Nvidia Corp has informed customers that prices for its artificial intelligence servers will rise by more than 15%. The increase applies to systems equipped with Vera Rubin and Grace Blackwell platforms, with shipments expected in early 2027.

The hike is driven by surging costs for memory chips, which constitute a significant portion of server bill-of-materials. Suppliers including Samsung Electronics, SK Hynix Inc, and Micron Technology have gained pricing power as AI infrastructure demand strains global supply chains.

Affected Platforms and Customers

The price adjustment is not uniform; it depends on the specific Nvidia chip generation and memory configuration used in each system. Server manufacturers building for major data center operators have already begun notifying their clients.

Key companies reportedly affected include:

• Microsoft Corp
• Alphabet Inc’s Google
• Oracle Corp

Nvidia did not immediately respond to requests for comment regarding the specific magnitude of increases for individual SKUs.

Market Context

The move reflects broader inflationary pressures in the semiconductor supply chain. Similar to recent price adjustments by Apple Inc and Qualcomm Inc, Nvidia is passing on higher input costs to enterprise buyers. This development comes as the company prepares to report second-quarter results on Aug. 26, with analysts projecting revenue of $92 billion, a 96% jump year-over-year. Third-quarter guidance is projected to reach $103 billion.

Nvidia shares closed at $214.72 on Friday, down 0.98% for the day. In the past 12 months, Nvidia shares are up by 19.41%.

What the Numbers Show

The correlation between rising memory costs and Nvidia’s pricing strategy highlights a dependency in AI hardware economics. While Nvidia’s revenue is projected to surge to $92 billion in Q2, the pass-through of component costs suggests that gross margin expansion may face headwinds if memory prices continue to outpace volume efficiencies. The 15%+ hike indicates that supply constraints are shifting bargaining power from chipmakers to component suppliers like SK Hynix and Micron.

SpaceXAI and Nvidia plan to launch a space-optimized Vera Rubin NVL72 system to orbit in the fourth quarter of next year, with significant scale expected in 2028.

SpaceXAI, the artificial intelligence division of SpaceX (NASDAQ: SPCX), announced Monday that it will deploy NVIDIA Corp. (NASDAQ: NVDA) Vera CPUs to power its next generation of agentic artificial intelligence applications. The strategic partnership aims to enhance the performance of AI agents while optimizing energy efficiency across SpaceXAI’s computing infrastructure.

The collaboration extends beyond CPU deployment. SpaceXAI plans to expand its infrastructure for Grok using the NVIDIA Vera Rubin platform. This system integrates computing, networking, data processing, and software technologies to support scaling toward gigawatts of computing capacity. The platform is designed to handle AI training, reasoning, and inference workloads across large data centers.

Vera Targets Agentic AI Workloads

Agentic AI systems utilize CPUs to coordinate tools, run code, process data, and manage simulations between model requests. SpaceXAI intends to use the NVIDIA Vera CPU for these tasks, ensuring that graphics processing units remain fully utilized for other operations.

The Vera chip features 88 NVIDIA-designed Olympus cores and high-bandwidth LPDDR5X memory. It delivers up to 1.2 terabytes per second of memory bandwidth. According to NVIDIA, Vera can complete certain AI, reinforcement learning, and data-processing tasks up to 1.8 times faster than x86 CPUs.

Mike Nicolls, president of SpaceXAI, stated that the technology should improve AI agent performance while allowing the company to extract more work from each watt of computing power.

Orbital Computing Expansion

The partnership also involves bringing NVIDIA’s accelerated computing technology into orbit. Elon Musk posted on X that SpaceX, in partnership with Nvidia, has designed a space-optimized Vera Rubin NVL72 system for launch to orbit in Q4 next year, with significant scale in 2028.

SpaceXAI’s first-generation Starmind AI satellite will use an optimized Vera Rubin NVL72 rack-scale system. NVIDIA and SpaceXAI plan to adapt this architecture to meet the specific power, cooling, bandwidth, and reliability demands of orbital computing.

Market Reaction

NVIDIA shares were down 1.88% at $210.68 at the time of publication on Monday, according to Benzinga Pro data.