NVIDIA: $4.85T GPU Giant & Nemotron AI Models (2026)
NVIDIA, founded 1993 by Jensen Huang, is the world's most valuable company at $4.85T with $215.9B FY2026 revenue. Builds Blackwell/Rubin GPUs and open Nemotron AI models.
NVIDIA is a Santa Clara, CA GPU and AI infrastructure company founded in 1993 by Jensen Huang, now the world's most valuable public company at a $4.85 trillion market cap (July 2026) with $215.9 billion in FY2026 revenue. It builds the Blackwell/Rubin GPU platforms, CUDA software, and the open-weight Nemotron 3 model family, and employs roughly 42,000 people worldwide.
NVIDIA, founded April 5, 1993 by Jensen Huang, Chris Malachowsky, and Curtis Priem, is a Santa Clara, CA hardware and AI company valued at $4.85 trillion (July 2026), the world's most valuable public company (NASDAQ: NVDA). It builds Blackwell and Rubin GPU platforms, CUDA, and the open Nemotron model family, reporting $215.9B in FY2026 revenue, mostly from data center GPUs.
Founded: 1993 · HQ: Santa Clara, CA, USA · Team: 40000-45000 · CEO: Jensen Huang · Funding: Public company (NASDAQ: NVDA) since January 1999; pre-IPO raised $20M from Sequoia Capital and Sutter Hill Ventures. No private funding rounds since IPO. · Valuation: $4.85T market cap (NASDAQ: NVDA, July 1, 2026)
About NVIDIA
NVIDIA was founded on April 5, 1993 by Jensen Huang, Chris Malachowsky, and Curtis Priem, three engineers who had previously worked at LSI Logic, Sun Microsystems, and AMD respectively. The company started with $40,000 in capital and a $20 million venture round from Sequoia Capital and Sutter Hill Ventures, and was incorporated in Delaware with headquarters in Santa Clara, California, where it remains today. The founding thesis was that graphics processing would become a distinct, programmable computing category separate from the CPU, a bet that took nearly two decades to fully pay off in AI. NVIDIA's core product lines now span six domains: the GPU and networking hardware business (Blackwell architecture shipping since 2024, the Rubin platform now in full production as of mid-2026 combining six chips including the Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch), the CUDA software platform, Nemotron open AI models for reasoning and agentic workloads, Omniverse and Cosmos for robotics and physical-world simulation, GR00T for humanoid robot foundation models, Alpamayo for autonomous driving, and Clara for healthcare AI. The Rubin platform claims up to a 10x reduction in inference token cost and a 4x reduction in GPU count needed to train MoE models versus Blackwell. In the last six months, NVIDIA debuted the Nemotron 3 family (Nano, Super, Ultra) of open models on December 14-15, 2025, led by Nemotron 3 Nano 30B-A3B, a hybrid Mamba-2/Transformer Mixture-of-Experts model with 31.6B total and 3.6B active parameters. It launched Rubin at CES in January 2026 with partner availability slated for H2 2026, unveiled a Windows laptop AI chip on June 1, 2026, and expanded its Cosmos world-model line with Cosmos Reason 2 and Cosmos Transfer 2.5. NVIDIA technology now runs 81% of the TOP500 supercomputer list and 90% of systems newly added to it. NVIDIA went public on NASDAQ (ticker NVDA) on January 22, 1999 at $12 per share, so there have been no further private funding rounds since; it is entirely public-market financed today. Its early venture backers, Sequoia Capital and Sutter Hill Ventures, exited long ago through the IPO and subsequent share appreciation, with Sequoia's early stake returning roughly 100x by the IPO alone. As of July 1, 2026, NVIDIA's market capitalization sits at approximately $4.85 trillion, making it the world's most valuable public company. NVIDIA's revenue comes overwhelmingly from data center hardware and software sold to hyperscalers, sovereign AI programs, and enterprises. Fiscal year 2026 revenue reached $215.9 billion, up 65% year over year, with Data Center segment revenue alone at $193.7 billion for the year and a record $62.3 billion in Q4 FY2026 (up 75% YoY). Beyond hardware margins, NVIDIA increasingly monetizes through NIM microservice subscriptions, DGX Cloud, and a fast-growing equity investment arm that had committed more than $40 billion across AI infrastructure partners (including IREN and Corning) by mid-2026. Jensen Huang has been founder, President, and CEO since 1993 and remains in the role in 2026; Forbes estimated his personal net worth above $200 billion this year, ranking him among the world's wealthiest individuals. Colette Kress, who joined in 2013 from Microsoft and Cisco, serves as CFO. The company employed approximately 42,000 people in fiscal year 2026, up from about 36,000 in fiscal 2025, a 16.7% year-over-year increase, with Engineering and IT among its largest departments. NVIDIA frames its mission as bringing "superhuman capabilities to every human, in every industry," and its vision as enabling "a world where everyone can experience the power of AI." Its five stated core values are innovation, intellectual honesty, speed and agility, excellence and determination, and one team. Unlike frontier AI labs, NVIDIA does not publish a Responsible Scaling Policy in the Anthropic/OpenAI sense; its research publication focus is on hardware-software co-design, Nemotron model technical reports (published on arXiv and via its research.nvidia.com labs), and open dataset releases including roughly 11,000 labeled agent-safety traces released alongside Nemotron 3 Nano. Competitively, NVIDIA's moat is distribution and ecosystem lock-in through CUDA, which no merchant GPU competitor (AMD, Intel) has replicated at scale, plus first-mover advantage in AI training and inference silicon. It faces rising competition from custom hyperscaler silicon (Google TPUs, Amazon Trainium, Microsoft Maia) on inference cost, and from AMD's MI-series on raw FLOPs-per-dollar; its Nemotron open models also compete directly with Qwen3, Llama, and GPT-OSS in the open-weights LLM race. NVIDIA's public AI Trust Center lists ISO 27001, ISO 27017, ISO 27018, ISO 27701, SOC 2, FedRAMP High, FISMA High, TISAX, and CAIQ among its certifications, reflecting the breadth of automotive, government, and cloud customers it serves. As a US-headquartered chip company operating globally, NVIDIA is directly exposed to US export control policy on advanced GPUs to China and other restricted destinations, a recurring swing factor in its data center revenue guidance through 2025-2026. It also falls under the EU AI Act's general-purpose AI provider obligations for its Nemotron model releases. Looking ahead, NVIDIA's near-term priorities are the H2 2026 partner rollout of Rubin-based systems with AWS, Google Cloud, Microsoft, and OCI, continued buildout of its $40B+ AI infrastructure equity portfolio, and closing the gap between Nemotron 3 Super/Ultra releases and the Nano tier already shipped.
Mission
To bring superhuman capabilities to every human, in every industry.
Products
- CUDA (GPU compute platform): https://developer.nvidia.com/cuda-toolkit
- Blackwell / Rubin GPU platforms (AI data center hardware): https://www.nvidia.com/en-us/data-center/
- Nemotron (Open reasoning and agentic AI models): https://developer.nvidia.com/nemotron
- NIM Microservices (Managed inference API): https://www.nvidia.com/en-us/ai/
- Omniverse / Cosmos (Robotics and world-model simulation platform): https://www.nvidia.com/en-us/omniverse/
- DGX Systems (AI supercomputer hardware): https://www.nvidia.com/en-us/data-center/dgx-platform/
Compliance
SOC 2, ISO 27001, ISO 27017, ISO 27018, ISO 27701, FedRAMP High, FISMA High, TISAX, CAIQ
Links
Website · GitHub · Twitter · LinkedIn · Blog · Docs
Frequently Asked Questions
What is NVIDIA and what do they build?
