TOPIC HUB

Nvidia AI News

Nvidia AI news from AiLookout: the latest on Nvidia GPUs, AI data center systems, open models, earnings, export rules and deals, each linked to its source.

Claude News Anthropic News OpenAI News Gemini News AI Agents News Meta AI News xAI and Grok News DeepSeek News Mistral News

Latest Nvidia news

95 stories.

Newest first, each with its sources linked.

About our Nvidia coverage

Nvidia designs the GPUs and networking systems that train and run most large AI models, and it increasingly publishes its own open models and software. This hub collects AiLookout’s Nvidia AI news, newest first: data center chips and systems, open-weight model releases, partnerships and investments, earnings and supply, export controls, and regulatory scrutiny of its deals.

Performance and efficiency figures from Nvidia are reported as Nvidia’s claims and linked to the announcement or technical post. Market, regulatory and supply-chain reporting is linked to the filing, agency or outlet it came from, so every number on the page can be traced back to a source.

Use this page to follow the hardware side of the AI race and the companies that depend on it. For the model makers that buy Nvidia compute, see our OpenAI news, Anthropic news and xAI and Grok news hubs. Every story here links its primary sources, shows its publication date, and discloses the AI assistance used to write it. New stories are added as they are published, and older coverage stays available further down the list.

Latest Nvidia AI News

A photorealistic low-angle view of dark GPU server racks in a data center hall with red accent lighting, no logos or text.

Industry6 min read

Nvidia caps Chinese business at 49% for data center clients

The Financial Times reports Nvidia is telling its data center partners to cut reliance on Chinese customers: from now on, cloud clients must sell at least 51% of services to non-Chinese companies, with the instructions delivered verbally and applying only to new contracts.

3 Nov 2026News

AI-generated extreme macro photograph of a bare GPU die on a circuit board under a single hard light. AI-generated.

Industry8 min read

Trajectory raises $45M from Sequoia to let domain experts post-train open models

An 18-person startup founded by three Stanford classmates a year ago raised a $45 million Series A at a $300 million valuation for a platform that lets lawyers and doctors retrain open models on their own feedback.

3 Nov 2026News

AI-generated close-up of the Meta Max humanoid robot prototype showing five-fingered hands and a screen face, dark cinematic lighting

Research5 min read

Meta builds a humanoid robot prototype with Fourier in three months

The 106-pound Max humanoid runs on Nvidia's Jetson Thor, mimics facial expressions on a screen, and is already testing a second version for household tasks.

1 Nov 2026News

AI-generated radio telescope array at night.

Research5 min read

SETI and Breakthrough Listen deploy agentic AI to hunt alien signals at the Allen Telescope Array

Breakthrough Listen and the SETI Institute have integrated agentic AI into the Allen Telescope Array in Northern California, using NVIDIA's Holoscan platform to let AI agents steer antennas, calibrate instruments, correct data errors, and run detection models from natural language prompts — automating the search for extraterrestrial intelligence at a level never before possible.

31 Oct 2026News

AI-generated entrepreneur in Toronto office.

Industry5 min read

Cohere raises up to $7 billion in Canada's largest startup funding to build sovereign frontier AI

Cohere announced the initial close of a Series E funding round targeting up to CAD $7 billion — the largest by any Canadian startup — with the Government of Canada taking a $1.6 billion equity stake, led by Schwarz Group with NVIDIA, Bell Ventures and others participating, valuing the company at $35.7 billion.

30 Oct 2026News

Industry5 min read

Huang and Musk clash over AI neocloud debt crisis as CoreWeave bond yields hit 15 percent

Nvidia CEO Jensen Huang says AI demand is booming, while Elon Musk warns of 'death and consolidation' among debt-funded AI cloud providers as CoreWeave bond yields reached 15% and CDS spreads hit 1,100 bps.

29 Oct 2026News

AI-generated image of a smartphone on an Apple-style developer desk.

