Nvidia projects $4T data center spending by 2030, eyes $20T valuation

Nvidia, still led by founder Jensen Huang, has made a bold prediction about the future of artificial intelligence infrastructure spending. Over the company's past few quarterly conference calls, Nvidia has told investors it expects that the world's annual data center capital expenditures could grow to up to $4 trillion（約640兆円） by 2030. At today's $5 trillion（約800兆円） market cap, that projection implies a path to a $20 trillion（約3200兆円） valuation over the next few years.

The logic underpinning this forecast centers on Nvidia's dominance in GPU (graphics processor) supply. Nvidia makes GPUs and the various products that support their use in data centers, and its processors have become the gold standard for high-performance parallel computing. The company captured the vast majority of the market in the early days of the AI arms race, which makes it incredibly difficult for data center operators to switch away from Nvidia's products now. This competitive advantage will only grow as more data centers are built. To contextualize current spending: the big four AI hyperscalers have estimated that they will spend a total of around $650 billion（約100兆円） on data center capital expenditures in 2026. When including spending by neoclouds, international players in markets such as China, and other rising stars like large language model developers Anthropic and OpenAI, actual total data center spend is estimated at something closer to $800 billion（約130兆円）. Huang's prediction of $4 trillion（約640兆円） by 2030 would therefore represent a fivefold rise from that baseline.

Nvidia's current financial momentum supports confidence in long-term growth. In its latest quarter, revenue grew 85% year over year, and Wall Street analysts expect nearly 100% revenue growth next quarter—all without any chip sales to China due to U.S. export restrictions. However, that situation may be changing. A U.S. official recently stated that 'very few' Nvidia H200 chips have been shipped to China. While that comment might initially sound negative, it can be read as a strong sign that Nvidia is returning to the Chinese market, as the U.S. government has begun relaxing restrictions somewhat. Even with those relaxations, the Chinese government has been putting roadblocks in the way of Nvidia's return, but if those barriers come down, Nvidia would gain access to a major market currently excluded from its revenue figures—a growth catalyst not yet reflected in the company's guidance.

Nvidia's $4 trillion（約640兆円） projection for global data center spending by 2030 rests on the company's entrenched position in the AI hardware market. The article notes that Nvidia makes GPUs and supporting products for data centers, and its processors have become the standard for high-performance computing. The company captured the vast majority of the market in the early days of the AI arms race, creating a lock-in effect: once data center operators choose Nvidia, switching becomes costly and difficult, an advantage that compounds as more data centers are built.

The math supporting the $20 trillion（約3200兆円） valuation target is straightforward: if global data center capex grows fivefold (from an estimated $800 billion（約130兆円） in 2026 to $4 trillion（約640兆円） by 2030) and Nvidia maintains its current market share, its revenue and profits would rise proportionally. The article notes that Nvidia would only need to quadruple from its current $5 trillion（約800兆円） market cap to hit $20 trillion（約3200兆円）, meaning the company could actually lose market share and still reach that target—a buffer that suggests the projection is not dependent on Nvidia's dominance remaining unchallenged.

A secondary catalyst lies in China. U.S. export restrictions have largely blocked Nvidia's access to the Chinese market, but recent signals suggest those barriers may be weakening. If restrictions ease and Nvidia regains meaningful sales in China, that growth is currently absent from the company's official guidance, potentially enabling even faster expansion than projected.