← Dailies

Daily · 2026-11-12 00:00 UTC

TSMC buys $1B in Anthropic and OpenAI; Gemini Decision 1 API

TSMC takes $1B in Anthropic and OpenAI stakes, Google ships a decisions-only model, AlphaProof verifies Signal, and AlphaGenome 2 reads DNA by cell type.

Source ↗

Ten stories from the last 24 hours, ranked by signal rather than volume.

1. TSMC approves up to $1B in Anthropic and OpenAI stakes — The board authorised up to $500 million each in Anthropic common stock and OpenAI preferred stock, and separately approved up to $500 million to set up a semiconductor investment company. Quartz · Nov 11
2. Google ships Gemini Decision 1 and the Gemini Decisions API — Up to 1,000 yes/no, choice or rubric-scoring questions in a single forward pass across text, images, audio, video and PDFs, at 150–250 ms end-to-end with no output tokens billed. Google says it is nearly 5x cheaper than Gemini 3.5 Flash-Lite on its shared benchmark suite. blog.google · Nov 11
3. DeepMind's AlphaProof Code verified Signal's cryptographic core in four weeks — Agents powered by Gemini Carbon wrote more than six million lines of proof code and flagged about one percent for human review, a verification job that normally takes years. DeepMind says it wants to extend the method to power grids and medical devices. The Decoder · Nov 11
4. AlphaGenome 2 predicts how mutations behave in specific cell types — Trained on more than 100 million individual cell profiles across 68 tissues, the model identifies nearly twice as many variants that change gene expression as the previous state of the art, and can make predictions about cell types it was not retrained on. The Verge · Nov 11
5. Fractile raises $720M for inference chips, before shipping silicon — The British startup's Series C was backed by the UK's Sovereign AI Fund and led by Thrive Capital, Founders Fund, Lightspeed and Redpoint. The British government put the valuation at over £5 billion; the Financial Times and a Reuters source reported roughly $7 billion. TechCrunch · Nov 11
6. OpenAI's quarterly revenue reportedly crosses $10B — That is almost a 50% jump from the $6.7 billion recorded in the second quarter, driven largely by business customers. TechCrunch · Nov 11
7. Anthropic says AI cut an AMD chip migration from months to three days — The company's in-house model, Mythos, reportedly ported Claude to AMD GPUs with almost no human intervention. Claude now runs on four chip families. The Decoder · Nov 11
8. Google Labs launches Tabletop, a shared space that generates group tools — Drop in links and files, then prompt it to build trip itineraries, dinner polls, expense splitters or tournament brackets, and share one invite link. Rolling out in the US, 18 and over, by waitlist. blog.google · Nov 11
9. England's universities told to return to exams and in-person presentations — The Office for Students says AI has made coursework "close to impossible" to assess, and warns that weak assessment design could breach registration conditions and bring sanctions. The Guardian · Nov 11
10. Police watchdog: facial recognition errors risk wrongful arrests — His Majesty's Inspectorate of Constabulary found some investigators treated retrospective facial recognition results as evidence rather than intelligence, acting without enough corroborating enquiries. The report also found the algorithm is still more likely to wrongly flag people from ethnic minorities. The Guardian · Nov 12

Try This Next

• 🔧 Try the Gemini Decisions API in Google AI Studio: send one document and a batch of yes/no checks, and watch a classifier collapse into a single call.
• 🎥 Read the writeup on AlphaProof Code and Signal — the clearest public look at AI doing formal verification of real-world code.
• 📝 Save a note on Fractile's stacked-DRAM approach before you price any inference hardware this year.

Content on Anagnorisis is summarized, paraphrased, and editorialized from publicly available sources for length and clarity. Original sources are linked where available. All trademarks belong to their respective owners.