Today's digest

AI news today: Wednesday, November 11, 2026

Anthropic begins its IPO roadshow; TSMC backs Anthropic and OpenAI; Google releases Gemini 4 Carbon

Full issue →

Get tomorrow's issue in your inbox.

  1. 01

    TSMC approves up to US$1 billion in investments across Anthropic and OpenAI

    Source: Taiwan Semiconductor Manufacturing Company (TSMC) / U.S. SEC filing

    In a November 10 Form 6-K, TSMC disclosed that its board approved investments of up to US$500 million in Anthropic common shares at Anthropic’s IPO price and up to US$500 million in OpenAI preferred stock. The same resolutions approved approximately US$40.9 billion in capital appropriations, mainly for advanced technology and packaging capacity, fab construction and 2027 research and development. These are board approvals; the filing does not say the investments have closed.

    Why it mattersThe world’s largest contract chipmaker is positioning itself as both a key supplier and a potential investor in the AI build-out, while continuing to commit heavily to manufacturing capacity. The approvals do not guarantee completion or promise either AI company preferential access to TSMC capacity. More on OpenAI →

  2. 02

    Anthropic launches the roadshow for its proposed public offering

    Source: Anthropic

    Anthropic said on November 10 that it had launched the roadshow for a proposed IPO of its Class A common stock. The proposed offering covers 825 million shares, including 801,351,354 new shares offered by Anthropic and 23,648,646 shares offered by existing stockholders. The expected price range is $70–$78 per share, and the company has applied to list on Nasdaq under the symbol ANT. Anthropic said its registration statement has been filed but is not yet effective, so the shares cannot be sold until it becomes effective. These are proposed terms, not a completed listing.

    Why it mattersThe proposed flotation could become a major public-market test of investor appetite for frontier-AI businesses, their compute commitments and their path to sustainable revenue. The final price, size and completion remain subject to the registration process and market conditions. More on Anthropic →

  3. 03

    Google makes Gemini 4 Carbon available for complex agentic and research work

    Source: Google

    Google announced on November 10 that Gemini 4 Carbon was becoming available across its products. Developers and enterprises can use it through the paid Gemini API in AI Studio and Gemini for business; access in the Gemini app, Antigravity and AI Mode in Search is rolling out to Google AI Ultra subscribers. Google positions Carbon for long-horizon agent workflows, complex coding and scientific reasoning, and is previewing a coordinated multi-agent "swarm" feature in Antigravity. The introductory API price is $2 per million input tokens and $10 per million output tokens; Google says those prices will rise to $4 and $20 respectively after November 30. Performance and safety comparisons in the announcement are Google's own claims.

    Why it mattersThe release expands the options for teams building agentic workflows, but access is tiered and the model may consume more tokens on difficult tasks. Buyers should compare end-to-end task cost, reliability and review effort rather than token rates or vendor benchmarks alone. More on Google & DeepMind →

  4. 04

    Google DeepMind says AI agents formally verified key Signal protocol code properties

    Source: Google DeepMind

    In a November 10 research post, Google DeepMind described AlphaProof Code, a multi-model agentic framework used with human experts to formally verify key security properties of the compiled code implementing the Signal protocol. DeepMind says agents powered by Gemini Carbon generated more than six million lines of Lean proofs, each machine-checked, while human reviewers focused on the specifications and the small portion flagged for further review. The work took four weeks, according to the researchers. DeepMind also says the system detected two previously fixed bugs when they were reintroduced into an isolated copy of the code. These are the researchers' reported results; the proof covers the properties specified and checked in this project, not every aspect of Signal or all software security.

    Why it mattersFormal verification can provide stronger assurance than testing alone when software must satisfy precise security properties. Agentic tools may make this work more scalable, but trustworthy specifications, independent review and careful definition of what has actually been proved remain essential. More on Google & DeepMind →

  5. 05

    IBM and Microsoft launch an enterprise AI-agent platform for regulated industries

    Source: IBM

    IBM and Microsoft announced Imagine Enterprise on November 10, a platform intended to help organizations build, operate and govern AI agents for industry-specific processes. IBM says it contributes more than 6,000 pre-built agents and assets, while Microsoft supplies AI, data, cloud and security technologies; both companies also offer forward-deployed engineering support for qualifying engagements. The initial industry frameworks cover financial services, insurance and retail. The companies describe the platform's capabilities and intended benefits; these are vendor claims, not independently measured outcomes.

    Why it mattersThe announcement shows enterprise AI moving from standalone chat tools toward governed agents connected to company systems and workflows. Buyers should assess integration effort, security controls, observability and measurable results before treating a large catalogue of agents as ready-to-use productivity. More on Microsoft →

New here? How to read this digest
Each story links to its original source. "Why it matters" explains the impact in plain language, with jargon explained. Tap a topic to see every story on it, or browse the archive.

More recent digests

Free daily digest

The AI news that matters, in five minutes.

5–10 hand-picked stories a day, each with a plain-language line on why it matters. Every item links to its source.

Free. One email a day, no spam. We only use your address to send AI Week Daily and never sell it. Unsubscribe any time with the link in every email, or reply to any email with “unsubscribe”. See the privacy note.

Free · Unsubscribe any time · Privacy

♥ Donate