AI Weekly Issue #490: Anthropic just had AI's biggest week of 2026

In five days, Anthropic shattered records and reshaped AI infrastructure. CEO Dario Amodei reported Q1 revenue grew 80-fold year-over-year, reaching an annualized run rate above $44B — the steepest single-quarter revenue jump any frontier AI company has publicly disclosed. To sustain that growth, Anthropic committed $200B to Google Cloud and signed a compute deal with SpaceX's Terafab chip project (a planned $55B Texas fab that may reach $119B total). On the product side, the company shipped Claude Code Auto Mode, which automatically picks the optimal Claude model and tools for each coding task, targeting ~90% of tasks without human intervention. It also opened its Agent SDK to all external developers and launched ten specialized financial-services agents alongside JPMorgan CEO Jamie Dimon.

Elsewhere in the AI world, the European Union struck a last-minute deal on AI Act compliance, pushing the deadline for high-risk AI systems from August 2026 to December 2027 — saving Q3 roadmaps. In the US, Pennsylvania became the first state to sue Character.AI under consumer-protection law, alleging a chatbot impersonated a licensed psychiatrist to a minor. Google DeepMind UK staff voted 98% to unionize, triggered by a classified Pentagon contract — the first such union at a top AI lab. DeepSeek is in talks for its first fundraising round at a $45B valuation via China's state chip fund. And a federal judge ruled that a jury will decide whether Microsoft secretly controlled OpenAI, a key question in Elon Musk's ongoing trial.