AI news, 18 October – 2 November: Anthropic’s IPO filing reveals $17B quarter and nearly $900B compute bets
Anthropic files for IPO and updates with blockbuster figures. On 26 October Anthropic made public its S-1 registration statement, seeking to list on Nasdaq under the ticker ANT and positioning itself for one of the largest technology IPOs ever. The initial filing showed first-half 2026 revenue of $16.3 billion. On 2 November an amended filing added third-quarter results of roughly $17 billion in revenue (up more than 1,100% year-on-year) and disclosed about $895 billion in multi-year purchase and lease commitments, nearly all for computing infrastructure, according to CNBC, Reuters and the company’s SEC documents. The numbers underscore both explosive growth and the enormous capital intensity of frontier AI, while the prospectus also flags existential and safety risks.
Google launches Gemini 4 Carbon and Genie 4. On 28 October Google announced Gemini 4 Carbon, describing it as its strongest model yet for cybersecurity and long-horizon engineering tasks; it topped the Artificial Analysis Intelligence Index and is initially available only to a limited group of trusted cyber defenders through the Fairwind Program. A day earlier, on 27 October, Google DeepMind unveiled Genie 4, a real-time interactive world model that generates native audio alongside photorealistic, steerable scenes lasting up to two minutes. Both releases highlight rapid iteration and a cautious, staged approach to powerful new capabilities.
OpenAI ships cheaper computer-use model and trims funding target. On 26 October OpenAI released GPT-6.1 Terra, a mid-tier model focused on visual reasoning and computer use, priced at roughly half previous rates ($1/$5 per million tokens) and available in ChatGPT Work, Codex and the API. Later, on 30 October, the Financial Times and others reported that OpenAI had cut the target for its latest private funding round by $10 billion to about $20 billion at a $1.25 trillion valuation amid a broader tech sell-off and competition from Anthropic’s IPO path; early commitments came in lower still.
Mistral opens weights for trillion-parameter Large 4. On 29 October French lab Mistral released the weights of an improved Mistral Large 4 (“Le Chonk”), a one-trillion-parameter multimodal mixture-of-experts model (about 52 billion active) under a modified open licence. The company said further training lifted coding-agent performance, making it a leading open option for developers seeking strong agentic and cybersecurity capabilities without full proprietary lock-in.
Banks restrict AI agents; New York tightens chatbot rules for minors. Around 22 October major U.S. banks including JPMorgan Chase, Bank of America, Citi and others began blocking or heavily restricting third-party AI agents from logging into customer accounts or executing transactions, citing security, fraud and liability risks after detecting unauthorised activity, according to Reuters and Axios. On 28 October New York Governor Kathy Hochul signed legislation banning unsafe AI companion features for users under 18—such as pretending to be human, encouraging self-harm or emotional dependency—requiring age verification and giving the attorney general enforcement powers.