Analysis highlights Nvidia's near-monopoly position in AI accelerators and the risks of market concentration.
Nvidia has announced Cosmos, which the company describes as "the world's first world foundation model" — AI models designed to understand how humans mentally model the world to predict and generate physics-aware videos for robotics and autonomous vehicles. During the keynote, Jensen Huang, founder and CEO of Nvidia, emphasized the significance of the moment: "The ChatGPT moment for robotics is coming. Like large language models, world foundation models are fundamental to advancing robot and AV development, yet not all developers have the expertise and resources to train their own. We created Cosmos to democratize physical AI and put general robotics in reach of every developer." Nvidia has made Cosmos available in three sizes — Nano, Super, and Ultra — ranging from 4 billion to 14 billion parameters. Robotics and automotive companies including 1X, Agile Robots, Agility, Figure AI, Foretellix, Fourier, Neura Robotics, Skild AI, Virtual Incision, Uber, and XPENG are already engaged with the technology.
Nvidia's strategic positioning extends far beyond Cosmos. The company has achieved dominance by controlling the chip infrastructure upon which most AI services run. Amazon Web Services, Microsoft Azure, and Google Cloud all rely heavily on Nvidia GPUs for their AI infrastructure and are among the largest buyers of Nvidia's high-end chips like the A100 and H100. The H100, during periods of shortage, commanded significant premiums. Nvidia has now introduced the GB200 Grace Blackwell superchip at $30,000 (around ₹25,00,600), which Huang described as an "engine to power this new industrial revolution."
Despite competitive efforts, few alternatives have gained traction. Microsoft's Maia and Cobalt chips have generated little public discussion since their summer preview, and analysts estimate Microsoft purchased as many as 485,000 of Nvidia's Hopper chips through 2024. Broadcom, which manufactures Tensor Processing Units for Google, represents potential competition, with Omdia Research estimating Google's TPUs could account for between $6 billion and $9 billion in AI semiconductor revenue for Broadcom in 2025. Nvidia's dominance has drawn regulatory scrutiny: the U.S. Department of Justice opened an antitrust probe into the chipmaker's more than 90% share of the datacenter GPU market, while in December, Chinese regulators launched an investigation into Nvidia's potential monopolistic practices. Nvidia reportedly held more than 90% share of the AI chip market in China before export restrictions took effect, with Huawei attempting to fill the resulting gap.
Alongside Cosmos, Nvidia has unveiled Project DIGITS, a $3,000 personal supercomputer with 1,000x the power of an average laptop. Powered by the Grace Blackwell Superchip, it will feature 128GB unified memory, 4TB SSD storage, and the ability to run models with up to 200 billion parameters when it becomes available in summer. Nvidia's expansion into personal computing beyond graphics chips—combined with the scale of these announcements—poses considerable challenges to competitors including Qualcomm, whose Snapdragon X chips represent a recent push, AMD's attempts at resurgence, and Intel, which faces increasingly difficult circumstances in the sector.