AI News from NVIDIA Blog
Latest coverage from NVIDIA Blog, summarized and scored for signal.
- AI Is Changing How Semiconductor Fabs Work — and Engineers Are Moving Faster Because of It — NVIDIA lays out its 'AI-defined fab' pitch, citing COMSOL and Applied Materials simulations running up to 10x faster with cuDSS as chipmakers face rising complexity and slow physical trials.
- Cities Worldwide Build and Scale AI Services With NVIDIA Physical and Agentic AI — NVIDIA is expanding physical and agentic AI deployments for city governments, with partners using Nemotron, Cosmos, Omniverse and Metropolis for resident-service agents, vision agents and shared municipal compute.
- Things That Go Beep in the Night: Agentic AI Joins the Search for Extraterrestrial Signals — Breakthrough Listen and the SETI Institute now let AI agents steer the Allen Telescope Array and process terabits per second of data on NVIDIA Holoscan in the search for alien signals.
- Heart of the Matter: How a Major Children’s Hospital Uses Open Source NVIDIA AI for Cardiac Care — Children's Hospital of Philadelphia is using open-source NVIDIA AI tools to generate 3D models of pediatric hearts in seconds, aiming to improve pre-surgical planning for children with congenital heart disease.
- Perplexity Portable Computer Is Now Available on Windows, Powered by NVIDIA RTX — Perplexity's Portable Computer AI agent is now available on Windows for NVIDIA GeForce RTX and RTX PRO GPUs with 24GB+ VRAM, running local agentic workflows on a post-trained Qwen 3.8 27B without sending data to the cloud or consuming Perplexity cloud credits.
- Skild AI Taps NVIDIA Physical AI to Teach Robots New Tasks From a Single Video — Skild AI's S1 robot foundation model learns previously unseen, long-horizon tasks from a single video demo using in-context learning — no retraining required. It achieves 66% step-level success vs. 9% for competing systems on novel multistep tasks, and the company hit $100M ARR 10 months after first commercial deployment.
- Physical AI Takes the Wheel: How the World’s Robotaxi Leaders Are Building With NVIDIA Technologies — NVIDIA details how every major commercial robotaxi program today runs on its three-computer stack — DGX for training, Omniverse/Cosmos for simulation, and in-vehicle compute — with VLA model improvements cutting trajectory prediction error by 43% in testing.
- d-Matrix Adopts NVIDIA NVLink Fusion for Rack-Scale XPU Deployment — d-Matrix will integrate its Raptor XPUs into NVIDIA's NVLink Fusion platform, gaining 6th-gen NVLink's 3 TB/s per-XPU all-to-all bandwidth and 3x lower latency than Ethernet — letting the inference-focused chipmaker skip building rack-scale infrastructure from scratch.
- NVIDIA Brings Real-Time AI to Broadcast, Sports and Global Streaming at IBC — NVIDIA expanded its AI for Media SDK suite at IBC 2026, with its Synthetic Video Detector reaching 99.3% accuracy on AI-generated video identification, and partnerships with Dalet, TwelveLabs, and Wowza embedding deepfake detection into broadcast workflows across 35,000+ deployments.
- Sparks Fly: NVIDIA Accelerates Local AI at IFA 2026 — At IFA 2026, NVIDIA announced RTX Spark mini PCs from Lenovo and Acer (shipping October), the free PAIR home-network inference router, 1.9x faster llama.cpp/vLLM builds, and native support for Nemotron 3.5 Lightning (30B), Meta Muse Glimmer (30B), and DeepSeek v4 Flash (284B MoE) on local hardware.
- ‘NBA 2K27’ With NVIDIA DLSS 5 Leads 26 New Games Coming to GeForce NOW — NVIDIA DLSS 5 makes its mainstream sports gaming debut in NBA 2K27 on GeForce NOW, using 3D-Guided Neural Rendering to replace traditional lighting pipelines—the first blockbuster title where neural rendering visibly shapes how skin, hair, and court surfaces look in real time.
- NVIDIA to Acquire Hugging Face — NVIDIA officially confirmed its $12.93 billion acquisition of Hugging Face, with CEO Jensen Huang pledging the platform stays open across all clouds and hardware — while NVIDIA's infrastructure scales the hub to its 18M developers and 200K enterprise customers globally.
