SANTA CLARA, CALIFORNIA / RankWire.AI / – Nvidia is planning to increase the prices of many AI server configurations by more than 15% in shipments scheduled for early 2027. These hikes affect models built on the Vera Rubin and Grace Blackwell platforms, with the magnitude of each increase depending on chip generation and memory configuration. Nvidia has not publicly announced a blanket price hike across all its server offerings, but server manufacturers have passed this new pricing information on to key data center clients.
Major data center operators, including Microsoft, Google, and Oracle, have been informed of these price changes by server manufacturers, as they purchase large volumes of AI infrastructure for their cloud and data center operations. The rising costs of memory have become a significant factor in the reported server price increases, as AI systems demand substantial amounts of high-bandwidth and server memory along with GPUs, CPUs, networking hardware, and storage. The ongoing demand from AI deployments has kept several memory categories tight throughout 2026.
TrendForce predicts that conventional DRAM contract prices will increase by 13% to 18% during the third quarter of 2026, while NAND Flash contract prices are projected to rise 10% to 15% in the same period. The research firm noted that server DRAM remains undersupplied as manufacturers prioritize capacity for AI-related applications. Although long-term supply agreements have constrained some price hikes, overall server memory costs continue to be high.
Memory costs climb amid high demand for AI servers
In May, Nvidia announced that Vera Rubin had entered full production with server manufacturers and supply-chain partners globally, with partner products using Rubin expected to launch in the second half of 2026. Vera Rubin integrates the Vera CPU, Rubin GPU, NVLink 6, ConnectX-9, BlueField-4, and Spectrum-6 networking technologies, succeeding Grace Blackwell as Nvidia’s latest rack-scale architecture for large AI workloads. System manufacturers are developing Rubin-based products for cloud providers, AI laboratories, and hyperscale data centers.
Meanwhile, Grace Blackwell remains a cornerstone of Nvidia’s current data center lineup. The GB200 NVL72 design connects 36 Grace CPUs with 72 Blackwell GPUs within a single liquid-cooled rack, designed to function as one NVLink computing domain for large AI models. The reported server price increases are varied depending on configuration, with memory capacity and chip generation playing key roles in pricing. Therefore, the 15% increase figure does not apply uniformly to all Nvidia server setups.
Expansion of Vera Rubin production influences Nvidia’s server lineup
As AI demand continues to absorb capacity, memory suppliers have shifted production toward server and high-performance products, reducing supply for some PC and consumer DRAM. According to TrendForce, this shift has limited the availability of certain DRAM products for personal computers. While server shipments have stayed strong through 2026, buyers have increased inventories of data center memory, and the market analyst expects supply tightness to persist into 2027 as demand growth outpaces new capacity, creating the cost environment behind Nvidia’s announced price adjustments.
Nvidia enters this pricing phase with record-breaking data center sales, having reported total revenue of $81.6 billion in its fiscal first quarter ending April 26. Data Center revenue hit a record $75.2 billion, up 92% from the previous year. When reporting those results in May, Nvidia projected second-quarter revenue of around $91 billion, plus or minus 2%. The company plans to release its fiscal second-quarter earnings on August 26.