AI Data Centers Squeeze Memory Supply, Coalition Warns

retained text

AI Data Centers Squeeze Memory Supply, Coalition Warns
With fabs “fully allocated,” the question is where the capacity is going – and what it means for non‑AI buyers.
A coalition of telecom, automotive, medical device, and retail groups is warning the Trump administration that surging AI data center demand is driving sharp increases in memory prices and constraining supply across key industries.
In a June 3 letter to Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent, the organizations argued that expanding AI data centers are consuming a growing share of memory production capacity, leaving less supply available for communications networks, automobiles, medical equipment, consumer electronics, and other products.
"Expanding artificial intelligence (AI) data centers consume an enormous share of available memory chip capacity," the coalition wrote. "The result has been an unprecedented surge in the price of memory chips and reduced supply of these chips for manufacturing and consumer-facing industries."
The groups warned the trend could increase the cost of telecommunications infrastructure, consumer electronics, and information technology products while disrupting automobile and medical device supply chains and delaying deliveries by federal contractors.
Signatories include NCTA – The Internet & Television Association, ACA Connects – America's Communications Association, NTCA – The Rural Broadband Association, the Telecommunications Industry Association, AdvaMed, the Medical Device Manufacturers Association, the Alliance for Automotive Innovation, the Retail Industry Leaders Association, and the National Retail Federation.
DRAM Prices Spike as Capacity Shifts to AI
Market data suggests mounting pressure in memory markets. Research firm TrendForce reported that conventional DRAM contract prices rose roughly 93%–98% quarter over quarter in the first quarter and said manufacturers are continuing to shift capacity toward high-bandwidth memory (HBM) and server-related products used in AI systems.
Stephen Sopko, semiconductor and deep tech analyst at HyperFrame Research, said the effects are no longer confined to AI infrastructure.
"Wider DRAM constraints are cropping up as well, not only affecting AI and tech but also embedded compute, automotive, telecom, medical devices, and other sectors," Sopko told Data Center Knowledge.
Analysts tracking the automotive sector have identified similar dynamics. S&P Global Mobility has noted that memory manufacturers are directing investment and production toward higher-margin AI and data center products, tightening supply for automotive customers and driving increases in automotive memory pricing.
For the telecom sector, the coalition's most direct concern is infrastructure cost. The groups specifically warn of higher costs for building, maintaining, and upgrading communications networks as operators continue investing in broadband expansion and capacity upgrades.
Sopko said operators should pay close attention to how semiconductor manufacturers allocate future production capacity as demand tightens across multiple markets.
"When the big fabs say they are fully allocated, the real question is: Where are those allocations going?" he said.
The letter does not identify which memory products are driving the concern. While much attention has focused on HBM demand from AI systems, the coalition argues the effects are spreading into broader memory markets.
Coalition Urges US to Expand Memory Capacity
The organizations urged the administration to work with memory manufacturers and buyers to address the imbalance. Their recommendations include expanding memory production capacity, strengthening supply-chain cooperation through trade agreements, reviewing CHIPS Act programs, reducing barriers to alternative sourcing, and monitoring market conditions.
One of Sopko's takeaways from the letter was that the coalition is asking policymakers to expand memory production rather than restrict AI infrastructure growth.
"In a time when factions are arising to limit data center expansion, this letter simply asks the government to remove red tape and do what it can to accelerate the production of memory," Sopko said.
Power systems, transformers, transmission equipment, cooling technologies, and advanced GPUs have already emerged as pressure points as developers race to build larger AI campuses. The coalition argues memory is emerging as the next critical bottleneck in the AI infrastructure buildout.

document fetched 2026-06-06 08:27