Europe's DARPA clones put €40M into AI that designs chips

Europe's answer to Nvidia just got a budget line. NADI, the Netherlands' brand-new DARPA-style agency, and Germany's SPRIND are putting €40 million into a challenge that pays small teams to let AI design the chips AI itself runs on.

The program is the first outing for NADI — the National Agency for Disruptive Innovation — which the Dutch government only stood up this month with €500 million in seed funding announced on Prinsjesdag. Together with Germany's SPRIND, the federal leap-agency that now gets a formal European peer, it is committing the €40 million over 20 months to small teams using AI to accelerate the design of chips for training and inference. The money is staged: up to €2.6 million per team across an eight-month first stage, then up to €7.0 million in a twelve-month second stage — a maximum of €9.6 million each, with follow-on funding "in the tens of millions" planned to carry winning designs to tape-out. Applications close November 30, with webinars in October and November.

The logic is Europe's standing chip grievance. NADI co-founder Jelle Prins argues that Nvidia GPUs are built for training and waste money at inference — "doing your shopping with a truck" when a trolley would do — and SPRIND's Jano Costard wants "several orders of magnitude" of acceleration in a design process that today takes years. The ambition is real; the scale is not. Forty million euros is roughly the per-team budget of the American labs' training runs, and a DARPA-style program only pays off if the breakthrough models actually get manufactured at scale. Still, two governments jointly funding AI-native chip design beats thirty more press releases about European AI sovereignty — and it puts real money behind a specific technical bet rather than a general one.

Germany's dominant legal publisher just bought control of Europe's sovereign legal AI. Noxtua, the Berlin legal-AI startup spun out of Oxford and Imperial research, has closed a Series C of more than €100 million — and C.H.BECK, the 260-year-old publisher behind the German-speaking world's largest legal database, becomes its majority shareholder in the largest investment of its history. Austrian legal publisher MANZ joined the round as a minority investor, while earlier backers including Global Brain, KDDI, CMS and Dentons are exiting.

The deal is the clearest sign yet that "sovereign AI" is becoming a product category with real revenue behind it. Noxtua builds research, analysis and drafting tools on curated legal data from European publishers, claims over 30,000 users and a 5x revenue jump in four months, and holds the BSI C5 and ISO certifications that make German law firms and courts actually able to buy it. When the company that owns the legal data also owns the model vendor, Europe gets a genuinely independent legal AI stack — and everyone else selling AI to lawyers just got a much tougher benchmark to clear.

What to watch: whether the NADI–SPRIND challenge actually produces tape-outs rather than papers — the follow-on funding decision will tell us by mid-2027 — and whether Noxtua's publisher-owned playbook gets copied in France and the Nordics.

If Brussels paid you to reinvent the chip, where would you start? Tell us in the comments.

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Microsoft finished pre-training its biggest in-house model yet — then used it to make a point about the company it bankrolls. Microsoft AI completed pre-training of MAI-Base-2, a mixture-of-experts foundation model with 80 billion active and 2.8 trillion total parameters, and says the recipe is 4.2–12.3x more pre-training FLOPs-efficient than DeepSeek-V4-Pro-Base. The run itself is the number worth keeping: 16,000 Nvidia GB200 GPUs for 38 days across August and September, at 21% average model

Amazon's new Bee devices listen all day — for $11.99 a month

Amazon turned its always-listening Bee line into a two-device family on Thursday, and NVIDIA opened a 2.8-trillion-token Indic corpus recipe. The connecting thread: both are bets that the next AI money is in memory — yours, or your model's. Amazon put two new always-listening devices on sale, and both now lead to a subscription. Bee Frontier, the next-generation wristband that also clips to clothing, ships now at $74.99 in Onyx, Lavender and Sage — up from the $50 predecessor — with seven-day

Perplexity open-sources its agent sandbox after 118 failed escapes

Perplexity handed its agent containment layer to the open-source world, and Washington's AI task force stopped reading briefings and started working. Perplexity open-sourced SPACE on Thursday — the sandbox it built to keep its own AI agents from getting out — and says nine frontier models failed to break out of it across 118 attempts. Each agent runs inside a small virtual machine with its own operating system, with credentials and network access managed outside the sandbox so an agent can re

Read more

Brain organoids classified Braille at 83% — in a trio, not alone

Suleyman says Microsoft's base model beat OpenAI's on a quarter compute

Amazon's new Bee devices listen all day — for $11.99 a month

Perplexity open-sources its agent sandbox after 118 failed escapes