Daily Highlights
- Anthropic files S-1 for Nasdaq IPO under ticker ANT
- Anthropic posts $16.29B H1 revenue against $80.7B net loss
- Mistral Large 4's 1-trillion-parameter open weights land today
- Alibaba Cloud, MetaX and Beijing AI fund take stakes in Kling AI
- Oracle launches Claw-powered agentic apps for supply chain
Today's Headlines
[Anthropic formally files S-1, set to become first frontier AI lab to list]On October 26, Anthropic filed its Form S-1 with the U.S. SEC to list Class A shares on the Nasdaq Global Select Market under ticker ANT, with Goldman Sachs and Morgan Stanley leading a 22-bank syndicate. The prospectus shows $16.29 billion of revenue in H1 2026 (up from $1.16 billion a year earlier) against an $80.7 billion net loss, of which roughly $72.4 billion was non-cash fair-value changes on convertible notes and SAFEs; Q2 delivered its first operating profit of $974 million, and the filing discloses more than $600 billion in infrastructure commitments over the next decade. (Source: Reuters)
[Mistral Large 4 open weights arrive today: Europe's first trillion-parameter open model]According to Reuters and other outlets, France's Mistral AI released the weights of Mistral Large 4 on October 27: 1 trillion total parameters with about 49 billion active per token, trained from scratch on 3,800 NVIDIA Grace Blackwell GPUs in Mistral's own European datacenters. The model scores 93% on Cybench and 93.3% resistance on prompt-injection benchmarks; with weights public the community can now quantize and self-host it, and Mistral plans a family of specialized models built on top. (Source: Reuters)
New Products & Tech
[Oracle launches Fusion Claw agentic applications for supply chain]On October 26 at Oracle AI World in Las Vegas, Oracle introduced Claw-powered Fusion agentic applications for supply chain operations. Fusion Claw is a governed agent execution runtime supporting reasoning, computation, simulation, modeling and continuous dynamic replanning, letting enterprises autonomously handle tasks from supply and transportation plan optimization to procurement spend analysis within authorized scopes. (Source: Oracle)
[Kimi apologizes for K3.1 Pro throttling, compensates subscribers with 25% monthly quota]On October 26, Moonshot AI publicly apologized after paying subscribers hit rate limits on Kimi K3.1 Pro and offered affected subscribers a 25% monthly quota compensation. K3.1 Pro had earlier topped open-weight coding benchmarks, driving demand that strained its compute. (Source: Caiwen News)
IPO & Fundings
[Alibaba Cloud, MetaX and Beijing AI industry fund invest in Kling AI]On October 26, it was reported that Kuaishou's Kling AI has taken investments from Alibaba Cloud, chip startup MetaX and the Beijing AI industry investment fund, among others. In August the state-backed AI fund had invested RMB 1.4 billion for about 1.14% of its domestic entity, implying a valuation above RMB 120 billion. (Source: Caiwen News)
Open Source & Research
[Mistral Large 4: a 1T-total/49B-active MoE open-weight model]Mistral Large 4 uses a mixture-of-experts architecture with 1 trillion total parameters and about 49 billion active per token (roughly 52 billion including embeddings and output layers). Mistral reports 93% on Cybench, 82% on a reproduce-and-patch vulnerability test, and 61.7% on DeepSWE v1.1; the released weights enable self-hosting, with license terms to check on the release page (its predecessor Large 3 was Apache 2.0). (Source: Mistral)
Company & People
[Anthropic's S-1 reveals customer concentration: five customers drive about a quarter of revenue]Anthropic's filing shows its five largest customers accounted for about 25% of revenue in the first nine months of 2026, with international revenue of $4.5 billion making up 27.8% of the total; the prospectus also discloses over $600 billion in infrastructure commitments over the next decade, including $450 billion in non-cancelable cloud computing contracts. (Source: CNBC)
Policies & Data
[Colorado AI rulemaking public comment period closes October 26]Colorado's formal rulemaking under its AI law closed its public comment period on October 26, with final rules expected to take effect in January 2027; in the same regulatory wave, California's SB 1000 on AI transparency has become law and the EU's ban on non-consensual synthetic imagery takes effect December 2. (Source: Vorp Labs)
Disclaimer: For reference only; exercise caution for investment or decisions.
Powered by AOYII.com
