NVIDIA agreed to acquire Hugging Face for $12.93 billion, folding the leading open-model registry and developer ecosystem into its hardware and cloud stack. The deal pairs NVIDIA’s supercomputing and CUDA advantage with Hugging Face’s community and tooling — accelerating integration but raising questions about openness, competition and platform neutrality.
OpenAI has published a 166‑page machine‑checked paper it says an internal model produced that proves finite‑time blowup for the three‑dimensional Navier–Stokes equations. The result includes a Lean formalization, but outside mathematicians are disputing priority, provenance and the scientific norms around AI‑assisted discovery.
NVIDIA confirmed RTX Spark N1X configurations and told partners the first laptops and small desktops will begin shipping in October 2026. The Grace CPU + Blackwell RTX superchip pairs up to 6,144 CUDA cores with unified memory (up to 128 GB) to make large LLMs, agents and pro creative workloads feasible on a single Windows PC.
