Jensen Huang: The Biggest Constraint on AI Right Now Is Memory

High‑bandwidth memory is produced by only a few major companies, including SK Hynix, Samsung, and Micron. The rapid expansion of AI data centers has pushed demand for these specialized memory chips sharply higher.

When supply falls short, several ripple effects can occur:

• Slower AI infrastructure rollout. Cloud providers and enterprises may have to wait longer for full GPU systems.
• Higher system costs. Scarce memory components can drive up prices across the AI hardware stack.
• Supplier power shifts. Memory manufacturers gain leverage in negotiations with chip designers and system builders.

In other words, the next phase of the AI race may hinge less on designing faster GPUs and more on scaling the manufacturing capacity for advanced memory.

China and the Uncertain Future of H200 Sales

Geopolitics adds another layer of complexity. Nvidia’s H200 AI chips are central to its next wave of data‑center deployments, but their availability in China remains uncertain.

During a recent trip to Beijing alongside U.S. President Donald Trump, Huang suggested he believes the Chinese market will eventually reopen more broadly to American chipmakers. However, he also acknowledged that regulatory approvals are still evolving.

The situation is complicated:

• The United States has granted licenses allowing certain Chinese firms to buy H200 chips.
• Chinese regulatory approval has not fully materialized.
• As of mid‑2026, no H200 shipments to China had actually occurred despite those approvals.

This means that what appears on paper as potential demand does not necessarily translate into immediate revenue for Nvidia.

Why Investors Are Watching This Before Nvidia’s Earnings

These supply‑chain and geopolitical issues matter especially now because Nvidia is about to report results.

The company’s Q1 fiscal 2027 earnings call is scheduled for May 20, 2026, covering the quarter that ended April 26.

Investors will likely focus on several key questions:

• How much HBM supply Nvidia has secured for upcoming GPU shipments
• Whether H200 and Blackwell systems can ship at scale
• Whether China sales could restart and contribute to revenue

Demand for AI infrastructure remains extremely strong worldwide. But Huang’s comments at Dell World highlight the central tension in the market: even when demand is overwhelming, the physical supply chain—especially advanced memory—can determine how fast the AI industry actually grows.

For Nvidia and the broader ecosystem, the next stage of the AI boom may be decided not just by algorithms or GPUs, but by how quickly the world can manufacture the memory needed to run them.