// NODE-02 · NEURAL FEED · DAILY TRANSMISSION //
May 24, 2026 · WINDYVIEWS.COM
OpenAI is filing a confidential S-1 with the SEC as early as May 22, aiming for a September 2026 listing at a valuation of roughly $852 billion to $1 trillion. Anthropic, which has also engaged investment banks, is targeting an October 2026 listing at a $900 billion valuation. The back-to-back filings would make 2026 the largest IPO year in tech history.
Anthropic projects $10.9 billion in Q2 2026 revenue and expects its first quarterly operating profit ever. SpaceX's own IPO prospectus revealed that Anthropic is paying SpaceX $1.25 billion every month through May 2029 for GPU compute capacity. Anthropic also acquired SDK-tooling startup Stainless on May 18.
Google announced Gemini 3.5 Flash and Gemini Omni alongside a wave of product launches: Gemini Spark (a personal AI agent), Universal Cart for AI shopping, Ask YouTube, Gmail Live, Docs Live, Android XR glasses, and a new "Neural Expressive" design language. Google AI Ultra was cut from $250 to $100/month, and Gemini replaced daily prompt limits with a compute-based system that refreshes every five hours.
An internal OpenAI model autonomously disproved a geometry conjecture that had stood unresolved for 80 years. Fields medalist Tim Gowers called the result "a milestone in AI mathematics," and the work has reignited debate over how soon frontier models will graduate from solving textbook problems to producing original research.
Andrej Karpathy announced he is joining Anthropic to lead a rebuild of its pretraining research team. The move follows months of senior poaching across frontier labs and signals Anthropic's intent to push back at OpenAI on raw model quality — not only product polish and safety.
Chinese open-weight models have grown from roughly 1% of OpenRouter usage in 2024 to more than 60% by May 2026. Developers are routing to models from DeepSeek, Qwen, Moonshot and Z.ai for cost and openness, eroding the inference-share lead U.S. closed labs held just a year ago.
Apple is reportedly preparing an "Extensions" capability that will let users pick third-party providers — including Google and Anthropic — to power Apple Intelligence features across iOS 27, iPadOS 27 and macOS 27. The shift effectively concedes that Apple's in-house models cannot match frontier labs and turns the OS into a routing layer.
Meta is laying off roughly 8,000 employees — about 10% of its workforce — while reassigning another 7,000 to AI-focused teams. The cuts hit middle management and non-AI product groups hardest as Mark Zuckerberg consolidates headcount around superintelligence and Reality Labs AI efforts.
On May 7 the EU Council and Parliament agreed to streamline the AI Act, easing some compliance burdens on general-purpose model providers. In the U.S., the White House's March National Policy Framework continues to clash with state laws — Colorado's AI Act takes effect June 30, and the DOJ's new AI litigation task force is preparing preemption suits against state regimes it deems "innovation-limiting."
Capital is the story of May 2026: OpenAI and Anthropic are walking simultaneous trillion-dollar IPOs into a market where compute contracts (Anthropic-SpaceX) are now bigger than most listed software companies. Product velocity has shifted to agents and ambient assistants — Google I/O leaned hard into Spark and XR glasses while Apple quietly conceded the model race by opening iOS to third-party AI. Underneath, Chinese open models have eaten 60% of inference share, math-grade research from frontier models is no longer theoretical, and regulators on both sides of the Atlantic are racing to either simplify (EU) or centralize (U.S.) the rulebook before the next product cycle ships.