"The AI competition is not just a race for computing power, but also a race for storage." Micron Technology CEO Sanjay Mehrotra made this assessment.
On the June 5th episode of the podcast "A Bit Personal," Sanjay participated in a rare in-home recorded deep interview. In addition to the usual industry insights, this personally colored conversation prompted him to voluntarily discuss his growth experiences, family influences, and career choices.
AI is still in its very early stages, which is one of Sanjay's core judgments.
In his view, as large models, Agent AI, and inference applications continue to evolve, AI requires not only stronger computing power but also stronger "memory capacity."
Longer contextual windows, larger model scales, and the ever-growing token consumption are all driving the continuous rise in storage demand.
The essence of AI is data, and data is inseparable from storage, making storage one of the most critical infrastructures in the AI enhancement process.
At the same time, the supply side is not adequately prepared. Sanjay pointed out that the current storage industry is facing not a short-term supply-demand mismatch but rather structural supply constraints. Advanced storage products require more wafers, and building new wafer fabs often takes three to four years, with subsequent capacity ramps equally lengthy.
More importantly, as technology nodes advance, the storage capacity output growth per wafer is decreasing. He predicts that the industry's tight supply situation is expected to continue beyond 2026.
When explaining why storage technology has been underestimated for so long, Sanjay candidly stated: "People often misunderstand memory and don't know how difficult it is to manufacture memory." From physics, chemistry, and materials science to ensuring that every single bit behaves correctly in mass production, the technological challenges involved are extremely high. He believes that the AI competition is also a storage competition, a point that has been overlooked by the market for a long time.
Looking at the longer-term perspective, Sanjay believes that the underlying logic of corporate and personal success has not changed. Whether driving a $200 billion investment plan or leading Micron through the storage industry cycles, the key words he repeatedly emphasizes are resilience, discipline, and long-termism. Investments must be based on data and fundamentals, and leaders must both see industry trends clearly and deeply understand technical details.
Just as he learned from his father, success requires both the resilience to persevere and the ability to seize opportunities at key moments.
Micron Technology CEO Sanjay Mehrotra's interview highlights are as follows:
Storage is the underestimated underlying bottleneck of AI, with its manufacturing difficulty and strategic value far exceeding market perception. AI is transitioning from a "race for computing power" to a "race for storage." As model scales increase, context windows lengthen, and token consumption rises, AI not only relies on stronger computing power but also on stronger "memory capacity." Without sufficient storage capacity and bandwidth, even the strongest computing power cannot be unleashed.
Structural constraints on the supply side determine that the storage shortage is not a short-term fluctuation but a long-term condition. Advanced storage products consume more wafers, and it takes three to four years to ramp up new wafer fabs, making capacity expansion equally lengthy. Meanwhile, the progression of technology nodes leads to a decreasing yield increase per wafer. Due to the supply-demand mismatch, supply constraints are expected to persist at least until after 2026.
People often underestimate the manufacturing difficulty of memory, but this is precisely the industry's deepest moat. From physics, chemistry, and material science to ensuring tens of trillions of bits are error-free in design and large-scale production, the engineering complexity is extremely high. The manufacturing difficulty of memory chips is no less than any other semiconductor field, and in many ways, it is even more challenging.
Success comes from resilience, discipline, and long-termism, rather than short-term trend judgments. Whether driving a $200 billion investment or navigating the cyclical fluctuations of the storage industry, leaders need to both see the industry trends clearly and delve into technical details. Just as his father did not give up after being rejected for a visa three times, success requires both the resilience to persevere and the ability to seize opportunities at key moments.
Storage is Becoming the Backbone of AI
When discussing the current historical position of the storage industry, Sanjay candidly states: "I have been in this industry for over 45 years. This is the most exciting moment I have ever experienced in the entire industry."
He further elaborated on the strategic significance of storage for AI:
"Without semiconductors, there is no AI. And storage is the backbone of AI, supporting the critical foundation for AI's continuous evolution."
In his view, the role of storage is no longer just a component in a device, but is directly carrying the "intelligence" itself: "Today, storage is not only keeping devices running, it is underpinning the 'intelligence' in AI, helping artificial intelligence become smarter."
With the increasing scale of models, exploding inference demands, and the rapid rise of Agent AI, the growth logic of storage needs is very clear to Sandeep: "As models get larger, as the demand for inference 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 larger capacity, higher performance, lower power consumption."
He also specifically mentioned the reliance of storage on tokenomics: "When you look at tokenomics, it also relies heavily on storage. As token usage grows, the context window becomes longer, KV cache demands increase, the models themselves are getting larger, AI needs not only computational power but also the ability to 'remember.'"
Supply Constraints to Continue Beyond 2026
For the market's most concerning supply-demand issue, Sandeep made a clear assessment: The supply constraints across the entire industry will continue beyond 2026, and will persist for quite a long time.
He explained the structural constraints on the supply side: "Building a fab takes a long time. From breaking ground to the first batch of wafers, it usually takes three to four years. Then you have to continue ramping up, gradually increasing production."
More importantly, the increasing technological difficulty is squeezing the output efficiency of each wafer: "The efficiency gains brought about by each new technology generation, that is, the incremental bits each wafer can bring, are decreasing."
Sandeep revealed that Micron had already anticipated this trend around 2021.
At that time, High Bandwidth Memory (HBM) accounted for less than 1% of the entire storage industry, but they foresaw that future generations of HBM would consume a large number of silicon wafers, causing a significant impact on the supply landscape: "So as early as 2021, we said that the industry needs new fabs built from scratch. It's just that no one really predicted that AI would explode at such a rapid pace."
Regarding the market's concern about the issue of "oversupply after supply catch-up," Sandeep did not directly rule it out, but he emphasized that current AI is still in its early stages, and the long-term structural growth in demand is the foundation of his confidence: "From the demand side, everything is still in a very, very early stage. We believe there is still a long way to go behind AI."
The Underlying Logic of a $200 Billion Investment: Discipline
Micron announced a $200 billion investment to build a storage manufacturing ecosystem in the United States, making it one of the most notable capital decisions in the semiconductor industry in recent years. Regarding the underlying logic of this decision, Sanjay repeatedly emphasized the word "discipline":
"Investment is never made blindly; it must be disciplined and based on data. You need to understand the technology, understand the applications, understand where these applications are heading. You also need to work closely with customers, understand where they are going in the future, and the role Micron plays in it."
He further explained the discipline at the execution level: "Today, we are investing in building a set of new fabs from scratch. The first step is to build the factories and infrastructure. After these factories are built, as we install the equipment and establish actual capacity, we will still maintain discipline—continuously assessing demand forecasts, evaluating how much growth technological advancements can bring, and assessing how product demand will change."
When asked if there had been any self-doubt, Sanjay's response was straightforward:
"We do not have self-doubt. We absolutely believe in the opportunity in storage, and today, this is very clear. Of course, in our business, it is always essential to maintain adaptability and agility."
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
