NVIDIA has notified key customers that AI server prices will increase by over 15% for most configurations starting early 2027, driven by soaring memory chip costs, according to Bloomberg.

The hike applies to servers powered by the Vera Rubin and Grace Blackwell flagship GPU platforms, with final adjustments varying by chip iteration and memory configuration. Tier‑1 OEMs serving Microsoft, Alphabet (Google), Oracle, and other hyperscalers have already begun issuing price‑alert notices to downstream clients.

NVIDIA dominates the global AI infrastructure market, and its pricing moves are widely watched as a bellwether for the AI supply chain. The company is set to report Q2 earnings on August 26, and investors will be closely monitoring how rising memory costs impact its margin outlook.

ICgoodFind Takeaway:
AI server price hikes are the latest sign of memory supply tightness feeding downstream. Expect cloud compute costs to follow suit – budget planning for AI capacity should factor in these increases.