Daytona is now a $1B company for one reason: AI agents need their own computers.
Sandboxes in 60 ms. Revenue up ~25x since January.
And its CEO says a CPU crunch is coming.
The UK just launched the world's first dedicated AI biosecurity lab.
£100M from government. Up to £200M more from OpenAI and Anthropic.
The job: spot biological threats earlier and speed up vaccines.
Jensen Huang put four $399 robot ducks on stage at GTC Berlin.
Now Hugging Face says Microduck preorders are crossing 20,000, with deliveries targeted before Christmas.
Google can now use your Gmail and Google Photos data to target ads.
The catch: it only applies if you've connected them to Gemini's Personal Intelligence.
Anthropic's own public benefit report admits four Claude models reached real third-party systems during cyber testing.
A misconfigured test environment was connected to the open internet.
NVIDIA is prepaying suppliers to lock up AI chip capacity through 2028.
2.5x more substrates, 300% more interposer capacity at Amkor, and a $1.1B US test plant.
Samsung Electro-Mechanics is the new name on the list.
Anthropic says Claude didn't just fuzz 123 open-source libraries. It proved 239 of their fuzz targets can never crash.
Machine-checked proofs, published openly.
Rivian just bought Third Dimension AI, a 15-person UK startup that rebuilds real driving scenes as 3D worlds, to train the self-driving system behind its planned Uber robotaxis.
https://t.co/cLcCkjbJ7z
Claude on Google Cloud now comes two ways: run by Google, or run by Anthropic with Managed Agents, skills and code execution on day one. Same models, same price, different data processor.
https://t.co/Yvqwwojb7Y