Desk
Broad artificial intelligence coverage.
Google’s generally available video model stretches native clips from 10 to 15 seconds, adds conversational editing and tighter reference lock, and drops to $0.07 per second at 720p. Cheaper, editable clips suit ad variants; provenance has to keep pace.
NVIDIA opens DSX OS — AICR, NVCRE, and NVSentinel — so GPU Kubernetes clusters get reproducible recipes, workload certification, and continuous health loops instead of hope-based handovers.
Nvidia’s profile of its Toronto engineer, known as llm_wizard and self-styled No. 1 Nemotron fan, is corporate soft power. His underlying argument holds: open models only democratize AI when someone teaches people to fine-tune, evaluate and ship them, and when the full recipe is public.
Proof-Driven Development Kit turns plain-language intent into Verus formal specs, proves agent code for every input, and labels tests as evidence where proofs can’t reach. On a 5,000-line Rust app it verified login rules without line-by-line review. It is Rust-only and a research prototype.
From Blackwell Ultra to Vera Rubin, Nvidia says HBM per GPU rose about 3% while agentic throughput per dollar rose up to 30x on SemiAnalysis AgentX. Its answer is a memory hierarchy, compact formats and KV-cache offload. The direction is right; the numbers need outside checks.
NVIDIA's developer deep dive names five co-design levers — NVFP4, Wide-EP, Dynamo, Vera CPU, and CMX — that push Vera Rubin to up to 30× GB300 throughput per megawatt on the same 288 GB HBM.
Nvidia’s cloud service is shipping DLSS 5 with 3D-Guided Neural Rendering in Hogwarts Legacy, tuned by Avalanche to keep the art style. It shows how streaming can deliver next-gen graphics without a new GPU, though the full experience sits behind the Ultimate tier.
Box Content Retrieval embeds Nvidia’s open Nemotron models in its core so agents get relevant chunks instead of reading whole files. Nvidia expects 4x speed, 20% fewer errors and 90% fewer tokens. The moat isn’t the open weights; it’s Box’s permissions and metadata.
The landscaper-turned-comedian, with about two million followers, told WIRED that all the big AI companies have approached him, including OpenAI offers and a reported xAI Tesla pitch. He keeps saying no, and his clips remain a public check on the labs’ everyday-usefulness pitch.
Amazon, Microsoft, Alphabet and Meta told Wall Street the spending continues into 2027 despite murky returns. Alphabet suffered its worst one-day drop on record, then rebounded as midterm results raised the odds of data center limits. The brake may come from outside the boardroom.
OpenAI’s specialist security model scores 48.5% on ExploitGym, helped produce exploits for 948 Linux kernel bugs and 81 new Firefox issues, and is rated Critical. It also stayed in scope on honeypot tests 99.93% to 100% of the time. Access breadth is the open question.
Together AI's new compiler keeps rewriting model forward passes as live traffic shifts, and the early numbers say a one-shot kernel tune is already the wrong mental model.