- NVIDIA and CrowdStrike Strengthen Agentic Cybersecurity Frontier — CrowdStrike launched SafeMind, an agentic cybersecurity system built on NVIDIA Nemotron open models and post-trained on CrowdStrike's 15+ years of threat data, shipping natively in the Falcon platform with an autonomous offense-defense coevolution loop designed to continuously harden customer environments.
- GeForce NOW Gives Gamers More Ways to Play at Gamescom 2026 — Nvidia brings DLSS 4.5 controls to GeForce NOW cloud gaming at Gamescom 2026, letting Ultimate members tune Super Resolution, Dynamic Frame Generation, and Ray Reconstruction in cloud-streamed games — alongside GOG single sign-on, Steam Machine support, and Firefox compatibility.
- NVIDIA NVLink Fusion Expands With NVHBM Custom High-Bandwidth Memory — NVIDIA expanded NVLink Fusion with NVHBM, a custom high-bandwidth memory that moves the controller into the HBM stack rather than the XPU die, delivering 30% more bandwidth, 15% lower power, and 25% more compute die area. Amazon's Annapurna Labs is first to adopt it for Trainium4.
- Leading Publishers Bring Blockbuster PC Games and Technology to NVIDIA RTX Spark — NVIDIA confirmed EA, Ubisoft, and Embark Studios for RTX Spark at Gamescom 2026, and shipped DLSS 4.5 with a second-generation transformer model for ray reconstruction — the Windows AI gaming platform launches this fall.
- How XPUs Meet a World-Class AI Factory — NVIDIA's NVLink Fusion lets hyperscalers plug custom XPU silicon into NVIDIA's proven AI infrastructure stack, delivering 3x lower end-to-end latency and 10x higher packet rate vs. off-the-shelf Ethernet. The program targets cloud builders who want custom silicon advantages without rebuilding networking and software stacks from scratch.
- With Groq 3 LPX in Full Production, NVIDIA Extends Vera Rubin Inference for Agents — NVIDIA's Groq 3 LPX is now in full production within the Vera Rubin NVL72 rack-scale system, hitting 3,400 output tokens/second at 4x the throughput of competing platforms for 100K-token agentic workloads. SpaceXAI, CoreWeave, and Nebius have committed to or deployed the platform.
- Up to 30x More Work Per Watt: NVIDIA Vera Rubin NVL72 Sets a New Efficiency Standard for AI Agents — On-silicon benchmarks show NVIDIA Vera Rubin NVL72 delivers 30x higher throughput per megawatt and 35x lower token costs vs. GB300 NVL72 on agentic workloads — measured using SemiAnalysis AgentX, which replays real recorded coding agent sessions rather than synthetic inference runs.
- Securing the Infrastructure of Intelligence — NVIDIA is partnering with SB Energy to secure 4.25 gigawatts of AI factory capacity at Ohio's PORTS-Pike campus, with OpenAI as anchor tenant — each generation of GPU deployments representing ~1.5 million GPUs and $150-200B in potential NVIDIA revenue.
- Universitas Gadjah Mada, Indosat and NVIDIA Open Indonesia’s First University AI Center to Develop Local AI Talent — Indonesia opened its first university-based AI technology center in Yogyakarta — a joint initiative by NVIDIA, Indosat, UGM, and the Ministry of Communication giving local researchers access to sovereign GPU compute and NVIDIA's full AI software stack.
- Class Is in Session: GeForce NOW Levels Up Linux, Chromebooks and More — NVIDIA's GeForce NOW Linux native app exits beta with a Flatpak release, and cloud-side DLSS Frame Generation improvements reduce streaming latency for 1440p and 4K gaming—no local hardware upgrades required.
- NVIDIA CEO Tops Glassdoor’s 2026 List of Best CEOs — Jensen Huang earned 99% employee approval to top Glassdoor's 2026 Best CEOs list — reintroduced after a five-year hiatus — with NVIDIA also ranking #1 on Glassdoor's Best Places to Work in Technology.
- NVIDIA AI Factory Compute Is Becoming an Investable Asset Class — NVIDIA partnered with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to mobilize over $500 billion in third-party capital for AI infrastructure — formally positioning GPU compute as an institutional asset class financeable like power plants or toll roads.
- Why Scaling AI Compute Performance Requires a New Power Architecture — NVIDIA, Google, and Microsoft are co-developing an 800 VDC direct-current power architecture for AI data centers through the Open Compute Project, with 80+ equipment partners building to the spec and NVIDIA's hybrid power rack arriving H2 2026 — letting existing AC facilities serve next-generation GPU racks without rewiring.