Nvidia is forecasting a $200 billion central processing unit market, a move that signals the chipmaker's ambition to challenge longstanding players in the CPU space. The forecast includes demand from China, even as existing export restrictions complicate sales to the country. The company's entry could reshape the artificial intelligence landscape, where its GPUs already dominate, but it also forces Nvidia to navigate a thicket of geopolitical tensions.
Why Nvidia Is Pushing Into CPUs
The $200 billion projection covers the total addressable market for CPUs over the coming years. Nvidia, best known for graphics chips that power AI training, has been quietly building CPU designs for data centers. Its own Grace CPU, announced in 2021, already pairs with its GPUs for high-performance computing. By targeting the broader CPU market, Nvidia is betting that the lines between general-purpose processors and specialized accelerators will blur — and that it can own both sides of that equation.
The company's move comes as AI workloads increasingly demand a mix of CPU and GPU muscle. Traditional CPU makers like Intel and AMD have responded by adding AI features to their chips, but Nvidia believes it can offer tighter integration. The forecast suggests Nvidia sees a chance to pull customers away from those incumbents, particularly in data centers where AI training and inference need fast data movement between processors.
China's Place in the Forecast
Nvidia's $200 billion CPU forecast explicitly includes demand from China, despite U.S. export controls that restrict sales of advanced chips to the country. The restrictions, updated in October 2022 and again in 2023, bar Nvidia from selling its top-tier A100 and H100 GPUs to Chinese customers. The company has since created lower-spec versions — the A800 and H800 — that comply with the rules but still face scrutiny.
Including China in the CPU forecast suggests Nvidia expects its CPU products to either fall outside the strictest export controls or that it can sell a version that complies. The company's CPU designs are less specialized for AI than its GPUs, which may give it more room to maneuver. But the geopolitical picture remains uncertain, and any new restrictions from Washington could shrink that part of the forecast overnight.
The Competitive and Geopolitical Hurdles
Nvidia's CPU push puts it directly against Intel, which has dominated the server CPU market for decades, and AMD, which has gained ground with its EPYC line. Both companies have deep relationships with cloud providers and enterprise customers. Nvidia will need to convince those buyers to redesign systems around its chips — a long and expensive process.
Geopolitically, the bet on China adds a layer of risk. The U.S. Commerce Department has shown it is willing to tighten export controls if it sees national security risks. Nvidia has already lost billions of dollars in potential GPU sales to China because of the rules. If the same logic extends to its CPUs, the forecast may prove optimistic.
Still, Nvidia's track record in AI gives it credibility. The same data center operators that buy its GPUs are natural targets for a combined CPU-GPU platform. The company will likely start with its Grace Hopper superchip, which pairs a Grace CPU with an H100 GPU, and then move into standalone CPUs for more general workloads.
For now, the $200 billion number is a target, not a guarantee. Nvidia has to deliver a product that can compete on price, performance, and power efficiency — and it has to do so while the rules of the global chip trade are being rewritten. The next test will come when Nvidia reveals further details about its CPU roadmap, expected later this year at its annual GTC conference.
