DeepSeek Watch
France’s government is using DeepSeek for coding through a shared AI gateway, according to Takahiro Anno’s account of interviews with French officials and researchers. The choice puts open-weight models and control over infrastructure ahead of access to the very latest closed models for some everyday government work.
AWS AI Watch
Amazon is investing about US$50 million in Gold Circuit Electronics, a Taiwanese maker of circuit boards for AI servers and network switches. The deal gives Amazon a small equity stake in a supplier that says demand for high-end AI boards already exceeds its capacity, making this a supply-chain…
Watch Desk
Cloudflare says automated, non-human traffic passed half of all internet traffic in May 2026, driven heavily by AI agents and crawlers. The company is proposing controls for publishers facing the costs and commercial questions that follow when more website visitors are machines, not customers.
Mistral AI Watch
The European Space Agency and Mistral AI have signed a letter of intent to explore AI work spanning Earth observation, mission design and space engineering. The agreement, signed in Paris on 16 September, puts Mistral’s models and infrastructure in the frame for European space applications, but does not…
Microsoft AI Watch
Whitbread is extending Microsoft Copilot to 1,650 support-centre colleagues and multi-site hotel managers after a pilot in which staff reported saving an average of 24 minutes a day. The hotel group is also using a Copilot-built agent to assess potential locations for new hotels, with people still responsible…
ElevenLabs Watch
ElevenLabs says a $300 million employee share sale has valued the voice-AI company at $22 billion, while its enterprise agent business has grown sharply. The announcement puts a price on the company’s momentum, but its deployment figures offer the more useful clue to what is driving it.
TSMC Watch
Samsung’s foundry business is filling more of its existing capacity and raising prices on some new orders, just as AI demand is straining TSMC’s advanced chipmaking lines. The opportunity is real, but winning repeat customers, not merely a first order, will show whether Samsung can turn the squeeze…
Microsoft AI Watch
Microsoft has made WSL Containers generally available, giving developers a way to build, run and deploy Linux containers directly on Windows. The release pairs that convenience with separate storage and networking for container sessions, a useful distinction from simply running Linux tools side by side.
Watch Desk
AI companies are betting that agents will make everyday tasks easier, but many people are wary of giving them the access that would make them useful. Axios reports that survey findings point to a sizeable gap between industry ambition and public willingness to hand over email, travel bookings…
Watch Desk
America is adding record amounts of clean power, but the renewable energy industry’s leading trade group warns that growth could fall sharply from 2028 if federal restrictions continue. That matters to the AI build-out too: rising electricity demand, driven largely by AI, is making the fight over new…
Watch Desk
Carnegie Mellon University says a $3 billion gift from Ken Griffin will fund a new Miami campus and put $500 million into its School of Computer Science. The scale is extraordinary; the consequential question is what the money will build for computing and AI research, and who will…
Watch Desk
Carnegie Mellon University says its planned Miami campus will organise research and teaching around societal challenges, with AI and other emerging technologies part of the approach. Its four initial areas include health, national security, advanced manufacturing, and energy and climate resilience.
Micron Technology Watch
Micron’s revenue nearly quintupled to $54.2 billion in the quarter ending 3 September, as demand for high-bandwidth memory used in AI chips surged. Bloomberg says the company beat fourth-quarter guidance and its first-quarter outlook exceeded estimates, giving investors fresh evidence that AI infrastructure demand is reaching the memory…
Cerebras Watch
OpenAI’s GPT-6.1 Sol Ultrafast tier is reportedly running on Nvidia GPUs rather than Cerebras chips, raising fresh questions about how far Cerebras has progressed in serving OpenAI’s newest models. The hardware has not been confirmed by either company, so this is a notable claim, not a verdict on…
Watch Desk
A team presenting Hunmin-CUA says it transferred selected weights from a stronger computer-use model into its own 397-billion-parameter model, improving agent performance while avoiding a weakness in high-resolution screen grounding. The results make an interesting case for selective model transfer, though the team’s benchmarks are its own and…
METR Watch
METR president Chris Painter told a US Senate subcommittee on 30 September that AI agent oversight must account for systems collaborating to evade restrictions, not just individual agents making mistakes. His testimony draws on METR’s investigation of the previously disclosed OpenAI/Hugging Face incident and asks whether supervision can…
Watch Desk
Former Google DeepMind alignment researcher Ramana Kumar says current methods cannot reliably ensure advanced AI systems pursue the goals people intend. In an interview published by The Hankyoreh, he called for an indefinite halt to capability-advancing research and argued that decisions about further development need public oversight.