Micron Says AI Memory Crunch Could Last Beyond 2027 as DRAM Demand Soars

Key Points

• Memory shortages may persist beyond 2027, with Micron expecting 2027 supply conditions to be even tighter than 2026 as AI-driven DRAM demand outpaces new fabrication capacity.
• AI workloads are intensifying demand for high-bandwidth memory (HBM), but shifting production toward HBM further reduces conventional DDR output—by roughly three bits of DDR for every bit of HBM3E, potentially worsening the broader supply crunch.
• Micron is securing long-term growth through strategic customer agreements with $22 billion in commitments and plans to raise U.S. investment to $250 billion, while also targeting future demand from HBM4E and physical AI applications such as robotics.

Micron Technology NASDAQ: MU expects memory-market supply constraints to persist beyond 2027 as demand driven by artificial intelligence continues to outpace the industry’s ability to add manufacturing capacity, Executive Vice President and Chief Business Officer Sumit Sadana said at KeyBanc Capital Markets’ Annual Technology Leadership Forum.

Sadana said customer demand signals have increased since Micron’s latest earnings report, leading the company to expect calendar 2027 to be “even tighter than 2026.” He said the company does not yet have visibility into when industry supply will catch up with demand.

“The number one constraint” for customers is DRAM, rather than power availability, real estate, data-center capacity or logic wafers, Sadana said. He attributed the imbalance in part to the difficulty and length of time required to construct and ramp leading-edge memory fabrication plants.

AI Demand Reshapes Memory Market

Sadana characterized the current environment as fundamentally different from prior memory cycles, citing the expansion of generative AI, agentic AI and future artificial general intelligence applications. He said agentic AI workloads can require five to 30 times more tokens than comparable chat-interface tasks, with deep-reasoning workloads requiring still more.

He also pointed to the rising importance of high-bandwidth memory, or HBM, in AI systems. According to Sadana, processors can sit idle while waiting for data from DRAM, making higher memory bandwidth and capacity necessary to improve system utilization.

Micron has previously discussed a tradeoff between HBM production and conventional DDR memory supply. Sadana said producing 100 bits of HBM can reduce DDR output by roughly 300 bits for HBM3E, a three-to-one trade ratio. He said that ratio could approach four-to-one with HBM4E, increasing pressure on conventional memory supply.

While data centers represent the most acute area of demand, Sadana said demand is elevated across market segments. He said some data-center customers cannot obtain more than half of the memory volume they seek, despite high pricing.

Customers are adjusting system memory configurations primarily because of constrained availability, rather than pricing, Sadana said. While lower average DRAM capacity can allow customers to ship more systems, he said it can also reduce processor utilization and create latent demand for higher-capacity configurations when supply becomes available.

Strategic Customer Agreements

Sadana discussed Micron’s strategic customer agreements, or SCAs, which he described as distinct from historical long-term agreements in the memory industry. The agreements cover multiyear periods, with the majority of SCA-related revenue expected to fall under terms extending through the end of calendar 2030, he said.

Unlike prior arrangements, Sadana said the SCAs include binding purchase commitments, take-or-pay provisions and no contractual exits for customers. At the time of Micron’s earnings report, the company had announced 16 agreements that included $22 billion in cash and cash-like commitments, including $18 billion in cash expected to be held on Micron’s balance sheet.

Some agreements use market-based pricing, while most volume covered by the agreements will include pricing bands, Sadana said. He said the floor prices are set at levels intended to generate gross margins above previous industry-cycle peaks.

Sadana said the agreements also support deeper engineering collaboration with customers, including product and research roadmaps extending beyond 2030. He cited Micron’s HBM3E product, which he said offered 30% lower power consumption than the next-best product, and its work with NVIDIA to bring low-power DRAM into data centers.

Investment and U.S. Manufacturing

Micron is increasing its planned U.S. investment to $250 billion from $200 billion over the coming years, Sadana said. The company is also investing across its manufacturing network in Japan, Taiwan and Singapore, as well as in back-end manufacturing in India.

He said Micron has committed $500 million to GlobalWafers for raw wafers and is participating in a broader $3 billion supply-chain investment effort.

Sadana described Micron as the only company investing in front-end memory-fab manufacturing in the U.S. He said Micron’s Idaho 1 facility is expected to come online in the middle of next year, with Idaho 2 expected at the end of 2028. The company also plans a New York fab cluster and investments in Virginia, including the introduction of 1-alpha DRAM technology.

Micron expects its U.S. manufacturing footprint to command a pricing premium and provide customers with greater supply-chain resilience, Sadana said.

HBM and Physical AI Opportunities

Looking ahead, Sadana said HBM4E will introduce opportunities for customized HBM products. He expects many HBM programs to use one or two suppliers because qualification and co-engineering processes are time-consuming and expensive.

He also highlighted “physical AI,” including robotics and humanoid robots, as an emerging long-term demand driver. Sadana said a humanoid robot could require hundreds of gigabytes of DRAM and terabytes of solid-state storage to support onboard computing, safety and responsiveness when cloud access is unavailable.

While robotics remains in its early stages, he said Micron expects the segment to grow later this decade and potentially enter a more rapid growth phase in the early part of the next decade.

About Micron Technology (NASDAQ:MU)

Micron Technology, Inc is a global semiconductor company that designs and manufactures memory and storage solutions. Its product portfolio includes dynamic random-access memory (DRAM), NAND flash memory, solid-state drives (SSDs), memory modules and embedded memory solutions for a wide range of computing and electronic devices. Micron supplies components used in data centers, enterprise and cloud infrastructure, client computing, mobile devices, automotive systems and industrial applications, and also markets consumer-facing products under the Crucial brand.

Founded in 1978 and headquartered in Boise, Idaho, Micron has grown into an international manufacturer with research, development and production facilities across multiple regions.

This instant news alert was generated by narrative science technology and financial data from MarketBeat in order to provide readers with the fastest reporting and unbiased coverage. Please send any questions or comments about this story to contact@marketbeat.com.

Continue following MarketBeat

Add MarketBeat as your preferred source on Google to see our latest stories in your feed.

Add As Preferred Source