By Britney Nguyen

Memory chips were once seen as an unglamorous business, but Micron is betting AI will change the game for good

Micron remains based in Boise, Idaho, and has survived multiple boom-and-bust memory cycles to become one of the world's top DRAM producers.

A little more than a year ago, Boise, Idaho, was known as a charming city boasting a tree-lined river greenbelt and a college football stadium with bright blue artificial turf.

Today, it's seen as a crucial hub of the artificial-intelligence boom.

Micron Technology (MU) was founded nearly half a century ago in a dentist-office basement in Boise. Against a backdrop of fierce competition between U.S. and Japanese memory-chip firms, twin brothers Ward and Joe Parkinson, alongside Dennis Wilson and Doug Pitman, started the semiconductor-design company focused on dynamic random-access memory, or DRAM. Today, that technology underpins inference, or the running of AI models.

Micron remains based in Boise and has survived multiple boom-and-bust memory cycles to become one of the world's top DRAM producers, alongside South Korea's SK Hynix (KR:000660) and Samsung Electronics (KR:005930). The AI-driven surge in demand for DRAM has propelled Micron's stock to the forefront of a tech-heavy stock market, as AI spending shows no sign of slowing down and memory supply looks far from catching up. As one of the last strongholds of American chip making, Micron is not only crucial to the AI boom, but to the wider semiconductor supply chain - and the U.S.'s place in it.

In April 2025, Micron's stock was changing hands for $64.72, valuing the company at $72 billion, according to Dow Jones Market Data. Since then, the stock has soared by 1,407%, to $975.56 a share at Thursday's close, and Micron is now worth $1.1 trillion - making it the 10th-most valuable U.S. company.

"The memory business historically has been a tough business, not hugely profitable, but these guys have persisted," said Willy Shih, a professor of management practice in business administration at Harvard Business School.

In the world of semiconductors, memory microchips are at the lower end of the food chain. They are used in electronics like your iPhone, helping you seamlessly switch between apps, or in your dishwasher when it recalls the appropriate cycles to run. Since the memory-chip business has largely been commoditized, companies like Micron usually could charge only slightly more for their chips than it cost to manufacture them. Sudden technological advancements, like personal computers or the iPhone, would spark fresh demand for memory chips - but within two or three years, new plants would build up supply and cause memory prices to sink.

But AI may be the biggest technological advancement the memory business has ever seen, and now Micron seems determined to break out of a cycle that last brought the company down just three years ago, experts noted.

"AI system performance is architecturally dependent on memory subsystem performance and capacity," Micron CEO Sanjay Mehrotra said on the company's latest earnings call in June. "This has given rise to more complex memory hierarchy that is providing greater differentiation opportunities for Micron than at any time in our history."

The consolidation advantage

Micron CEO Sanjay Mehrotra.

Micron's founding came a decade after Intel's (INTC) start as a maker of DRAM and static random-access memory chips, which it dominated the memory market with in the 1970s. Texas Instruments (TXN), which now primarily makes analog chips, and the now-defunct Mostek were also major players in the memory market. Three of Micron's co-founders had worked at Mostek, according to the Los Angeles Times, and the company's first contract was to design a 64K memory chip for their former employer.

In its early days, Micron sold products including sticks of DRAM called single in-line memory modules, or SIMMs, that were used in PC motherboards.

By the 1980s, U.S. companies found it increasingly difficult to compete with subsidized Japanese DRAM makers, which were able to win market share by offering higher-quality chips at a lower cost.

"Micron was founded at perhaps the hardest possible time to win market share in memory," Chris Miller, a professor of international history at Tufts University, told MarketWatch.

While Intel eventually exited the memory market and pivoted to microprocessors, Micron managed to keep a place there by "being as aggressive as anyone in being efficient," said Miller, who chronicled Micron's rise as part of his 2022 book "Chip War: The Fight for the World's Most Critical Technology."

Micron raised $300,000 in seed money from local Boise business leaders, and in 1980 it received a $1 million investment from J.R. Simplot, who had made a fortune from potatoes, for a 40% stake in the company.

Throughout the following decades, Micron was also "very deliberate about not overextending itself" during the memory trade's upcycles, Miller said, which gave it the ability to buy less-disciplined competitors during downcycles.

By the late 1990s, Japan's memory giants were losing their edge to South Korean firms, forcing companies to consolidate into what became Japan's only DRAM manufacturer, Elpida Memory.

Micron still manufactures its chips at a fab in Japan today, as well as at sites across the world - including in Taiwan, Singapore and, soon, at home in Boise.

"Micron had an interesting survival strategy in leveraging Japan and picking up the pieces in these downcycles," Harvard professor Shih told MarketWatch.

