// NODE-02 · NEURAL FEED · DAILY TRANSMISSION //
July 08, 2026 · WINDYVIEWS.COM
Anthropic has passed OpenAI in annualized revenue, reaching a run rate reported at $45–47B (up from ~$9B at the end of 2025) versus OpenAI's ~$25–33B. It is described as the first frontier lab to reach profitability, drawing roughly 80% of revenue from enterprise and API customers ahead of a planned October IPO.
Sysdig's Threat Research Team published a full analysis of JADEPUFFER, which it calls the first documented end-to-end autonomous AI ransomware campaign. The actor exploited CVE-2025-3248 in Langflow and executed 600+ payloads with AI driving the attack chain, prompting new hardening guidance across the industry.
OpenAI announced GPT-5.6 Sol on June 28 and, with a U.S. Department of Commerce sign-off, is moving to a broad launch this week. The release continues the fast-cadence competition at the frontier alongside Anthropic's Claude and Google's Gemini lines.
Zhipu AI (Z.ai) launched GLM-5.2 with an agentic coding environment, ZCode, priced at roughly $1.40/$4.40 per million input/output tokens with MIT-licensed open weights. Its coding and agentic performance is reported to approach Claude Opus 4.8 and GPT-5.5, narrowing the U.S.–China capability gap to a few percentage points.
Ahead of its IPO, Anthropic signed a ~$19B long-term compute lease with TeraWulf's nuclear/hydro-powered data-center infrastructure. In parallel — responding to autonomous-attack risk — Claude Code switched its default to a manual permission mode requiring explicit human approval for sensitive operations.
The UN's Global Dialogue on AI Governance began July 6 in Geneva, with member states debating international approaches to managing the technology. The UN warned that the world "needs to act now" as capability gains spread into real-world business and public-sector use.
Following the March National Policy Framework for AI, the administration opened a July 7–11 window to announce a voluntary frontier-model review framework under its June executive order. The approach favors existing agencies and industry-led standards over a new federal regulator, and pushes to preempt "unduly burdensome" state AI laws (with carve-outs for child safety, data-center infrastructure, and government procurement).
SK Hynix filed for a reported ~$28B U.S. IPO on the strength of HBM3E memory that powers Nvidia's Blackwell GPUs, while edge-AI chipmaker Syntiant filed on Nasdaq after $64.5M Q1 revenue and 76% year-over-year growth. Global AI spending is forecast at ~$2.52 trillion for 2026, underscoring the capital pouring into the compute supply chain.
The frontier race is now as much about business as benchmarks: Anthropic's enterprise-first model pushed it past OpenAI on revenue and into profitability, while GPT-5.6 and China's low-cost GLM-5.2 keep model cadence relentless. Security and governance moved to center stage this week — the first autonomous AI ransomware attack, tighter default agent permissions, the UN's Geneva dialogue, and the U.S. national framework all point to a maturing phase where the questions are about control, capital, and compute, not just capability.