Nvidia Stock Slips 15% as GPU Spot Prices Fall and Memory Chips Surge
Original: Nvidia is a victim of the compute marketplace it created
Why This Matters
Signals a structural shift in AI infrastructure investment from GPUs to memory components.
Nvidia's stock has fallen 15% since its May 2026 peak despite growing revenue projections. GPU spot prices on open compute markets have declined steadily, while DRAM memory chip prices have surged tenfold over the past year, making memory companies like Micron the new AI infrastructure darlings.
Nvidia, long the dominant force in AI infrastructure, has experienced a difficult stretch since May 2026. Its stock price has dropped 15% from peak levels, and on a price-to-earnings basis, the company now trades below the S&P 500 average — meaning investors are paying less per dollar of projected Nvidia profit than for the typical large U.S. company. Meanwhile, Micron, one of the world's largest DRAM manufacturers, has nearly tripled in value over the same period.
The shift reflects a fundamental change in the data center bottleneck. The GPU shortage that alarmed the industry in 2024–2025 has partially eased, while demand for high-bandwidth memory (HBM) — the specialized chips that move data in and out of processors — has vastly outpaced supply. DRAM spot prices have risen tenfold since 2023, according to data from Datatrack, with no major technical breakthrough driving the spike; the industry simply underestimated how much memory the AI buildout would require.
On the GPU compute side, data from the compute marketplace Ornn shows that spot pricing for an Nvidia H100 GPU peaked around $3.20 per hour in May 2026 and has since fallen steadily. Ornn co-founder and CTO Wayne Nelms attributed the disparity to straightforward supply-demand dynamics: GPU supply has grown faster than the market absorbed, while memory supply constraints remain severe.