NVIDIA is a Santa Clara, California hardware and AI company founded in 1993, originally built around programmable graphics processors and now the dominant supplier of AI training and inference hardware. Its core product lines include the Blackwell and newly-shipping Rubin GPU platforms, the CUDA software stack that developers use to program those GPUs, NIM microservices for managed model inference, and the open-weight Nemotron 3 family of language models aimed at agentic AI use cases. It also builds Omniverse and Cosmos for robotics and physical-world simulation, GR00T for humanoid robot foundation models, Alpamayo for autonomous driving, and Clara for healthcare AI. Developers reach these products through developer.nvidia.com, Hugging Face for open model weights, and build.nvidia.com for hosted APIs. The Rubin platform, now in full production as of mid-2026, claims up to a 10x cut in inference token cost versus Blackwell. As of July 2026 NVIDIA is the world's most valuable public company, with data center hardware sales as its overwhelming revenue driver.
Who founded NVIDIA and who is the CEO?
NVIDIA was founded on April 5, 1993 by Jensen Huang, Chris Malachowsky, and Curtis Priem. Huang previously worked at LSI Logic and AMD as a chip designer, Malachowsky came from Sun Microsystems, and Priem also came from Sun; the trio believed graphics processing would become its own category of programmable computing distinct from the CPU. Jensen Huang has served as founder, President, and CEO continuously since 1993, making him one of the longest-tenured tech CEOs in the industry. Colette Kress, who joined in 2013 from Microsoft and Cisco, has served as CFO since then. As of 2026, Forbes estimates Huang's personal net worth at more than $200 billion, placing him among the world's wealthiest individuals. There has been no CEO change since founding; NVIDIA has no separate founder trust or dual-class governance structure of note beyond standard public-company board oversight.
How much funding has NVIDIA raised?
NVIDIA raised roughly $20 million in venture capital before its IPO, led by Sequoia Capital along with Sutter Hill Ventures, on top of an initial $40,000 in founder capital in 1993. It went public on NASDAQ under ticker NVDA on January 22, 1999 at $12 per share, and has raised no further private funding rounds since; all growth capital since 1999 has come from public equity markets and retained earnings. As of July 1, 2026, its market capitalization stands at approximately $4.85 trillion, making it the most valuable public company in the world. For fiscal year 2026 (ended January 25, 2026), NVIDIA reported revenue of $215.9 billion, up 65% year over year, with data center revenue alone reaching $193.7 billion for the year. Rather than raising funding, NVIDIA has become an active investor itself, committing more than $40 billion in equity stakes across AI infrastructure partners like IREN and Corning by mid-2026. There is no IPO or acquisition rumor relevant here since NVIDIA has been public for over 25 years.
What products does NVIDIA make?
NVIDIA's flagship hardware line is its GPU platforms: the Blackwell architecture shipping since 2024, and the newly production-ready Rubin platform (early 2026), a six-chip system combining the Vera CPU, Rubin GPU, NVLink 6 Switch, ConnectX-9 SuperNIC, BlueField-4 DPU, and Spectrum-6 Ethernet Switch, aimed at hyperscalers and enterprise data centers. On the software side, CUDA remains the dominant GPU programming platform with no widely-adopted equivalent from competitors. NVIDIA also publishes the open-weight Nemotron 3 model family (Nano released December 2025, with Super and Ultra following in H1 2026), a hybrid Mamba-Transformer Mixture-of-Experts architecture aimed at cost-efficient agentic AI, available free on Hugging Face under the NVIDIA Open Model License and via paid managed APIs like build.nvidia.com, OpenRouter, and DeepInfra at roughly $0.05 per 1M input tokens and $0.20 per 1M output tokens. Additional product lines include DGX systems (turnkey AI supercomputers), Omniverse and Cosmos for robotics/world-model simulation, GR00T for humanoid robots, and Clara for healthcare AI. Rubin-based systems are set for partner availability from AWS, Google Cloud, Microsoft, and Oracle Cloud in the second half of 2026.
Where is NVIDIA headquartered and how big is the team?
