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]][^c7], 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[^c2], reaching approximately $4.7 trillion by February 2026[^c18] and $5.7 trillion by May 2026 — a valuation exceeding Germany's projected GDP[^c22], holding an estimated 92% of the data center GPU market[^c6]. As of January 2026, NVIDIA employed 42,000 people worldwide[^c17].
The company's core businesses include data center computing, gaming graphics, professional visualization, automotive platforms, and robotics. The CUDA software platform, introduced in 2006 and celebrating its 20th anniversary in 2026, created an ecosystem that "creates a formidable 'lock-in' effect"[^c3] with over 6 million developers worldwide[^c15]. The [[ai-enterprise]] software platform reached a $1 billion annualized revenue run rate in 2023[^c8]. For fiscal year 2025, NVIDIA reported record revenue of $130.5 billion[^c4], followed by fiscal year 2026 revenue of $215.9 billion, up 65%[^c9]. In the first quarter of fiscal 2027, the company posted record revenue of $81.6 billion, up 85% year-over-year[^c23], with Data Center revenue reaching $75.2 billion. Concurrently, NVIDIA restructured its financial reporting into two segments — Data Center and Edge Computing — consolidating its Gaming, Professional Visualization, and automotive revenue lines into a $6.4 billion Edge Computing segment[^c27][^c34]. The board authorized an $80 billion share buyback program and raised the quarterly dividend from $0.01 to $0.25 per share[^c33].
NVIDIA has expanded beyond chip design into data center networking, autonomous vehicle platforms, industrial digital twins, humanoid robotics, space computing, quantum computing, and enterprise AI agent infrastructure. At GTC 2026, the company announced a combined Blackwell and Vera Rubin purchase order pipeline targeting $1 trillion through 2027[^c10], previewed the Feynman architecture for 2028[^c13], unveiled the Ising family of open AI models for quantum computing[^c19], launched the NemoClaw platform for enterprise AI agents[^c20], and entered space computing with orbital data center platforms[^c21]. In optical networking, NVIDIA partnered with Corning in a multiyear agreement to expand US manufacturing of optical connectivity solutions[^c11], including a $500 million investment[^c14]. In December 2025, NVIDIA executed a $20 billion licensing and talent acquisition deal with [[groq]], acquiring its LPU inference technology and engineering team[^c16]. On the geopolitical front, the US authorized H200 chip sales to approximately ten Chinese companies[^c12], while CEO Jensen Huang joined President Trump's delegation to China for negotiations over AI chip exports.
In May 2026, NVIDIA declared the arrival of "the era of useful AI," with Huang stating that "What took months now takes weeks. What took weeks now takes days"[^c26] as agentic AI workloads drove accelerating demand. At COMPUTEX 2026 in June, Huang confirmed that the [[blackwell-rubin|Vera Rubin]] platform was in full production with a supply chain twice as large as Grace Blackwell, reducing rack assembly time from two hours to five minutes[^c31]. NVIDIA launched the Vera CPU, a processor purpose-built for AI agents. Independent Phoronix benchmarks showed the Vera CPU delivering a 1.5x overall performance advantage over a 128-core x86 processor and compiling a Linux kernel in 20 seconds, described as "the most formidable competition to Intel and AMD x86_64 processors ever realized"[^c56][^c57][^c58]. NVIDIA also entered the consumer PC market with the RTX Spark superchip (developed with MediaTek) for Windows-on-Arm laptops[^c28]. Huang described the RTX Spark as "as big of a deal as the reinvention of the phone into what we now know as the smartphone"[^c53]. The company also announced Cosmos 3, an open world foundation model for physical AI, and Nebius launched the Physical AI Living Lab, a six-month program providing European robotics startups with access to NVIDIA's physical AI development tools including Cosmos, Isaac Sim, and Isaac Lab[^c35].
At GTC Taipei at COMPUTEX, NVIDIA unveiled the [[alpamayo]]-powered reasoning autonomous vehicle system, including Alpamayo 2 Super, a 32-billion-parameter model with 360-degree surround-view perception and meta-action outputs[^c52]. Huang announced the DRIVE Hyperion platform as a unified Level 4-ready foundation for robotaxi fleets, stating that "autonomous mobility is entering its industrial scaling moment." NVIDIA also announced the Factory Operations Blueprint (FOX), a reference design for autonomous factory manager agents built with NemoClaw, with Foxconn projecting an 80% improvement in root cause analysis time[^c39], and began shipping Spectrum-X co-packaged optics (CPO) switches delivering up to 400 Tb/s throughput[^c54].
On June 15, 2026, NVIDIA returned to the corporate bond market for the first time in five years, pricing a $25 billion senior notes offering that drew $85 billion in investor demand[^c36]. The following day, NVIDIA achieved a clean sweep in MLPerf Training v6.0, setting time-to-train records across all workloads including training the 671-billion-parameter DeepSeek-V3 model in 2.02 minutes on 8,192 Blackwell Ultra GPUs[^c37]. NVIDIA also released XR AI in public beta, an open-source library for building AI agents that operate on AR glasses and XR headsets[^c38].
