Nvidia’s AI chips are set for another price hike. According to Bloomberg, Nvidia has notified its largest customers—including server manufacturers building AI data centers for tech giants like Oracle and Microsoft—that server prices will riseby over 15%This means computing costs across the entire AI industry will rise sharply in the short term
The price increase is unsurprising—but the magnitude is staggering
Nvidia has already raised GPU prices this year, and this 15% increase applies to full server solutions—not just GPUs. Given Nvidia’s dominance in the AI training chip market, holdingover 80%share, few competitors can effectively hedge against this move
Multiple factors are driving this round of price hikes: tariff policies raising component costs, surging AI demand straining production capacity, and Nvidia’s massive R&D investment in its Blackwell architecture. Nvidia will release its Q2 earnings report next Wednesday (August 27), when further details may emerge
Who will feel the pain first?
- AI startupsMost sensitive to compute costs—this 15% increase could directly squeeze profit margins and even disrupt fundraising timelines
- Cloud service providersAWS, Azure, Google Cloud, and others will likely pass costs on to end users, potentially raising API call prices
- Open-source model communitiesIndividual developers and researchers relying on cloud GPU rentals will face higher barriers to entry for open-source AI
- Chinese AI companiesAffected by export controls and already facing difficulties acquiring chips, global price hikes further intensify cost pressures
The ripple effect of rising compute costs
This is not an isolated price increase. Over the past month, we have seen multiple related signals: Nvidia backed OpenAI’s data center construction with a $10.5 billion guarantee; Starcloud secured $250 million to build an orbital data center; private equity giants such as Apollo heavily invested in GPU lending businesses. The entire AI infrastructure sector is undergoingUnprecedented capital intensification.
When chip prices rise alongside surging data center construction costs, AI model training and inference costs will inevitably increase. Ultimately, consumers and enterprise users of AI products will bear these costs. For enterprises weighing the “build vs. rent” decision for AI compute, this decision window is rapidly narrowing.
Summary: The era of compute inflation has arrived
Nvidia’s 15% price hike is neither the first nor the last. Amid surging AI demand, geopolitical disruptions to supply chains, and Nvidia’s de facto monopoly,Structural increases in compute costsWill be one of the dominant themes in the AI industry over the next two to three years. For AI tool users and developers, selecting higher-value models and inference solutions will become increasingly critical.
Follow the latest updates and value-for-money comparisons of AI tools atAI Dash——Discover the best AI tools.