2026-05-10
By Vadym · Generated with AI, curated by me
NVIDIA revealed it has committed over $40 billion to AI equity deals in 2026 — on the same day JPMorgan confirmed its AI investments have already self-funded through $2 billion in savings. The infrastructure layer is solidifying fast, and the workforce is starting to push back.
NVIDIA has committed more than $40 billion to equity investments across the AI ecosystem in the first five months of 2026, with $30 billion going to OpenAI alone — the company’s largest-ever single investment. Additional stakes include up to $3.2 billion in glass manufacturer Corning and $2.1 billion in data center operator IREN, plus participation in roughly two dozen private startup rounds. Jensen Huang frames this as ecosystem deepening; critics note NVIDIA is investing directly in its own largest customers, raising questions about circular capital flows.
JPMorgan Chase has formally moved AI spending out of experimental R&D and into core infrastructure, placing it alongside cybersecurity and data centers within its $19.8 billion technology budget for 2026. The bank reports 500+ AI use cases in active production, a 10–11% productivity gain in engineering and operations, and a 95% reduction in anti-money laundering false positives. Jamie Dimon confirmed the program has already self-funded through $2 billion in operational savings across 150,000+ employees.
Anthropic doubled the five-hour usage limits for Claude Code across Pro, Max, Team, and Enterprise plans this week, drawing directly from its compute deal with SpaceX. Colossus 1 — SpaceX’s Memphis data center with 220,000 NVIDIA GPUs and 300 megawatts of capacity — now backs Anthropic’s production inference. Claude Opus API rate limits were also raised significantly, with the company noting constraints will ease further as the full compute allocation comes online.
Approximately 1,000 workers at Google DeepMind’s London office voted 98% in favor of seeking union recognition from the Communication Workers Union and Unite, over the company’s AI contracts with the U.S. Pentagon and Israeli military. Organizers gave Google management 10 working days to voluntarily recognize the unions before a formal legal process begins. Workers are demanding reinstatement of DeepMind’s 2018 pledge not to develop AI for weapons or surveillance — quietly removed from Google’s website in February 2025. Organizers have raised the possibility of “research strikes,” a potential first for a frontier AI lab.
Connecticut’s SB 5 — the AI Responsibility and Transparency Act — passed with rare bipartisan support: 131–17 in the House and 32–4 in the Senate, and awaits the Governor’s signature. The law requires large generative AI providers (1 million+ monthly users) to embed provenance data in all synthetic media, sets disclosure and anti-discrimination rules for automated employment decisions, and creates new safety rules for AI companion chatbots. Key provisions take effect October 1, 2026.
OpenAI released two new Realtime API models: GPT-Realtime-Translate, which translates live speech from 70+ input languages into 13 output languages while keeping pace with the speaker, and GPT-Realtime-Whisper, which transcribes speech in real time as it is spoken. Both run on the same low-latency infrastructure as GPT-Realtime-2, priced at $32 per million audio input tokens and $64 per million audio output tokens. The additions shift OpenAI’s voice infrastructure from conversation-only into real-time translation and transcription use cases.
Three things happened in parallel this week: NVIDIA bought control of the AI supply chain, JPMorgan proved the returns are real, and the people building the frontier models voted to go on strike over what those models are actually being used for. The capital layer and the workforce layer are now moving in opposite directions at speed. Connecticut drew a regulatory line. OpenAI shipped translation infrastructure most enterprise IT teams haven’t planned for. Anthropic solved a capacity problem by renting compute from a competitor’s parent company. None of this is noise — it’s the industry telling you where the pressure is building. Watch the workers, watch the regulators, watch who controls the GPUs.
— Boba
Curated by Vadym