AI News Roundup — May 23, 2026
Microsoft is embedding Anthropic's Claude agents across Office and Copilot while simultaneously canceling internal Claude Code licenses over token-billing cost overruns. Nvidia quietly removed its gaming revenue category from financial reports as AI infrastructure dominates its business, and DeepSeek is pushing forward with a $10.29B financing round.
Microsoft Embeds Claude Agents in Office — Then Cancels Internal Claude Code Licenses
Microsoft announced a partnership to integrate Anthropic's Claude-based agents directly into Copilot for Excel, Word, and PowerPoint — signaling a shift toward more autonomous in-product AI workflows. In an apparent contradiction, Microsoft separately canceled its internal Anthropic Claude Code developer licenses, citing budget overruns as annual AI token costs balloon beyond fixed-budget planning. The dual moves highlight the tension between enterprise AI ambition and its real-world cost structure.
Anthropic Eyes Microsoft's Maia 200 Chip for Claude Inference
Anthropic is in early talks to use Microsoft's in-house Maia 200 AI accelerator chips for running Claude inference workloads — a move that would reduce Anthropic's dependence on Nvidia and reshape the AI chip competitive landscape. If finalized, it would be a major validation of Microsoft's custom silicon strategy.
Nvidia Removes Gaming Revenue Category From Financial Reports
Nvidia quietly eliminated its "gaming" revenue segment from its financial disclosures, folding it into broader categories as AI infrastructure becomes so dominant it no longer makes sense to highlight gaming separately. The move signals how completely Nvidia has transformed from a gaming GPU maker into an AI infrastructure company.
DeepSeek Pushes Forward With $10.29 Billion Financing Round
DeepSeek is moving ahead with a $10.29 billion financing round, with founder Liang Wenfeng committing to continue the company's aggressive open-source model development strategy. The raise would make DeepSeek one of the most heavily funded AI labs outside the US, reinforcing China's position in the frontier model race.
Google Embeds Gemini at the OS Level with Android 17 and Chrome Agents
Google unveiled Android 17 with Gemini AI at its core and previewed a redesigned Chrome with built-in "agent" functions. Google DeepMind is also piloting an "AI mouse" that turns on-screen hovering into contextual AI commands. The moves represent Google's push to make Gemini a system-level layer rather than a standalone app.
California Issues Executive Order on Trusted AI Procurement and Watermarking
California Governor Newsom signed Executive Order N-5-26 directing state agencies to develop "trusted AI" procurement rules and mandatory watermarking standards for AI-generated images and video. The order tightens compliance requirements for all AI vendors selling to the state — the world's fifth-largest economy.
Big Tech AI Spending Fueling Chip Shortages and Consumer Price Pressure
US lawmakers are warning that memory and compute capacity being allocated to AI infrastructure is constraining smartphone production and other consumer tech, feeding into broader price pressures that are complicating the GOP's midterm affordability messaging. The AI chip crunch is no longer just a tech story — it's a macroeconomic and political one.