Intel and AMD face server hardware squeeze as AI demand surges

Surging demand for AI computing is pushing the semiconductor industry toward a record $1.3 trillion in revenue this year, but it’s also making one thing increasingly scarce: server hardware.

The strain can be felt throughout the entire AI supply chain, particularly regarding Intel’s and AMD’s x86 server CPUs provided to the cloud market. Research firms such as Gartner, IDC, and Trendforce all acknowledge that demand remains resilient, memory prices are climbing steadily, and supply chain issues are likely to persist until 2027.

Their predictions jointly reveal a much bigger picture. Gartner concentrates on the spike in chip sales and prices, IDC monitors server delivery activity, and TrendForce monitors the deliveries of components. Put this way, it can be assumed that the most important problem that the industry currently faces is not the growth of demand but rather getting hardware produced in enough quantity. The news comes in the wake of Nvidia‘s push into the CPU industry, which adds more challenges for both Intel and AMD, as Cryptopolitan reported earlier.

AI silicon now takes 30% of every chip dollar

Gartner announced on April 8 that semiconductor earnings are expected to exceed $1.3 trillion globally by 2026, which means a surge of 64% on an annual scale, which is the highest growth the research organization has recorded in 20 years. It is estimated that chips designed specifically for AI will represent about 30% of gross sales.

A large segment of the expenditure can be attributed to the race of hyperscale cloud providers in building their AI infrastructure. According to the prediction of Gartner, the investment in AI infrastructure is expected to increase around 50% in 2026, resulting in heightened demand for CPUs, bespoke AI accelerators, and the servers that enable them.

According to Rajeev Rajput, senior principal analyst at Gartner, the industry’s third consecutive year with double-digit growth reinforces the central position of semiconductors in the AI technology stack.

While some manufacturing bottlenecks have started to be resolved, AI infrastructure continues to take in nearly all of the new production capacity, which leaves mainstream server components almost unavailable in the market.

Memflation pushes DRAM and NAND to fresh highs

Memory has become one of the biggest pain points.

Gartner uses the term “memflation” to describe the AI-driven surge in memory prices as demand continues to outpace supply. The firm expects DRAM prices to rise 125% in 2026 and NAND flash prices 234%, driving memory revenue from $216.3 billion last year to $633.3 billion.

The DRAM contract price is likely to increase again from 13% to 18% in Q3 while NAND flash contract prices are likely to see a rise of 10-15%, according to TrendForce.

Consumer demand for PCs and mobile devices has weakened, with buyers resisting higher prices, but this hasn’t relieved the pressure on data centers. Memory manufacturers continue to focus on the high-margin server segment, resulting in a tight supply overall.

x86 servers are the bottleneck, not the demand

The imbalance is evident in the latest numbers from IDC.

In the first quarter of 2026, the global revenues for server systems went up 30.7% year-on-year while the number of servers shipped rose only by 3.3%. Companies are now paying much more for slightly elevated hardware capacities, with a shortage of memory technology being the main constraint.

According to TrendForce, x86-based servers that use RDIMM memory are still the most optimal solution for many AI operations that require multitasking capability. As a result, Intel and AMD continue to dominate in terms of AI infrastructure growth even if memory shortage hampers the deployment of the platforms.

The main point for corporate customers is that there are now other elements that account for delivery times, in addition to processors. Even if Intel and AMD boost production through the construction of more factories and making supply contracts, issues with memory and other server parts can still lead to long delivery times for complete systems.

Buyers are being warned on the terms they sign

Analysts say buyers should think beyond today’s prices. Big hikes in prices will likely reduce after the first half of 2026, but it still predicts that prices will remain high in 2027. The consulting firm advises firms in IT not to enter long-term supply contracts that have unappealing prices. Rajput also mentions that memflation could “destroy, or at least delay, non-AI demand into 2028.”

The IDC predicts that NAND and memory supplies will remain tight until at least the first half of the year 2027, but the server market is expected to expand at a compounded annual rate of 25.1% until 2030.

Overall, the forecasts from Gartner, IDC, and TrendForce imply that the landscape of the AI infrastructure race is changing. The question is not whether businesses want to construct AI systems anymore, but whether they are able to obtain the necessary hardware at a reasonable price.

Don’t just read crypto news. Understand it. Subscribe to our newsletter. It's free.