Nvidia’s share price has fallen roughly 15% from its May peak, even as the company’s projected revenue continues to grow. Bloomberg reported the details: the upshot is that investors are now paying less per dollar of Nvidia’s expected earnings than they do for the typical S&P company.
Where the money is going: memory makers gain
Although capital continues to flow into AI infrastructure, much of it is now moving toward memory companies. Micron — one of the world’s largest DRAM manufacturers — has nearly tripled in value over the same period, signaling that memory capacity has become the bottleneck for modern data centers and the new hot trade for investors.
The basic story is straightforward: the GPU shortage that looked alarming last year has eased somewhat, while data centers urgently need more memory. DRAM, the standard memory chip found in computers and servers, consists of high-bandwidth components optimized to move data in and out of processors quickly — a service that has become far more valuable as demand surged.
Prices diverge: concrete numbers and trends
Spot prices for DRAM — what buyers pay on the open market rather than under long-term contracts — have risen dramatically; the article states DRAM prices increased tenfold over the past year. By contrast, spot prices for time on an Nvidia H100 GPU have fallen after peaking: Ornn’s data show an hourly spot peak of about $3.20 around May, followed by a steady decline.
This divergence helps explain Nvidia’s weaker relative valuation: Nvidia’s corporate value is tied closely to the price of compute, which has been falling, while memory-linked companies benefit from rising DRAM prices.
Why this happened: supply and demand dynamics
Wayne Nelms, co-founder and CTO of Ornn, frames the imbalance as a simple supply-and-demand issue. Large cloud providers — Google, Amazon, and Microsoft — and companies like OpenAI have developed custom processors to reduce dependence on Nvidia; even if those chips don’t match Nvidia’s newest models, they are sufficient to push down compute prices.
Nelms told the author: “More GPU and accelerator players are entering the market. Everyone wants to make their own silicon, but no one is making their own DRAM. Until there’s a major technological breakthrough on HBM [high-bandwidth memory], a shift in supply and demand, or someone new [enters the market in memory], I think things will more or less persist as we see today.”
Implications and summary
The situation is frustrating for Nvidia and is partly a consequence of its own success. By proving the value of compute for AI, Nvidia became the center of a market many players now want a piece of — while simpler, less glamorous companies that provide essential components, like memory manufacturers, see outsized gains. For the near term, investors and data-center builders are focused on the memory bottleneck, strengthening DRAM makers even as GPU spot prices and Nvidia’s market premium decline.