Models5 min read

Hugging Face releases open Swift packages for on-device speech and decisions on Apple platforms

Hugging Face's mattt released two open-source Swift packages — AnySpeechRecognizer and AnyDecisionModel — that let Apple developers run speech recognition and decision models entirely on device, with a choice of open models including NVIDIA's Parakeet, Alibaba's Qwen3-ASR and decision models from Jev or Ollama.

29 Oct 2026News

AI-generated image of a courthouse with a lawyer on the steps.

Industry5 min read

DOJ asks Nvidia questions about its $12.9 billion Hugging Face deal

The US Department of Justice has asked Nvidia for information about how its $12.9 billion acquisition of Hugging Face could impact competitors and harm competition in AI, Reuters and CNBC reported on 28 October 2026, at the same time the antitrust staff is examining the deal.

29 Oct 2026News

AI-generated image of headphones and an audio waveform editor on a desk.

Models5 min read

NVIDIA releases Nemotron 3.5 Transcribe, an open-weight speech model that transcribes 40 languages

NVIDIA released Nemotron 3.5 Transcribe on 28 October 2026, an open-weight model combining transcription, speaker attribution, translation and audio question answering in one checkpoint: 40 transcription languages, 31 translation languages and up to eight speaker labels per recording.

28 Oct 2026News

AI-generated editorial-style image of a data center aisle with server racks and a worker in a lab coat.

Industry5 min read

OpenAI plans to buy Google's AI chips, The Information reports

OpenAI is planning to buy Google's AI chips, The Information reported on Tuesday, citing a person familiar with the matter, deepening ties between the two rivals.

28 Oct 2026News

Industry5 min read

Naver signs $9 billion AI infrastructure deal with Brookfield

Naver signed a $9 billion agreement with Brookfield to secure AI computing infrastructure, clearing the path for NVIDIA's $1 billion investment that makes it Naver's third-largest shareholder.

27 Oct 2026News

An NVIDIA GeForce GTX 1070 Founders Edition graphics card on a white background.

Industry5 min read

NVIDIA and Meta expand their Muse partnership as NVIDIA invests $3 billion

NVIDIA and Meta said on 27 October they will deepen their collaboration on the Muse assistant and Muse models, and NVIDIA will invest $3 billion in Meta's equity offering.

27 Oct 2026News

A SpaceX Falcon 9 first stage landing at Landing Zone 1 during the CRS-13 mission, trailing engines lit against a blue sky.

Industry5 min read

SpaceX locks in a $40 billion GPU financing lease with Apollo

SpaceX signed a binding commitment for about $40 billion in GPU financing with an Apollo-managed entity, in what both companies call the largest committed GPU financing ever.

26 Oct 2026News

Industry5 min read

Cadence unveils Maestro, AI models for chip design that can run in-house

Cadence launched Maestro, a portfolio of domain-specialized AI models for its chip and system design agents, built partly on NVIDIA Nemotron and secured with NVIDIA OpenShell.

26 Oct 2026News

Industry9 min read

NVIDIA quietly lowers Vera Rubin memory and NVLink bandwidth specs

NVIDIA's Vera Rubin pages now list 19.2 TB/s of HBM4 bandwidth per GPU instead of 22 TB/s and 17 percent less NVLink bandwidth, while networking went up. NVIDIA has not explained the change.

26 Oct 2026News

Industry9 min read

NVIDIA floats a $2,000 per GPU yearly fee for its AI agent software

Jensen Huang said at GTC Berlin that running NVIDIA's agent software costs about $2,000 per GPU per year. NVIDIA has published no terms, and its AI Enterprise licence already lists at $4,500.

26 Oct 2026News

Industry8 min read

Musk says SpaceX closed $40 billion of GPU debt and predicts a neocloud shakeout

Elon Musk says SpaceX closed $40 billion of callable debt for GPUs arriving in early 2027, and that lenders have shut out AI-only cloud companies. Reported rates ran near 7.75 to 8.3 percent.

