Memory Chip Shortage Intensifies as AI Data Centers Claim Supply

AI Infrastructure Consuming Memory Production Capacity

A coalition representing telecommunications, automotive, medical device, and retail industries has warned the Trump administration that explosive growth in AI data center construction is creating severe constraints in memory chip markets, driving unprecedented price increases and supply shortages across critical sectors.

In a June 3 letter to Commerce Secretary Howard Lutnick and Treasury Secretary Scott Bessent, nine industry organizations argued that AI data centers are consuming a disproportionate share of memory production capacity, leaving insufficient supply for communications networks, automobiles, medical equipment, consumer electronics, and other essential products. The coalition includes NCTA – The Internet & Television Association, the Alliance for Automotive Innovation, AdvaMed, and the National Retail Federation, among others.

"Expanding artificial intelligence data centers consume an enormous share of available memory chip capacity," the groups wrote, warning the trend could disrupt automobile and medical device supply chains while increasing costs for telecommunications infrastructure and federal contractors.

Why It Matters

The memory shortage represents a new constraint on AI infrastructure buildout that extends beyond the widely reported bottlenecks in power systems, transformers, and GPUs. More significantly, it signals that AI's resource demands are now affecting industries far removed from technology, potentially slowing automotive production, delaying medical device manufacturing, and raising costs for broadband expansion—all sectors critical to economic competitiveness and public welfare.

DRAM Prices Surge as Fabs Prioritize AI Products

Market data confirms the coalition's concerns. Research firm TrendForce reported that conventional DRAM contract prices increased approximately 93-98% quarter over quarter in the first quarter of 2026, as manufacturers shifted production capacity toward high-bandwidth memory (HBM) and server products used in AI systems.

"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," Stephen Sopko, semiconductor and deep tech analyst at HyperFrame Research, told Data Center Knowledge, which first reported the story.

S&P Global Mobility has documented similar dynamics in the automotive sector, noting that memory manufacturers are directing investment toward higher-margin AI and data center products, tightening supply for automotive customers and driving price increases.

For telecommunications operators, the immediate concern centers on infrastructure costs. The coalition specifically warned of higher expenses for building, maintaining, and upgrading communications networks as operators continue investing in broadband expansion.

"When the big fabs say they are fully allocated, the real question is: Where are those allocations going?" Sopko said.

Coalition Calls for Production Expansion, Not AI Limits

The organizations urged the administration to work with memory manufacturers and buyers to address the supply 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.

Notably, the coalition is not asking policymakers to restrict AI infrastructure growth. "This letter simply asks the government to remove red tape and do what it can to accelerate the production of memory," Sopko observed.

The coalition argues that memory is emerging as the next critical bottleneck in AI infrastructure buildout, joining power systems, transformers, transmission equipment, cooling technologies, and advanced GPUs as pressure points in the race to build larger AI campuses.

While the letter does not specify which memory products are driving the concern, analysts note that while HBM demand from AI systems has received significant attention, the effects are now spreading into broader memory markets affecting conventional DRAM and other products.

The details were first reported by Shane Snider at Data Center Knowledge.