SAN FRANCISCO, January 24, 2026 — Data centers are facing a surprising bottleneck in the age of artificial intelligence: not graphics processing units (GPUs) themselves, but the central processing units (CPUs) needed to support them. This unexpected twist is contributing to rising costs and extended lead times for essential hardware, impacting enterprises across the board.
How Pricing Might Increase for Enterprises
The surge in demand for advanced chips has created shortages at foundries producing Intel’s 10/7 nanometer nodes, which currently handle the majority of the company’s production. These wafers require up to three quarters to move through the fabrication process, meaning Intel anticipates being constrained until at least the second quarter of 2026, when increased chip production is projected.
Currently, manufacturing capacity for Xeon processors is fully booked for 2026, with distributors reporting varying lead times. Custom silicon programs are facing waits of 6 to 8 months, with some orders extending into 2027.
Memory is emerging as the most significant bottleneck in data centers, with prices expected to climb more than 65% year-over-year in 2026. NAND Flash is also projected to see increases of up to 25%. Some products have already experienced price inflation exceeding 1,000% since 2025, and substantial new memory capacity isn’t anticipated until 2027 or 2028.
What will happen with memory prices? While some anticipate continued volatility, forecasts suggest memory prices may stabilize later this year as more capacity comes online in 2027.
Moor’s Sag offered a slightly more optimistic outlook, predicting that memory prices on the client side will “probably stabilize this year until more capacity comes online in 2027.”
How Enterprises Can Prepare
Diversifying suppliers is the most effective strategy for enterprises at this time. While adding complexity, a multi-supplier approach allows data center operators to better absorb price fluctuations by shifting orders to companies with more robust supply chains or better planning.
- CPUs are becoming a critical bottleneck in AI infrastructure, despite not being directly replaced by GPUs.
- Foundry shortages of Intel 10/7 nodes are driving up costs and extending lead times.
- Memory prices are soaring, with increases exceeding 65% expected in 2026.
- Supplier diversification is crucial for mitigating risk and absorbing price shocks.
