NVIDIA
NVIDIA is an American technology company specializing in graphics processing units (GPUs), accelerated computing, and artificial intelligence. Founded in 1993 by [[jensen-huang]], [[chris-malachowsky]], and [[curtis-priem]], the company invented the GPU in 1999, which "sparked the growth of the PC gaming market, redefined computer graphics, and ignited the era of modern AI"[^c1]. By 2025, NVIDIA had grown into the world's most valuable semiconductor company, surpassing a $4 trillion market capitalization, and holds an estimated 92% of the data center GPU market[^c2][^c5]. As of January 2026, NVIDIA employed 42,000 people worldwide.
In the first quarter of fiscal 2027 (ended April 26, 2026), NVIDIA posted record revenue of $81.6 billion, up 85% year-over-year, with Data Center revenue reaching $75.2 billion[^c6]. The company restructured its financial reporting into two segments — Data Center (with Hyperscale and ACIE sub-segments) and Edge Computing — reflecting its transformation from a graphics company to an AI computing platform. Driven by AI demand, NVIDIA returned approximately $20 billion to shareholders in the quarter and announced an $80 billion share repurchase authorization.
In July 2026, NVIDIA's Vera Rubin platform received its first independently measured customer validation, with CoreWeave recording a 10x performance-per-watt improvement over Blackwell on DeepSeek-R1 across the full Pareto frontier[^c11]. Vera Rubin NVL72 entered full production in June 2026, with racks running at CoreWeave, Google Cloud, Microsoft Azure, and Oracle Cloud Infrastructure across more than 350 factory sites in 30 countries[^c16]. Nebius Group brought up and began validating its first full Vera Rubin NVL72 rack at its Finland data center in late July, two days after NVIDIA disclosed a 9.3% passive stake in the company[^c17]. The Vera CPU posted a 1.8x performance advantage over a 128-core AMD EPYC Turin on SPEC 2026 agentic workloads, and reinforcement learning sandbox benchmarks from Perplexity and Prime Intellect confirmed 1.5x to 1.9x improvements over x86 alternatives in tool execution and sandbox startup time. The roadmap also hit a packaging constraint in July 2026, when the four-die Rubin Ultra GPU was cancelled and the revised 2027 GPU was reported to deliver roughly half the announced compute and memory bandwidth[^c24].
NVIDIA announced two landmark sovereign AI infrastructure partnerships in Korea during President Jae Myung Lee's visit to San Francisco. With NAVER and Brookfield, the company will expand Korea's national AI factory from 55 MW to 200 MW, with NVIDIA investing $1 billion and Brookfield funding up to $9 billion[^c9]. Separately, SK Group and NVIDIA announced a $500-billion-plus comprehensive partnership spanning a 2-gigawatt SK Telecom DSX AI factory using Vera Rubin and long-term HBM4 memory co-development with SK hynix[^c10]. These announcements followed the July 16 launch of the world's first national AI infrastructure for physical AI in Japan, a Noetra consortium project with 27,500 Rubin GPUs at 140 MW capacity backed by up to ¥1 trillion in METI funding[^c8].
In telecommunications, Nokia launched the industry's first commercial AI-RAN platform on July 15, built on its anyRAN software and NVIDIA's Aerial AI-RAN platform, targeting more than 100% spectral efficiency gains by 2028 through GPU-powered baseband processing[^c12]. The partnership is backed by NVIDIA's USD 1.0 billion equity investment in Nokia made in October 2025[^c22]. NVIDIA's broader national AI strategy framework, outlined in July, positions AI factories as "the bedrock of modern economies" and identifies five ingredients of sovereign AI capability[^c13].
NVIDIA's physical AI portfolio expanded through 2026. At GTC Taipei, the company launched Cosmos 3, an open world foundation model built on a mixture-of-transformers architecture for physical AI reasoning, world simulation, and action generation[^c25]. In the autonomous vehicle segment, NVIDIA and Uber announced a partnership to scale a global autonomous fleet toward 100,000 vehicles starting in 2027, supported by a joint AI data factory built on Cosmos and the DRIVE AGX Hyperion 10 level 4-ready reference architecture[^c26]. In networking, NVIDIA delivered its first co-packaged-optics Quantum-X InfiniBand Photonics switch to AI cloud provider Lambda in June 2026, moving silicon-photonics CPO technology from engineering samples to production deployment[^c27].
NVIDIA's inference software stack was shown to compound optimizations for up to 20x throughput improvement, cutting token costs on Blackwell by up to 5x on the [[deepseek-v4|DeepSeek V4]] model within a month. The open-source Dynamo framework for distributed inference serving demonstrated up to 50x MoE throughput gains on GB300 NVL72 systems. At ICML 2026, NVIDIA had 74 accepted papers, with approximately 2,000 papers citing NVIDIA GPUs[^c14]. Reports indicated that NVIDIA planned no new gaming GPU releases in 2026 for the first time in nearly 30 years, as GDDR7 memory shortages driven by AI accelerator demand forced prioritization[^c15].
NVIDIA's financial relationship with OpenAI deepened beyond equity in the summer of 2026. By late July, the company was in talks to provide roughly $250 billion in financial guarantees to help OpenAI lease a 10-gigawatt data center in southern Ohio being developed by SB Energy, a SoftBank-controlled subsidiary[^c20]. The discussions would mark NVIDIA's first role as a de facto infrastructure financial guarantor at that scale, using its balance sheet to backstop OpenAI's lease and the project's financing[^c18]. The period also brought renewed skepticism about AI infrastructure spending: investor Michael Burry doubled his short positions in Nvidia and Micron in July, arguing that much of the demand for AI chips was not being driven by end customers[^c21].
NVIDIA's U.S. manufacturing expansion continued with Blackwell wafers in volume production at TSMC's Phoenix facility and new AI supercomputer plants under construction with Foxconn in Houston and Wistron in Fort Worth. In July 2026, Wistron inaugurated its D1 AI Smart Factory in Fort Worth, the first U.S. facility to mass-produce NVIDIA GB300 Grace Blackwell Ultra superchips[^c19]. Public First estimated that NVIDIA-driven AI demand contributed $485 billion to U.S. GDP in 2026 and supported over 100,000 jobs[^c7]. On June 22, NVIDIA unveiled Halos for Robotics, extending its autonomous vehicle safety architecture to industrial and humanoid robots with Agility Robotics as the first adopter. NVIDIA also appointed former Goldman Sachs Vice Chairman Suzanne Nora Johnson to its board of directors effective July 13, 2026. In June, SpaceX revealed its AI1 orbital data-center satellite and said the initial orbital data centers will use Nvidia hardware, extending NVIDIA's reach into space-based compute[^c23].
NVIDIA is led by founder and CEO Jensen Huang, who has served as president and chief executive since the company's founding in 1993.