26 Oct 2026News

The former Unterweser nuclear power plant across a marsh and tidal mudflats under a grey sky: a tall grey chimney on the left, a large weathered concrete containment dome in the centre, and trees and industrial structures around them.

Industry7 min read

PreussenElektra and NVIDIA will study AI factories at Germany's former nuclear plant sites

E.ON's nuclear-dismantling arm and NVIDIA will match eight shut German plant sites against AI data-centre needs. No site or investment has been decided.

25 Oct 2026News

Research9 min read

NVIDIA open-sources Flora, the engine that labels 200,000 hours of robot video a day for GR00T

NVIDIA's Flora is three open-weight models that caption robot videos and score each step. NVIDIA says it beats GPT-6 Astra on its own test at an estimated 125x lower cost.

25 Oct 2026News

Two large graphics cards stacked on a red cloth: a black Gigabyte RTX 3090 with three fans on top, and a longer Asus ROG Strix RTX 4090 with three fans and red and blue accents below, on a blurred backdrop.

Tools & Agents9 min read

Open-source gpu-unlock lifts the half-rate tensor limit on RTX 3090 and 4090 cards

A Linux kernel module from developer Zagitta, and a new fork for the RTX 4090, raise GeForce FP16 and BF16 Tensor Core math with FP32 accumulation to the full rate before NVIDIA's driver loads.

25 Oct 2026News

Four NVIDIA H100 graphics cards stacked on top of each other, with gold ribbed heatsinks and black NVLink bridges, on a white marble surface fading into a black background.

Industry7 min read

a16z says Nvidia GPU rental rates are up since January, with the H200 about 160 percent higher

a16z, citing Ornn Data, says every NVIDIA GPU it tracks rents for more than on 1 January 2026, with the H200 up about 160 percent, the B200 about 120 percent and the H100 about 35 percent.

24 Oct 2026News

Tools & Agents9 min read

Cohere's Model Vault beta runs inference in attested, encrypted H100 enclaves on Azure

Cohere opened a public beta of confidential computing on Model Vault. Prompts and responses are encrypted end to end into attested CPU and GPU enclaves that customers verify before sending data.

24 Oct 2026News

Models8 min read

NVIDIA releases Nemotron 3.5 Bolt E4B, a 4GB model for local voice agents

NVIDIA's Nemotron 3.5 Bolt E4B 4GB is a Gemma 4-based small model trained with reinforcement learning only. It leads 4B peers on instruction following but loses some vision skill.

24 Oct 2026News

Industry8 min read

Higgsfield passes $1 billion in annualized revenue as TCV and NVIDIA extend its Series B

AI video platform Higgsfield says it crossed a $1 billion revenue run rate in September, 18 months after launch, as TCV and NVIDIA joined an extension of its August Series B.

24 Oct 2026News

Industry7 min read

Firebird gets a US license for 45,000 NVIDIA GPUs in Kazakhstan as Jensen Huang visits

Firebird signed a deal with Kazakhstan for up to 500 MW of AI capacity and a US export licence for 45,000 Grace Blackwell GPUs, as Jensen Huang met President Tokayev in Astana.

24 Oct 2026News

Three small two-legged Microduck robots in blue, white and graphite with single round camera eyes stand around a white ball on a green court, while a person in striped socks reaches down to one of them.

Industry9 min read

Microduck, Hugging Face's $399 robot duck, nears 20,000 preorders after Huang's GTC demo

Pollen Robotics' 25 cm biped went viral at NVIDIA's GTC Berlin keynote, and Clement Delangue says preorders are crossing 20,000 ahead of a before-Christmas delivery target.

24 Oct 2026News

AI-generated illustration of a single gold-topped data-center GPU module mounted on a black metal server tray, with a red network cable plugged into it and a small amber status light glowing, on a dark background.

