Nvidia’s stock has fallen 15% since its May peak, even as projected revenue continues to grow, leaving the company trading cheaper relative to earnings than the S&P average. Meanwhile, memory chipmaker Micron has nearly tripled in value over the same period, as high-bandwidth memory emerges as a new bottleneck for AI data centers. The shift reflects an easing GPU shortage combined with surging demand for memory components needed to move data through processors.
Compute pricing data from marketplace Ornn showed the spot price for an hour of Nvidia H100 GPU time peaking near $3.20 in May before steadily declining, mirroring Nvidia’s stock trajectory. By contrast, DRAM spot prices have risen roughly tenfold over the past year, according to data from Datatrack, despite no major technological change in memory chip design.
Wayne Nelms, co-founder and CTO of Ornn, attributed the divergence to supply and demand dynamics, noting that companies including Google, Amazon, Microsoft and OpenAI have developed custom AI chips to reduce reliance on Nvidia, helping push compute prices down. Nelms said no comparable shift has occurred in memory production, with no new entrants or major breakthroughs in high-bandwidth memory technology, suggesting current pricing trends are likely to persist absent significant supply or technological changes.
