AI hardware & industry

12 Nov 2026NVIDIA's board announced that Jensen Huang, who co-founded the company 33 years ago, has resigned as president, chief executive and board member with immediate effect. Colette Kress, the chief financial officer since 2013, is interim chief executive and will stay on as CFO while the board searches inside and outside the company for a permanent successor. Harvey C. Jones, a former chief executive of Synopsys, becomes the board's first-ever chair. The board says it expects operations, customer commitments and the product roadmap to continue without interruption. NVIDIA's announcement gives no reason for the resignation. I have left out press reports about the circumstances, because I could not verify them from a primary source. NVIDIA will report its third-quarter results on 17 November as planned.

21 Oct 2026At GTC in Berlin, NVIDIA announced the DSX Pod AI Factory: seven Vera Rubin racks packed into a shipping container weighing about 28,000 kg, which can be delivered by lorry. NVIDIA says it can cut the time for a new AI data centre to start serving from about six months to 14 days. CNBC reports that deliveries are expected from early 2027 and that analysts estimate the cost at more than $7 million per rack.

A5X on Vera Rubin NVL72

21 Oct 2026Google Cloud's A5X instances, built on NVIDIA Vera Rubin NVL72 systems, are due to become generally available next month, with expansion to Europe planned. NVIDIA says Google plans to deploy up to 960,000 Rubin GPUs. The first named customer is London-based Ineffable Intelligence, founded by David Silver.

Google TPUs on Oracle Cloud

27 Oct 2026Google and Oracle announced that Google's TPU chips will be offered through Oracle Cloud Infrastructure, making Oracle the first hyperscaler outside Google to offer them. Capacity is due from calendar 2027. Google says its eighth-generation TPUs, announced in April, reach up to 12.6 petaflops of FP4 compute per chip. It is a notable shift: TPUs have until now been a reason to use Google's own cloud.

28 Oct 2026NVIDIA says RTX Spark Windows laptops are now available from ASUS, Dell, HP, Lenovo, Microsoft and MSI, with up to 128 GB of unified memory and up to 1 petaflop of FP4 AI compute. The 4 GB Nemotron 3.5 Bolt E4B model is now out for PCs, along with a new open-weight speech model, Nemotron 3.5 Transcribe.

5 Nov 2026OpenAI published a community compact for the compute campuses it owns and operates, and a policy blueprint for lawmakers. It promises to pay the full cost of the power and grid upgrades it needs, publish water and electricity use, and avoid NDAs with public officials. At its planned Effingham County, Georgia, campus, OpenAI says its agreement with Georgia Power lets the utility cut the power delivered to it by up to 1,000 megawatts when demand is high. It also cites a $4.8 million grant for free school meals in Effingham County and a joint $6 million childcare commitment in Pike County, Ohio. See the policy page for the commitments in detail.

5 Nov 2026Thinking Machines Lab says its new model's first pre-training phase covered 8.4 trillion tokens and ran for 68 hours on NVIDIA GB300 NVL72 racks, with training carried out in NVIDIA's 4-bit NVFP4 format. It says the main reinforcement-learning run used 2,048 GPUs for nine days and more than 4 × 10²³ floating-point operations. These are the company's own figures, and it says more detail on the NVFP4 recipe will follow.

Oct 2026DeepSeek has published DeepInfer on GitHub, an inference engine for its DeepSeek-V4.1-Flash model. It supports NVIDIA's Blackwell-generation (SM100) GPUs and Huawei's Ascend 950 chips. The public release covers offline inference only. DeepSeek says its KV cache system, disaggregated serving and load balancing are not included. Support for a Chinese accelerator alongside NVIDIA is the notable part, in the same week that the US ordered federal civilian agencies to remove DeepSeek (see the policy page).