Micron CEO Interview: "Memory" is the Overlooked Bottleneck of AI, Supply Constraints May Extend Beyond 2026

Original Author: Li Jia

Original Source: Wall Street News

"The AI race isn't just a computing race; it's also a storage race." This is the assessment from Micron Technology CEO Sanjay Mehrotra.

In a rare, in-depth interview recorded at his home for the June 5 podcast episode "A Bit Personal," Sanjay opened up beyond the usual industry insights. The personal conversation led him to voluntarily discuss his upbringing, family influences, and career choices.

One of Sanjay's core judgments is that AI is still in its very early stages.

In his view, as large models, Agent AI, and reasoning applications continue to evolve, AI requires not only greater computing power but also stronger "memory capabilities."

Longer context windows, larger model sizes, and ever-increasing token consumption are all driving a sustained surge in storage demand.

The essence of AI is data, and data cannot exist without storage. Therefore, storage will become one of the most critical infrastructures in the process of enhancing AI capabilities.

Meanwhile, the supply side is not adequately prepared. Sanjay pointed out that the current challenge for the storage industry is not a short-term supply-demand mismatch but a structural supply constraint. Advanced memory products consume more silicon wafers, while building new fabs typically takes three to four years, followed by a lengthy production ramp-up period.

More importantly, as technology nodes advance, the increase in storage capacity output per wafer is diminishing. He predicts that the industry's tight supply situation is expected to continue beyond 2026.

When explaining why storage technology has long been underestimated, Sanjay stated directly: "People often misunderstand memory and don't realize how difficult it is to manufacture." From physics and chemistry to materials science, and then to ensuring that every single one of trillions of bits behaves correctly during mass production, the underlying technical difficulty is immense. He believes the AI race is also a storage race, a point that has been long overlooked by the market.

From a longer-term perspective, Sanjay believes the fundamental logic behind the success of enterprises and individuals hasn't changed. Whether driving a $200 billion investment plan or steering Micron through the cyclical nature of the storage industry, the key concepts he repeatedly emphasizes are resilience, discipline, and long-termism. Investments must be based on data and fundamentals. Leaders need both a clear vision of industry trends and a deep understanding of technical details.

As he learned from his father, success requires both the resilience to persevere and the ability to seize opportunities at critical moments.

Key Insights from Micron Technology CEO Sanjay Mehrotra's Interview:

Storage is the underestimated bottleneck for AI; its manufacturing difficulty and strategic value far exceed market perception. AI is extending from a "computing race" to a "storage race." The increase in model size, longer context windows, and surging token consumption mean AI relies not only on stronger compute but also on stronger "memory capabilities." Without sufficient storage capacity and bandwidth, even the most powerful compute power cannot be unleashed.

Structural constraints on the supply side mean the storage shortage is not short-term volatility but a long-term state. Advanced memory products consume more wafers, and building new fabs takes three to four years, with capacity ramp-up equally lengthy. Simultaneously, advances in technology nodes lead to diminishing output per wafer. Under this supply-demand mismatch, tight supply is expected to last at least until after 2026.

People always underestimate the difficulty of manufacturing memory, but this is precisely the industry's deepest moat. From physics, chemistry, and materials science to design and ensuring trillions of bits are flawless in mass production, the engineering complexity is enormous. The difficulty of manufacturing memory chips is no less than any other semiconductor field, and in many ways, it is even harder.

Success comes from resilience, discipline, and long-termism, not short-term trend-chasing. Whether driving a $200 billion investment or navigating the cyclical fluctuations of the storage industry, leaders need both a clear grasp of industrial trends and a deep dive into technical details. Just like his father, who didn't give up after being denied a visa three times, success requires the resilience to persist and the ability to seize opportunities at critical moments.

Storage Is Becoming the Backbone of AI

Discussing the historical position of the storage industry today, Sanjay stated directly: "I've been in this industry for over 45 years. This is the most exciting time I have ever witnessed for the entire industry."

He further elaborated on the strategic significance of storage for AI:

"Without semiconductors, there is no AI. And storage is precisely the backbone of AI, the key foundation supporting the continuous evolution of AI."

In his view, storage's role is no longer just a component within a device; it directly carries 'intelligence' itself: "Today, storage isn't just about making devices run; it is supporting the 'intelligence' within AI, helping artificial intelligence become smarter."

With the scaling of models, the explosion of reasoning demand, and the rapid rise of Agent AI, the logic for growing storage demand appears clear to Sanjay: "As models get larger, and as reasoning demand continues to grow, as AI moves from training to inference, from data centers to the edge, the demand for storage will only increase – it needs higher capacity, better performance, and lower power consumption."

He specifically mentioned tokenomics' dependence on storage: "When you look at tokenomics, it also heavily relies on storage. As token usage grows, context windows become longer, KV cache demand increases, and the models themselves are getting larger. AI needs more than just compute power; it needs the ability to 'remember.'"

Supply Tightness Will Persist Beyond 2026

Regarding the market's most concerning issue of supply and demand, Sanjay gave a clear judgment: The entire industry's supply tightness will persist beyond 2026 and will last for a considerable period.

He explained the structural constraints on the supply side: "Building a fab takes a long time. From breaking ground to the first wafers coming out, it typically takes three to four years. Then you have a ramp-up period to gradually increase output."

More critically, the rising technical difficulty is compressing the output efficiency per wafer: "The productivity improvement from each new technology generation, meaning the bit growth per wafer, is decreasing."

Sanjay revealed that Micron had anticipated this trend as early as around 2021.

Back then, High Bandwidth Memory (HBM) accounted for less than 1% of the total storage industry, but they already foresaw that future generations of HBM would consume large amounts of silicon wafers, significantly impacting the supply landscape: "So as early as 2021, we said the industry would need new fabs built from the ground up. It's just that no one really predicted AI would explode at such a rapid pace."

Regarding the market's fear of a "return to oversupply after supply catches up," Sanjay didn't explicitly rule it out, but he emphasized that AI is still in its early stages, and long-term structural growth in demand is the basis for his confidence: "From the demand side, we are still in the very, very early stages of all this. We believe AI has a long, long way to go."

The Underlying Logic of the $200 Billion Investment: Discipline

Micron's announcement of a $200 billion investment to build a memory manufacturing ecosystem in the U.S. is one of the most attention-grabbing capital decisions in the semiconductor industry in recent years. Regarding the underlying logic of this decision, Sanjay repeatedly emphasized the word "discipline":

"Investments are never made blindly; they must be disciplined and grounded in data. You have to understand the technology, understand the applications, and understand where these applications are headed. You also need to work closely with customers to understand where they are going and what Micron's role will be."

He further explained the discipline at the execution level: "Today, we are investing in building several new fabs from the ground up. The first step is to build the shell and the infrastructure. Once these fabs are built, we will still maintain discipline when installing equipment and building actual capacity – continuously evaluating demand forecasts, assessing how much growth technological advancements can bring, and analyzing how product demand will evolve."

When asked if he ever had self-doubt, Sanjay's answer was straightforward:

"We don't have self-doubt. We absolutely believe in the memory opportunity, and that is very clear today. Of course, in our business, it's always important to remain adaptable and agile."