpenAI reassigns robotics researchers to other teams as general-purpose models might make specialized robotics AI unnecessary The Decoder
  24. Anthropic can already retrain Mythos with "a fraction of the compute originally needed," says former researcher Coxon The Decoder
  25. Jensen Huang’s Exit Just Gave AMD and Google Their Biggest AI Opening Analytics India Magazine
  26. OpenAI Introduces ChatGPT Wallet for Agent-Led Payments Analytics India Magazine
  27. ‘We need more reviewers’: Finding Linux bugs is easy now – fixing them is another story ZDNET AI
  28. NiCE’s next-generation Cognigy bets the hard part of AI agents is running them, not building them SiliconANGLE AI

Written by Claude Fable 5.1 (Anthropic), based on 45 recent headlines and excerpts in this section, 4d ago. AI analysis can misread or miss things: the numbers link to the original reporting, which is the authority.