Nvidia’s contract server builders have told Microsoft, Google, and Oracle that prices on AI server systems will rise more than 15% starting on shipments in early 2027, hitting the company’s flagship Vera Rubin and Grace Blackwell configurations.
The price increase, first reported by Bloomberg and confirmed by Fortune, is driven by soaring memory-chip costs from Samsung, SK Hynix, and Micron that Nvidia can no longer absorb even at its 75% gross margin. The pass-through is the first broad hyperscaler-facing sticker shock of the Nvidia Rubin era.
Memory Costs Force Unprecedented Pass-Through
Rising DRAM and HBM memory prices have squeezed Nvidia’s ability to absorb component cost increases across its full product line. The price hike varies by chip generation and memory configuration, with systems containing Vera Rubin accelerators facing the steepest increases. The move ripples the DDR5 and HBM memory squeeze into finished AI server systems just as hyperscalers finalize their 2027 capital expenditure budgets.
Memory chip costs have climbed 30-40% year-over-year as AI training workloads demand ever-larger memory configurations. HBM supplies remain constrained despite aggressive capacity expansion by Samsung and SK Hynix, while DDR5 demand from AI inference systems has added further pressure on an already tight market.
Hyperscalers Face Tricky 2027 Budget Decisions
The timing of the increase is significant. Hyperscalers are locking in their 2027 spending plans now, and a 15%-plus jump on finished systems could force difficult trade-offs between compute capacity and networking or storage investments. Google, Microsoft, and Oracle have all announced massive AI infrastructure buildouts for next year, with combined planned spending expected to exceed $200 billion.
Nvidia holds roughly 90% of the AI accelerator market, leaving hyperscalers with limited alternatives. AMD’s MI400 series, expected in early 2027, could eventually provide competitive pressure, but Nvidia’s software ecosystem and developer tooling remain significant switching barriers.
The price hikes come at a delicate moment for Nvidia itself. The company is ramping production of its Vera Rubin architecture while still clearing Blackwell inventory, and any perception of pricing friction could slow the adoption cycle for its next-generation products.
discussion