On June 4, 2026, Senator Elizabeth Warren invited Jensen Huang to testify before the Senate Banking Committee on June 11 regarding NVIDIA's compliance with US export control laws and its business in China[^c29]. The following day, a broad semiconductor selloff erased approximately $1.3 trillion in market value, with NVIDIA losing more than $300 billion in market capitalization, triggered by Broadcom's weak earnings and amplified by strong US jobs data and a SemiAnalysis report on Vera Rubin memory configurations[^c30]. Later that week, NVIDIA director Mark Stevens sold 500,000 shares worth approximately $110 million, a 7.25% reduction in his position[^c32].
At ISC High Performance 2026 in Hamburg, the latest TOP500 list revealed that NVIDIA technology powers over 400 (81%) of the world's 500 fastest supercomputers, with its GPUs used in 238 systems and networking in 376 systems[^c55]. NVIDIA also swept the top eight positions on the Green500 energy-efficiency list. The company announced the Vera Rubin platform as a scientific supercomputer delivering more than 7 exaflops of AI for science and 5 petaflops of native FP64 performance in a single rack, with supercomputing centers including LRZ (Blue Lion), NERSC (Doudna), and Los Alamos National Laboratory (Mission, Vision, Veritas) adopting the platform[^c44]. Earlier, the Grace Hopper-powered JUPITER supercomputer at Forschungszentrum Jülich — Europe's first exascale system — demonstrated four scientific breakthroughs including a brain foundation model (CytoNet), a Gordon Bell Prize-winning climate simulation, a world-record 50-qubit quantum computer simulation, and a 6G AI collaboration with Ericsson[^c41]. Separately, NVIDIA announced 35 new AI HPC supercomputers in development across 23 European countries, equipping over 3 million researchers — Europe's largest one-year supercomputer expansion[^c43].
On June 22, NVIDIA announced that the Rubin generation is the world's first 100% liquid-cooled AI infrastructure, with coolant operating at up to 45 °C in a closed-loop system that eliminates new water consumption entirely[^c42]. A 50-megawatt hyperscale facility can save over $4 million annually in cooling costs. The company also unveiled Halos for Robotics, extending its autonomous vehicle safety system to cover robotics and physical AI, with Agility Robotics as the first adopter[^c45]. The Blackwell-Next codename was discovered in Linux 7.2 kernel patches, hinting at NVIDIA's next-generation GPU architecture work beyond the public roadmap[^c46].
Following COMPUTEX, Huang visited South Korea, where he announced a multiyear, multiplatform partnership expansion with SK Group[^c48]. SK hynix signed a separate multiyear technology partnership with NVIDIA to co-develop memory for Vera Rubin and expand into personal AI and physical AI markets[^c49]. In China, NVIDIA began pitching its Vera CPU as a strategic product that could bypass GPU export restrictions, with availability as soon as August 2026[^c50], while Marcio Aguiar, NVIDIA's Latin America chief, stated at the Rio Web Summit that NVIDIA's China AI chip market share had fallen from approximately 95% in 2022 to effectively zero under tightening controls[^c51]. Separately, OpenAI was reported to be in talks to lease a 10-gigawatt data center campus in Ohio, with NVIDIA expected to both supply hardware and provide a financial guarantee for the lease[^c47].
On June 23, semiconductor stocks experienced a second severe selloff triggered by Korean media reports that NVIDIA's Rubin platform production forecasts had been revised downward, leading SK Hynix to slow HBM4 expansion[^c63]. The Philadelphia Semiconductor Index dropped 6.3%, and NVIDIA fell approximately 4%, briefly pushing its market capitalization below $5 trillion[^c62]. The global selloff erased South Korea's KOSPI index by nearly 10%, triggering circuit breakers, while memory chipmakers Micron and SanDisk each fell more than 13%. At the company's Annual Shareholder Meeting on June 24, Jensen Huang declared that "useful AI has arrived and it is profitable"[^c59], reported full-year revenue of $216 billion (up 65%) with $103 billion in operating cash flow and $410 billion returned to shareholders[^c61], confirmed that the Vera Rubin architecture was in full production, and committed to returning 50% or more of free cash flow to shareholders annually[^c60].
On June 25, OpenAI and Broadcom released the Jalapeño inference chip while Qualcomm launched the Dragonfly data center platform[^c65], and the NVDA/SOXX ratio was reported to have declined 58% from its July 2025 peak, reflecting a broadening of the AI trade beyond NVIDIA. On June 28, Firmus Technologies announced a partnership to deploy 170,000 NVIDIA GPUs in Indonesia[^c66], expanding NVIDIA's Southeast Asian infrastructure footprint.
NVIDIA is led by founder and CEO Jensen Huang, who has served as president and chief executive since its founding. Huang is a member of the National Academy of Engineering and has received the Robert N. Noyce Award, the Semiconductor Industry Association's highest honor[^c5].