Nvidia's Open Speech Model Does Three Jobs
Nvidia's Nemotron 3.5 Transcribe handles transcripts, translation and speaker labels on one GPU, and tops the open models on Hugging Face's English leaderboard.
Read the storyNVIDIA
Reporting and context on NVIDIA, with a closer look at the decisions and developments that matter.
Follow NVIDIA
Explore the coverage Get AI Secret’s free daily briefing, including NVIDIA and the wider AI industry.
At a glance
NVIDIA develops accelerated computing hardware and software. Coverage follows its GPUs, CUDA ecosystem, AI infrastructure and business decisions.
- Focus
- GPUs; CUDA; AI systems
- Reference
- NVIDIA
Reviewed 19 October 2026. Dates retain the precision of the source; retrospective disclosures are identified in the event. Fiscal-year, quarterly and Data Center sales are labelled separately; none is presented as ARR.
Company milestones
Revenue, funding, product launches and major company events, newest first. Expand the earlier milestones for more history.
Q2 FY2027 revenue reaches $96.2B
Quarterly revenue grows 106% year over year; Data Center contributes $89.0B, up 117%. These are quarterly sales, not ARR.
Source: NVIDIAVera Rubin ramps into full production
NVIDIA reports partner deployments of its new platform, extending the hardware roadmap beyond Blackwell.
Source: NVIDIA
Show earlier milestonesHide earlier milestones2026–1993 · 10 milestones
Earlier revenue, financing and company milestones.
FY2026 revenue rises to $215.9B
Full-year revenue increases 65%; Data Center revenue reaches $193.7B. Fiscal 2026 ends in January 2026.
Source: NVIDIAAnnounces a 10GW OpenAI infrastructure plan
The companies sign a letter of intent. Planned capacity and contemplated investment are not recorded as delivered capacity or revenue.
Source: OpenAI / NVIDIAFY2025 revenue reaches $130.5B
Annual sales grow 114%; fourth-quarter Data Center revenue is $35.6B, showing the scale of the generative-AI infrastructure expansion.
Source: NVIDIAIntroduces Blackwell
The new GPU platform integrates compute, networking and software for large-scale model training and inference.
Source: NVIDIACompletes the $7B Mellanox acquisition
High-performance networking becomes a larger part of NVIDIA’s data-center platform alongside GPUs.
Source: NVIDIAIntroduces the DGX-1 AI system
The integrated GPU server combines accelerators, interconnects and optimized software for deep-learning development.
Source: NVIDIAAlexNet demonstrates GPU-trained deep learning
The ImageNet breakthrough uses NVIDIA GPUs, marking an important step in the shift from graphics hardware to AI computing.
Source: NVIDIA retrospectiveIntroduces CUDA
The programming platform opens GPU acceleration to general-purpose computing, becoming a foundation for later deep-learning workloads.
Source: NVIDIAGoes public
NVIDIA’s investor FAQ dates its public-market debut to January 22, before its later expansion into general-purpose GPU computing.
Source: NVIDIANVIDIA is founded
The company starts in computer graphics before expanding into accelerated computing and AI infrastructure.
Source: NVIDIA
The latest on NVIDIA
Newest firstRelated reporting
Briefs, articles and relevant sections, newest first. Explore older coverage below.
More coverage
1 min read · Members
Nvidia, AMD Put $4 Billion Into Meta
Nvidia and AMD will buy $4 billion of Meta's stock sale, and Meta is buying more of their chips at the same time.
1 min read · Members
Huang Quotes $2,000 a Year per GPU
Huang prices Nvidia's agent software at about $2,000 a GPU yearly, under half its AI Enterprise list price, and a shareholder already sees $10 billion.
1 min read · Members
Nvidia Eyes Germany's Shut Nuclear Sites
Nvidia and PreussenElektra will study turning Germany's former nuclear sites into AI data centers, reusing grid links and land, with no investment decided.
1 min read · Members
Nvidia Opens Flora, Its Robot Video Labeler
Nvidia says its open Flora models label robot video 125 times cheaper than GPT-6 Astra, a figure projected from an assumed GPU rental rate.
1 min read · Members
Nvidia's Smaller DGX Spark Starts at $4,999
Nvidia calls its $4,999 DGX Spark 64GB accessible, yet it costs more than the 128GB original did at launch, and two units cost about $10,000.
1 min read · Members
Huang Tells Kazakhstan to Export Intelligence
On his first visit, Huang urged Kazakhstan to export intelligence, as Firebird lined up 45,000 Nvidia GPUs and Tokayev gave him a black belt.
1 min read · Members
OpenAI, Anthropic and Nvidia Set Lobbying Records
Nvidia, Anthropic and OpenAI each set lobbying records this summer; OpenAI, under half a million a quarter until late 2024, spent $1.53 million.
1 min read · Members
Nvidia Builds a Cage for AI Agents
Nvidia launched an agent safety stack that splits thinking, acting and policing across three of its chips, plus confidential bare metal for model builders.
1 min read · Members
Nvidia Pays to Lock Up Chip Substrates
Nvidia is co-funding Samsung Electro-Mechanics substrate lines, reportedly with about $2 billion, while it pays several suppliers ahead as substrates run short.