NVIDIA is headquartered in Santa Clara, California, where it has been based since its 1993 founding. The company employed approximately 42,000 people in fiscal year 2026, up from about 36,000 in fiscal year 2025, a 16.7% year-over-year increase, and up further from roughly 29,600 in fiscal 2024. Its largest internal departments include Information Technology (around 4,494 employees) and Engineering (around 3,571 employees), reflecting the scale of both its chip design and software organizations. NVIDIA maintains major engineering hubs in Santa Clara, and globally across offices supporting its data center, automotive, robotics, and gaming business units. The company has not announced layoffs during this growth period; headcount has grown consistently alongside its data center revenue expansion. No major country-entry announcements were highlighted in 2025-2026 beyond continued expansion of European supercomputer partnerships spanning more than 35 systems across the continent.
What is NVIDIA's mission or research focus?
NVIDIA states its mission as bringing "superhuman capabilities to every human, in every industry," paired with a vision of enabling "a world where everyone can experience the power of AI." Its five stated core values are innovation, intellectual honesty, speed and agility, excellence and determination, and one team. Unlike frontier AI labs such as Anthropic or OpenAI, NVIDIA does not publish a formal Responsible Scaling Policy; its safety-relevant research output instead centers on releasing open datasets, including roughly 11,000 labeled agent-safety traces published alongside Nemotron 3 Nano to help developers evaluate tool-using agent behavior. Its active research areas, documented via research.nvidia.com and arXiv technical reports, include hybrid Mamba-Transformer MoE architectures for efficient inference, long-context retrieval (RULER benchmark work), robotics foundation models (GR00T, Cosmos), and autonomous driving perception (Alpamayo). NVIDIA CEO Jensen Huang has framed 2026 as the year "agentic AI has arrived, doing productive work," positioning the company's research agenda around making agentic AI computationally affordable at scale.
Is NVIDIA compliant with SOC 2, GDPR, HIPAA?
NVIDIA's AI Trust Center lists SOC 2 (SSAE 18-based reporting), ISO 27001, ISO 27017 (cloud security controls), ISO 27018 (PII protection in public cloud), and ISO 27701 (privacy information management) among its certifications, alongside FedRAMP High and FISMA High for US government cloud workloads, TISAX for automotive customers, and CAIQ for cloud security self-assessment. HIPAA-specific eligibility was not explicitly confirmed in NVIDIA's public trust center documentation reviewed for this profile, so it is not listed among confirmed certifications here. NVIDIA's privacy policy is published at nvidia.com/en-us/about-nvidia/privacy-policy, and its terms are covered under its general legal information page. As a hardware and platform company rather than a data processor for most of its business, NVIDIA's GDPR obligations center on its own customer and developer account data rather than customer inference data, which for Nemotron models is governed by whichever API provider (build.nvidia.com, OpenRouter, DeepInfra, or self-hosted deployment) a customer chooses. Nemotron model releases fall under the EU AI Act's general-purpose AI model provider obligations.
Who are NVIDIA's main competitors?
NVIDIA's most direct hardware competitor is AMD, whose Instinct MI-series GPUs compete on price-per-FLOP for AI training and inference, though NVIDIA maintains a substantial CUDA software ecosystem lock-in that AMD's ROCm stack has not matched in developer adoption. A growing threat comes from hyperscaler-built custom silicon: Google's TPUs, Amazon's Trainium, and Microsoft's Maia chips let those companies run inference workloads on their own hardware, reducing their NVIDIA GPU purchases for internal use even as they remain customers for external cloud offerings. In the open-weight AI model space, NVIDIA's Nemotron 3 family competes directly with Alibaba's Qwen3-30B-A3B (which edges out Nemotron 3 Nano on MMLU-Pro, 80.9% versus 78.3%, though Nemotron claims 3.3x higher throughput), Meta's Llama models, and OpenAI's GPT-OSS line. NVIDIA's clearest differentiator remains full-stack integration: owning silicon, interconnect (NVLink), and software (CUDA, NIM) in one package that competitors assemble piecemeal. A category where NVIDIA currently has no strong proprietary answer is closed frontier chat models comparable to GPT-5 or Claude; Nemotron targets efficient open agentic use cases rather than a general consumer chatbot.