SANTA CLARA, CALIFORNIA / RankWire.AI / – Nvidia is planning to implement price increases exceeding 15% on numerous AI server configurations scheduled for shipment in early 2027. The adjustments impact systems built around Vera Rubin and Grace Blackwell technologies. Final price changes vary depending on chip generation, memory capacity, and system architecture. Nvidia has not announced a single companywide increase that encompasses all server models. Instead, manufacturers assembling AI systems have communicated revised pricing to major data center clients.
Microsoft, Google, and Oracle are among the leading cloud providers purchasing large quantities of accelerated computing hardware. Their data centers utilize AI servers for tasks like model training, inference, and cloud services. During 2026, memory costs have emerged as one of the most significant financial pressures in these systems. Modern AI servers integrate GPUs with high-bandwidth memory, server DRAM, storage, and fast networking components. The strong demand for these parts has resulted in tight supplies across various segments of the memory market.
TrendForce forecasts a 13% to 18% increase in contract prices for conventional DRAM during the third quarter of 2026. Additionally, it predicts NAND Flash contract prices will rise between 10% and 15% within the same period. Server DRAM remains highly constrained as memory manufacturers divert more capacity towards AI and data center applications. These rising memory costs have driven up the expense of constructing advanced computing systems, forming a crucial element of the pricing environment for next-generation AI servers.
Memory pricing pressures intensify within AI infrastructure
According to Nvidia, Vera Rubin reached full production with system integrators and supply-chain partners in 2026. Systems leveraging the platform are expected to become available in the latter half of the year. Rubin features the Vera CPU and Rubin GPU combined with NVLink 6 and multiple networking technologies. The platform aims to support large-scale AI workloads in cloud and hyperscale data centers. It succeeds Grace Blackwell as Nvidia’s latest rack-scale computing architecture.
Grace Blackwell remains a fundamental platform in current AI data center deployments. The GB200 NVL72 system integrates 36 Grace CPUs with 72 Blackwell GPUs within a liquid-cooled rack. Nvidia designed this setup to function as a single, expansive NVLink computing domain. Price adjustments linked to these servers depend on hardware configuration specifics rather than a uniform percentage. Variations in memory capacity, processor generation, and rack design influence each server’s final cost.
Demand for servers sustains tight memory supply conditions
Memory manufacturers have shifted increased production towards server and high-performance modules as AI demand continues to absorb supply capacity. TrendForce reports this shift has reduced the availability of certain PC and consumer memory categories. Data center operators maintained large-scale purchases of server memory through 2026, and the research firm predicts that server DRAM availability will remain constrained into 2027, with demand surpassing new supply. This ongoing environment continues to influence component costs across AI infrastructure.
Nvidia enters this pricing cycle following another quarter of record data center revenue, reporting fiscal first-quarter revenue of $81.6 billion for the period ending April 26, 2026. Data Center revenue reached $75.2 billion, representing a 92% increase compared to the same quarter in the previous year. Nvidia also provided guidance for second-quarter revenue at around $91 billion, plus or minus 2%. The company is set to announce its fiscal second-quarter results on Aug. 26, offering an update on its latest financial performance.