
What happened
Rising prices for DRAM memory chips are forcing server makers to charge Nvidia's largest customers—including Microsoft, Google, and Oracle—price increases exceeding 15% for systems shipping early next year, equipped with Nvidia's Vera Rubin and Grace Blackwell chips. Memory chipmakers Samsung Electronics, SK Hynix, and Micron now control most of the world's DRAM production and have gained unprecedented leverage.
Why it matters
Nvidia's AI accelerators depend on paired DRAM capacity to function effectively. Server cost inflation ripples through the entire data center industry and threatens to slow the massive AI infrastructure build-out already complicated by project delays, labor shortages, and capital constraints. Nvidia itself has a 75% gross margin and charges tens of thousands of dollars per chip because supply from Taiwan Semiconductor Manufacturing Co cannot meet demand—yet memory costs are now pressuring even its biggest customers upward.
What to watch
Major cloud operators—Amazon, Microsoft, Google, and Meta—are all pursuing in-house chip programs to reduce Nvidia dependence, but their success hinges on securing their own DRAM supply from Samsung, SK Hynix, and Micron. Nvidia has also recently raised prices for gaming-oriented PC graphics cards, signaling broader pricing pressure across its product line.
Summaries like this, in your inbox every morning.
The shortage of DRAM memory has become a bottleneck in the AI infrastructure race. While Nvidia's own chips remain in acute demand—allowing the company to maintain a 75% gross margin and charge tens of thousands of dollars per unit—the chipmaker's customers are now being squeezed by memory costs they cannot easily control. Samsung Electronics, SK Hynix, and Micron collectively dominate DRAM production and have leveraged the surge in AI infrastructure demand to raise prices significantly. The size of increases varies depending on chip generation and memory configuration, but some customers face over 15% hikes starting early next year.
This cost pressure complicates the industry's ability to execute massive AI data center build-outs. The report notes that project delays, labor shortages, tightening capital markets, and community resistance have already strained many plans; rising server costs add another layer of friction. Major cloud operators—Amazon, Microsoft, Google, and Meta—all have in-house chip programs underway, but these remain long-term hedges; for now, they remain dependent on Nvidia purchases and must also compete for limited DRAM supply from the same three memory makers. Nvidia has also recently raised prices on gaming-oriented PC graphics cards, signaling that cost pressures extend beyond data center chips.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
The Motley Fool argues Amazon is the AI company of the two, since AWS was just 21% of first-half 2026 revenue…

Hewlett Packard Enterprise stock is down 5% at $55.87

Vercel data published Thursday shows TypeSafe AI's Jev, a lightweight decision model for choosing among predet…

Chunghwa Telecom president Rong-shy Lin said the shift from conversational AI to autonomous AI agents is resha…

CoreWeave fell 5% in premarket trading Friday and was on track to finish the week down 15%, after rebounding f…

Apple and Meta said they would jointly release a new open-weight AI model designed to run on high-end Macs, ai…
