Good morning. It's Monday, May 11th, 2026. Welcome to today's Voicecast.

**Story one.** OpenAI is going after the enterprise services market head-on. The company announced this morning a new subsidiary called the OpenAI Deployment Company, backed by more than four billion dollars in initial investment. It'll be staffed immediately through the acquisition of Tomoro, an AI consulting firm bringing about 150 forward-deployed engineers to the new unit. This is a direct answer to Anthropic's enterprise JV with Blackstone and Goldman Sachs, announced just last week. The API era is ending — the big labs are now building armies of deployment engineers to own the corporate stack, top to bottom.

**Story two.** The U.S. government's voluntary AI testing program just got a lot less voluntary in practice. Microsoft, Google, and xAI have agreed to give the U.S. government early access to their unreleased AI models for national security testing before public release. NIST now has pre-release testing agreements with all five major labs. And on Monday, an industry group recommended that passing safety review should be a requirement for winning government contracts. The trigger here is Anthropic's Mythos model, which can reportedly find and exploit zero-days across major operating systems. The government is trying to close the horse-gate after the horse showed up.

**Story three.** Nvidia has pushed its AI equity investment commitments past forty billion dollars in 2026 — and the pace is accelerating. The anchor is its thirty-billion-dollar bet on OpenAI. But this week alone, Nvidia struck two more massive deals: up to three-point-two billion in Corning, the optical and materials science company critical for AI data center infrastructure, and two-point-one billion in data center operator IREN. That's seven multi-billion-dollar investments in publicly traded AI companies in the first four months of the year. At this point it's less like venture capital and more like vertical integration — a pattern that's starting to draw regulatory eyebrows about circular deals.

**Story four.** Alibaba has officially launched agentic shopping on Taobao, integrating its Qwen AI platform directly into the marketplace. Instead of keyword searches, consumers can now browse, compare, and buy by chatting with an AI agent. The scale is notable: Qwen has already hit 300 million monthly active users across Alibaba's consumer apps, with 140 million first-time AI shopping experiences logged during the Chinese New Year campaign. If you've been wondering when conversational commerce would actually hit at scale — it just did, in the world's largest e-commerce market.

**Story five.** And finally, President Trump heads to Beijing on Tuesday for his first face-to-face meeting with Xi Jinping in over six months. The summit runs through Thursday. Trade and rare earth export controls are top of mind, but AI is now firmly on the agenda alongside Iran, Taiwan, and nuclear arms. When the world's two superpowers are negotiating AI policy at the same table as nuclear proliferation, that tells you everything about where we are.

That's all for today. Have a good week.