Industry10 min read

Lava finds 32 Nvidia B300 GPUs on China-linked networks through open monitoring endpoints

Unprotected Nvidia DCGM Exporter pages revealed 32 export-controlled B300 GPUs on China-geolocated IP addresses, part of more than 12,000 GPUs anyone online could see. Nvidia calls smuggling a nonfactor.

23 Oct 2026News

Industry7 min read

inDrive and Pony.ai plan robotaxi tests in Kazakhstan on NVIDIA Hyperion

inDrive, Pony.ai and NVIDIA signed an MOU to explore Gen-7 robotaxis on the inDrive app, starting in Astana. inDrive told TechCrunch supervised tests should start in the first half of 2027.

23 Oct 2026News

Industry7 min read

CoreWeave launches one-command templates for Isaac Sim, CARLA, MuJoCo and Cosmos 3

CoreWeave's Physical AI Launchables package tested robotics and driving simulation jobs as templates you start with one command, with agent skills for Claude Code and Codex. Prices were not published.

23 Oct 2026News

Industry8 min read

Agility Robotics moves Digit 5 humanoid onto NVIDIA IGX Thor ahead of Nasdaq listing

NVIDIA becomes the exclusive onboard compute supplier for the general availability Digit 5, which also adds Halos safety models. Early access is due in the first half of 2027.

23 Oct 2026News

Industry8 min read

NVIDIA adds Samsung Electro-Mechanics as it plans 2.5x more AI chip substrate capacity

At GTC Berlin, NVIDIA unveiled deals with six suppliers to expand substrates, packaging, testing and optical engines, including a $1.1 billion advance for a new KYEC test plant in the US.

23 Oct 2026News

Tools & Agents7 min read

Liquid AI launches RIPL to serve small language models in under 50 ms

Liquid AI's new enterprise engine runs its LFM2.5 models inside live search, ads and payment requests, and the company reports 5.6 to 28.4 times lower p95 latency than Qwen3.5 on vLLM in its own tests.

22 Oct 2026News

The OpenAI, Anthropic and NVIDIA logos in white and green across the top of a dark background, above three overlapping cream paper forms with grey lines and tick boxes.

Industry6 min read

OpenAI, Anthropic and NVIDIA each filed their biggest lobbying quarter yet

New federal disclosures show NVIDIA spent $2.15 million, Anthropic $2.11 million and OpenAI $1.53 million on in-house lobbying from July to September 2026, a record for each.

22 Oct 2026News

Industry7 min read

NVIDIA launches Minerva to trace AI factory faults from a single part to the job it slows

NVIDIA's new Minerva AIOps platform maps GPUs, networking, storage and jobs into one graph and uses machine learning and an AI assistant to find failing parts. IREN uses it; Dell, Lenovo and Supermicro will sell it.

22 Oct 2026News

Industry7 min read

NVIDIA Confidential Bare Metal lets AI labs lock their models on rented GPU servers

NVIDIA's new Confidential Computing mode aims to protect model weights and data on dedicated third-party GPU servers; Google Cloud's A4X Max, Cohere Model Vault and Lambda plan to support it.

22 Oct 2026News

Research7 min read

Perplexity's JOLT trains one model to be both teacher and student

Perplexity and NVIDIA researchers' JOLT trains a privileged teacher and a student in one model; on MATH it matched GRPO with 13x fewer completion tokens, and JOLT+ led every reported test.

22 Oct 2026Research

Tools & Agents8 min read

ServiceNow makes Arc generally available, a work agent policed by NVIDIA OpenShell

ServiceNow's Arc agent is now generally available on Mac. It plans and runs multi-day tasks across email, Teams and ServiceNow, with IT rules enforced at runtime by NVIDIA OpenShell; hardware revocation comes later.

22 Oct 2026News

Tools & Agents9 min read

NVIDIA expands Cosmos with Data Factory blueprints and Cosmos 3 Nano simulation models

At GTC Berlin, NVIDIA released five Cosmos Data Factory blueprints for synthetic training data, extended Cosmos Framework to models like GR00T and Qwen, and added Cosmos 3 Nano variants for closed-loop policy testing.

