Nvidia (NASDAQ: NVDA) has told some of its largest customers that the price of servers containing its AI chips will rise by more than 15% in many cases, Bloomberg reported, with soaring memory costs the driver. The increases apply to systems shipping in early 2027, including those built on Nvidia’s flagship Grace Blackwell and next-generation Vera Rubin platforms, and the size of each hike depends on the chip generation and memory configuration. It is a striking move from the most powerful company in the AI chip supply chain, and it lands three days before Nvidia reports earnings.
The news matters beyond the headline number because of what is forcing it. This is the AI boom inflating the cost of its own inputs, a dynamic some have started calling “AI-flation,” where demand for artificial intelligence drives up the price of the components AI itself depends on. When even Nvidia, with gross margins around 75%, passes memory costs through to buyers, it signals that the squeeze has reached the very top of the semiconductor stack.
What Nvidia Told Its Customers
The specifics, per Bloomberg, are that contract manufacturers assembling servers for large data-center operators such as Microsoft, Google, and Oracle have notified their customers of the coming increases. The reporting is single-sourced; Reuters, which summarized it, said it could not immediately verify the details, and the hikes described apply to new systems shipping early next year rather than orders already under contract. Nvidia has not commented publicly.
Even with those caveats, the direction is unmistakable, and it follows a pattern. Nvidia raised prices on its GeForce graphics cards earlier this month, and the pressure has already reached consumers elsewhere: Apple lifted prices on Macs and iPads in June, with CEO Tim Cook citing memory costs, and Amazon raised prices on Echo and Kindle devices. The AI server hike is the same force reaching the most expensive hardware in the chain, where a 15% increase on rack-scale systems that sell for millions of dollars adds hundreds of thousands per rack.
The Memory Squeeze Behind It
The cause is a historic shortage in memory chips. High-bandwidth memory and server DRAM are essential to AI accelerators, and their prices have climbed at record rates as the three big memory makers, Samsung, SK Hynix, and Micron (NASDAQ: MU), shift capacity toward the high-margin AI parts and starve everything else. Conventional DRAM contract prices rose an estimated 90% to 95% quarter over quarter in the first quarter of 2026, with a further 58% to 63% projected in the second, according to TrendForce data reported by Tom’s Hardware.
Two consecutive quarters of extraordinary DRAM contract-price increases are the cost pressure now reaching AI servers. Source: TrendForce data via Tom’s Hardware (Q2 projected) · Chart: FinanceFeedsMemory now accounts for roughly a quarter of the bill of materials for a high-end AI server rack, per Deloitte estimates, which is why a memory spike forces a server price hike rather than getting absorbed. And the shortage is not resolving quickly: analysts at Gartner expect the crunch to persist at least through the first half of 2027. That hands enormous leverage to the memory suppliers, a shift FinanceFeeds tracked in its look at how institutions are positioned across semiconductor stocks, where Micron sits alongside Nvidia as a core AI-infrastructure holding.
Investor Takeaway
The hike is Bloomberg-sourced and unverified by Reuters, and it applies to systems shipping in early 2027, so it is a forward signal about margin pressure rather than a change to revenue Nvidia will report on Wednesday.
What It Means for Nvidia’s Margins
For Nvidia, passing the cost through is a show of pricing power, particularly given its more than 70% share of the data-center AI chip market and margins near 75%, which give it room to raise prices without obviously losing customers that have few alternatives. That is the bullish read for anyone weighing AI stocks.
But the move cuts both ways. Higher server prices raise the cost of the AI buildout for the hyperscalers doing the buying, and if memory inflation keeps climbing, the question becomes whether Nvidia can keep passing it through without denting demand or its own margins. That tension, mapped in FinanceFeeds’ bull and bear breakdown of NVDA, is the one this news sharpens. The stock offered no cheer ahead of the print, easing to about $215 and down roughly 5% on the week.
Nvidia drifted lower into its earnings date, down about 5% on the week. Source: TradingViewThe Q2 Report on Wednesday
Nvidia reports second-quarter fiscal 2027 results on Wednesday, August 26, with Wall Street expecting revenue to roughly double from a year earlier. The price-hike news frames the key question for that report: whether rising memory costs are starting to pressure the margins that have defined Nvidia’s run, or whether its pricing power is strong enough to pass the bill straight through.
The AI-capex backdrop, from Nvidia’s trimmed OpenAI-linked financing guarantee to its stake in cloud provider Nebius, makes the stakes for Wednesday’s guidance unusually high.
Investor Takeaway
The central question Wednesday is margins, since the price hike suggests memory inflation is real, so watch Nvidia’s gross-margin guidance more than the revenue beat the Street already expects.