
Jensen Huang's Spy Scandal Dominates as Anthropic Drops New Models
Nvidia's CEO is in freefall over an alleged affair with a Chinese spy, while Anthropic ships new models and drops $3.4B on gas turbines.
Subscribe free →
Nvidia's Jensen Huang Is in Full Crisis Mode — and His Job May Be on the Line
Jensen Huang has admitted to being "unfaithful" to his wife with Alice Huang, an executive at Megaspeed (an AI compute firm renting Nvidia GPUs) who U.S. intelligence has identified as a suspected Chinese spy. Nvidia's board has opened a formal investigation, the stock dropped over 4%, and Yale governance expert Jeffrey Sonnenfeld puts the odds of Huang keeping his job at roughly 50-50. Meanwhile, Huang has quietly withdrawn from both an Anthropic event and the ElevenLabs Summit.
Why it matters: The CEO of the company that supplies the chips powering virtually all advanced AI development is now under a corporate investigation with national security implications — this affects everything from Nvidia's China market access to U.S. chip export policy.

A Short Seller Saw This Coming Six Months Ago
Culper Research flagged Jensen Huang's unusually close relationship with Alice Huang back in early 2026, well before the scandal broke publicly. Nvidia's response at the time was to call the relationship "normal" — a characterization that securities lawyers now say could become very expensive if it's ruled a material misstatement to investors.
Why it matters: The paper trail of warnings Nvidia ignored hands plaintiff attorneys exactly the ammunition they need for securities fraud suits, and multiple law firms are already racing to file first.

Anthropic Ships Claude Fable 5.2 and Mythos 5.2
Anthropic released two new Claude models today — Fable 5.2, optimized for tough software engineering problems, and Mythos 5.2, aimed at broader technical reasoning and vision tasks. Both are faster than their 5.1 predecessors and claim improvements across coding, reasoning, and multimodal understanding.
Why it matters: Anthropic is pushing model releases at an aggressive cadence while competitors do the same, keeping pressure on OpenAI and Google even as the company deals with internal RSI concerns among its workforce.

Anthropic Drops $3.4 Billion on Natural Gas Turbines
Anthropic has committed $3.4 billion to purchase natural gas turbines, joining a growing list of AI companies locking up fossil fuel power infrastructure to feed their data center appetites. The move is a bet that the company will need massive dedicated power capacity — but also carries real risk if energy strategy shifts or cleaner alternatives arrive faster than expected.
Why it matters: AI labs are now making multi-billion-dollar infrastructure bets that tie them to fossil fuels for years, putting them on a collision course with their own stated climate commitments.

Meta's Muse Ember Is Gunning for the Cheap Agent Market
Meta released Muse Ember, a lightweight model built specifically for agentic task execution, priced at just $0.10 per million input tokens. That's a fraction of what competitors like Anthropic's Haiku or Luna charge, and positions Meta as the low-cost option for developers building agent pipelines that need to run thousands of cheap, fast calls.
Why it matters: Pricing at this level commoditizes agent execution and could force rivals to cut prices or differentiate on capability — the race to the bottom in inference pricing is accelerating.

Apple Starts Charging $50 for More AI Features
Apple has quietly introduced $49.99 "Intelligence Usage Packs" — one-time purchases that extend access to Apple Intelligence features like Logic Pro's AI assistant beyond their default usage caps. First spotted in Logic Pro, this marks Apple's first move toward monetizing AI features beyond the base device cost.
Why it matters: Apple is testing whether users will pay à la carte for AI on top of hardware they already bought, which could become a significant revenue stream — or a significant backlash waiting to happen.
Quick Hits
- →Waymo robotaxis are still illegally passing school buses despite a software fix rolled out nearly a year ago, federal regulators say — TechCrunch
- →AI hardware startup Nominal tripled its valuation to $3B in just 8 months, with defense giant Anduril as a key customer — TechCrunch
- →Automattic launched Spacefast, a vibe-coding deployment platform taking direct aim at Lovable — TechCrunch
- →Meta is rolling out AI age-detection globally on Instagram to catch teens lying about their birthdays — TechCrunch
- →COP31's new AI-for-climate pledge is getting roasted as performative while AI data centers keep guzzling fossil fuels — The Verge
AI TLDR