Nvidia says its forecast for $200 billion CPU market includes China

Nvidia CEO Jensen Huang said the company’s new Vera CPUs open a $200 billion market, and he indicated China is included in that opportunity. He also reiterated that H200 chips have been licensed for shipment to China, though no deliveries have been made and Chinese approval remains absent. The article suggests continued long-term demand in AI semiconductors despite U.S.-China technology tensions.

Analysis

The market is underappreciating how quickly AI demand is broadening from model-training hardware into the control layer of inference and orchestration. That matters because CPU-led workloads tend to be more diversified across enterprise IT, sovereign cloud, and edge deployments, which reduces customer concentration risk and makes the AI capex cycle less exposed to a single bottleneck in GPU supply. For NVDA, this is not just incremental TAM expansion; it is a defense against a future where buyers optimize around cheaper, more available compute architectures and pressure margins on flagship accelerators.

The second-order winner is TSM, because any real mix shift toward higher-volume CPUs and heterogeneous AI systems increases the number of leading-edge wafers, advanced packaging steps, and interconnect complexity per deployed system. Even if export restrictions cap China near-term revenue, the geopolitical fragmentation itself supports duplicate supply chains and more localized inventory buffers, which is structurally positive for TSM capacity utilization over the next 12-24 months. The loser set is less about named competitors and more about incumbent x86/merchant silicon vendors that face a faster erosion of design wins as AI customers consolidate around vertically integrated platforms.

The key risk is timing: China upside is a 6-18 month catalyst, not a next-quarter story, and licensing approvals can still be reversed or slowed by either government. Near term, the stock may already reflect the idea that AI demand is durable, so the bigger risk is a sentiment air pocket if deliveries remain symbolic rather than monetized. The contrarian read is that the market is still too focused on the geopolitical headline and not enough on product-mix diversification, which could make NVDA’s long-run earnings power less cyclical than the bear case assumes.

Key Decisions for Investors

• Long NVDA / long TSM as a 3-6 month pair: express the view that AI compute diversification expands node demand, with TSM offering lower policy risk and NVDA retaining optionality on CPU attach. Use a tighter stop on NVDA if China headlines turn restrictive again.
• Buy NVDA 6-12 month call spreads instead of outright equity: upside is tied to a re-rating of long-duration AI demand, while the spread limits premium burn if China remains largely non-monetized.
• Add TSM on pullbacks over the next 2-4 weeks: the stock should benefit from any confirmation that heterogeneous AI architecture increases advanced foundry and packaging intensity. Risk/reward improves on weakness tied to macro or FX noise rather than thesis deterioration.
• Hedge NVDA long exposure with a short basket of legacy x86/AI-adjacent hardware names over 3-6 months: the thesis is that platform consolidation and custom silicon adoption compress the value of standalone compute vendors.
• If H200 shipment progress becomes visible, trade a short-dated NVDA upside call structure for a 1-2 month catalyst window; if deliveries remain stalled, take profits quickly because the market will likely fade the China narrative.