22 Oct 2026News

Tools & Agents8 min read

Red Hat brings NVIDIA's OpenShell agent sandbox and BlueField monitoring to OpenShift

At GTC Berlin, Red Hat and NVIDIA said OpenShift can watch and isolate AI agents from BlueField DPUs with DOCA Argus, added a governed model-access design, and readied RHEL for NVIDIA Vera CPUs.

22 Oct 2026News

Industry8 min read

Bolt picks NVIDIA Alpamayo software for its Lucid robotaxis, starting in Madrid in 2027

Bolt has chosen NVIDIA's Alpamayo Stack to drive its European robotaxi programme, starting with data-collection cars in Madrid in 2027 and planning at least 25,000 Level 4-ready Lucid vehicles on NVIDIA Hyperion.

22 Oct 2026News

Models9 min read

NVIDIA releases Kumo Relational, a model that predicts straight from business databases

NVIDIA's Kumo Relational reads linked database tables and predicts churn, demand or fraud in one pass, without training or feature engineering. NVIDIA says it beats supervised models on 41 benchmark tasks.

21 Oct 2026News

Models8 min read

NVIDIA releases Nemotron 3.5 Embed, Rerank and Parse open retrieval models

Five open models search document pages as images and text, rerank evidence for agents and parse PDFs into structure. NVIDIA says reranking cut agent tokens per query by about 14 percent.

21 Oct 2026News

Industry7 min read

Agile Robots unveils Agile ONE SN, a reference humanoid built on NVIDIA Isaac GR00T

At GTC Berlin, NVIDIA showed Agile ONE SN, a 1.4-metre, 31-degree-of-freedom research humanoid with Jetson Thor onboard and two 20-degree-of-freedom hands, due in early 2027, alongside Isaac OS and Isaac Lab 3.0.

21 Oct 2026News

Industry7 min read

Google Cloud says its A5X Vera Rubin instances reach general availability next month

Google Cloud and NVIDIA say A5X instances on Vera Rubin NVL72 go generally available in November, with Europe to follow and a plan for up to 960,000 Rubin GPUs. David Silver's Ineffable went first.

21 Oct 2026News

Six colored ribbon renders of designed biomolecular structures on a white background: a protein-protein complex, a protein bound to DNA, a double-helix, two enzyme pockets around small molecules, and a ringed protein assembly.

Research10 min read

NVIDIA and the Baker Lab release RFD4-Proteina, an open protein-design model

The BSD-licensed BioNeMo model from NVIDIA and UW's Institute for Protein Design designs proteins, enzymes, binders and antibodies, and ships with weights, code and wet-lab results.

21 Oct 2026News

Tools & Agents9 min read

Ant Group and vLLM launch Humming, a kernel library for quantized inference

The Apache-2.0 library compiles quantized GEMM kernels on demand for every NVIDIA GPU since Turing; its authors report up to 43% more DeepSeek-V4.1-Flash throughput on eight H20s.

21 Oct 2026News

Industry7 min read

NVIDIA unveils DSX POD, containerized Vera Rubin AI factories it says deploy in 14 days

Announced at GTC Berlin, the reference design has partners pre-build complete Vera Rubin data centres in containers, with 17 partners targeting deliveries in early 2027.

21 Oct 2026News

An endoscopic view inside the abdomen during robotic gallbladder surgery: a metal grasper enters from the left and a hooked instrument from the upper right over yellow and pink tissue, with round console icons in the corners and small CMR Surgical marks.

Research9 min read

NVIDIA releases Cosmos3-H-Dreams, a real-time surgical simulator trained on real operations

NVIDIA's Cosmos3-H-Dreams generates a surgical robot's camera view live as instruments move, trained on CMR Surgical Versius gallbladder operations. Weights and code are open; it is a research tool, not for patient care.

20 Oct 2026News

All AI news