Nvidia eyes $1 trillion in orders as AI chip demand outpaces supply

Nvidia's chip supply still cannot keep pace with AI demand two years into the boom, CEO Jensen Huang said.

Nvidia expects $1 trillion in orders for its Blackwell and Vera Rubin platforms through 2027, as demand for AI chips continues to outstrip the company's ability to produce them. The Santa Clara, California-based company posted Q1 FY2027 revenue of $81.61 billion, up 85% from a year earlier, with data center networking alone growing 199% to $14.8 billion.

"We have supply for very, very robust growth, but we're still supply constrained," Huang said at Computex, the international technology trade show in Taiwan. The CEO's comments came as Nvidia prepares to launch its Vera Rubin platform, designed for the emerging agentic AI era where AI systems can act autonomously on behalf of users.

Nvidia shares trade at 31.3 times trailing earnings, a valuation that some analysts consider inexpensive relative to the company's growth trajectory. The company's market capitalization stands near $5 trillion, making it the world's most valuable publicly traded company. Huang told investors they "should be very excited to buy stock at a cheaper price" following a market dip tied to rate-hike concerns after the latest jobs data.

Vera Rubin and the Agentic AI Opportunity

The Vera Rubin platform represents Nvidia's next major product cycle, targeting what Huang described as the agentic AI inflection point. Unlike traditional AI models that respond to prompts, agentic AI systems can plan, execute tasks, and make decisions independently, requiring substantially more computing power. Nvidia's new Vera CPU, developed specifically for these workloads, could prove even more popular than its GPUs, according to Huang.

The company expects $20 billion in stand-alone CPU revenue through the end of the year, tapping into what Huang estimates is a $200 billion addressable market in the CPU space driven by agentic AI. That positions Nvidia to challenge Intel and Advanced Micro Devices in a market those two companies have dominated for decades.

Supply Constraints Persist Despite Capacity Expansion

Nvidia has invested heavily in expanding manufacturing capacity, but demand continues to outrun supply. The company's chips are manufactured primarily by Taiwan Semiconductor Manufacturing Co., which trades at 37.1 times trailing earnings and faces its own capacity limitations. TSMC must roughly double its capacity to meet AI demand, Huang has said previously.

The supply-demand imbalance extends beyond Nvidia's core GPU business. Marvell Technology, which Huang publicly endorsed as a candidate to join the trillion-dollar club, has reported exceptional AI-related bookings. Marvell posted Q1 FY2027 revenue of $2.42 billion, up 28% year over year, with data center revenue accounting for 76% of total sales. Nvidia and Marvell collaborate on NVLink Fusion technology and silicon photonics, which use light rather than electrical signals to transmit data between chips.

Huang has also warned that the U.S. faces structural disadvantages in energy production that could constrain AI infrastructure buildout. "The United States is woefully behind in energy production," he said in a recent interview, noting that data centers require enormous amounts of electricity. The CEO has advocated for broader AI adoption across society, arguing that the technology will create jobs and economic growth rather than destroy them.

Nvidia shares, trading at roughly 31 times forward earnings, reflect a market that has priced in continued dominance but may not fully account for the Vera Rubin cycle or the agentic AI opportunity. If Huang's demand outlook proves accurate, the current multiple could look conservative against the revenue trajectory implied by $1 trillion in platform orders through 2027.

This article is for informational purposes only and does not constitute investment advice.