Micron eventually acquired Elpida in 2013 after the Japanese company filed for bankruptcy, giving it access to technology that, at the time, was "superior to Micron's technology," Wedbush analyst Matt Bryson told MarketWatch.

The premium on high-bandwidth memory

For most of its existence, Micron has been viewed as a commodity supplier, given that DRAM was typically made to industry standards and therefore interchangeable between companies. That also meant customers were reluctant to pay very much, and profit margins were small.

"If you're a public company and your margins are small, people don't like that," Jack Gold, principal analyst at J.Gold Associates, told MarketWatch. "If you want to invest money in a low-margin business, people don't like that either."

Not only has the AI boom changed the game by creating shortages that have allowed Micron and its South Korean peers to raise prices for DRAM, but the intense memory needs of AI have changed the technology itself too.

The emergence of high-bandwidth memory, or HBM, which involves stacking multiple DRAM chips, has offered "a longer-term favorable shift for memory," Bryson said.

Compared to traditional DRAM products, HBM is a custom solution that offers diversification between memory suppliers. Large customers are asking Micron, Samsung and SK Hynix - the world's top HBM producers - for products that better fit their applications, meaning the companies can't exactly compete on price every quarter, Bryson noted.

A lesson in supply-chain resilience

In 2011, global automakers found themselves in a crisis when Japan's Renesas (JP:6723) (RNECY) had to take its logic-chip fabrication plant in the city of Naka out of commission after it was damaged by the devastating earthquake that struck the country that March.

Renesas supplied about 40% of the world's microcontrollers for automotive engines at the time, many of which came from the Naka plant, according to the New York Times.

Recounting a visit to the fabrication plant months after the earthquake, Shih said he remembers how Japan "had mobilized the country to bring it back on stream because so many companies were dependent on it."

Executives at Japanese carmaker Toyota (TM) (JP:7203) told Shih the lesson they learned was that the silicon they depended on was coming from "geopolitically and geologically unstable parts of the world," including Taiwan, he recalled.

That led Toyota to fare better during the COVID pandemic, when competitors including Ford (F) and General Motors (GM) ran out of chips due to a habit of buying them in short cycles, Shih noted, with Toyota having amassed a large inventory following 2011.

To Shih, that highlights the importance of Micron to the wider semiconductor supply chain. The company is the only DRAM producer with cutting-edge production capabilities outside of South Korea - offering crucial diversification not only for memory chips for AI, but also for those that go into phones, refrigerators and cars. Additionally, it employs more than 50,000 people across the U.S., Japan, Taiwan and Singapore, adding to supply-chain resilience.

The company uses a lot of the same advanced chip-making technology as Taiwan Semiconductor Manufacturing (TSM) (TW:2330), Shih said, "except they happen to make memory rather than logic technology." It also "is one of the few American companies" with the capability to do 3-D packaging for HBM, he added.

Micron's headquarters in Boise also gives it an advantage against homegrown competitors, according to Gold. Not only is it a lower-cost part of the country compared to the birthplace of semiconductors in Silicon Valley, but engineers are less likely to be poached from far away, he said.

"[Forty-seven] years ago, a small team of engineers in Boise, Idaho, set out to change the world of computing," a Micron spokesperson said in a statement to MarketWatch - emphasizing that while the company has become "a global leader in memory and storage," it's stayed close to its roots.

A new type of agreement

In 2022, the memory industry became flooded with chips - an oversupply issue that led to a fall in prices in 2023 that lasted into early 2025. After the pandemic, many of Micron's end customers had built up inventory ahead of an optimization period, creating an extended downcycle, Bryson said.

"I think there was a level of disbelief that there would be a significant pickup because it felt like oversupply had been there forever," he said.

The Wall Street analyst community had been able to see spending on AI chips continuing for training, Bryson noted - but beyond that, questions revolved around how AI models would continue to progress and whether the spending on AI was sustainable. Additionally, Bryson said he thinks chip manufacturers would have thought more about adding capacity sooner if they had anticipated an uptick in AI inference fueling demand.

Micron wasn't alone in underestimating the level of demand for AI-linked products. Bryson also pointed to Intel, which found itself without enough supply of central processing units as they became more essential to inference workloads.

And although Micron was a recipient of funding from the Biden administration's Chips and Science Act, Bryson said the company didn't seem to be pushing hard on building its facilities in New York and Idaho to bring capacity online.

In an attempt to manage the boom-and-bust cycle this time, Micron has introduced a new type of long-term agreement that it has said will help smooth out demand trends from its big customers.

The company announced its "strategic customer agreements" in March that include a pricing floor, with guaranteed margins above previous cycle peaks. At the time, the company had a single five-year agreement. By June, Micron said it had secured 16 agreements.

In Shih's view, the long-term agreements are a way for Micron to "smooth out these boom-and-bust cycles that they've been a victim to in the past."

Micron's new memory-chip supply deals are also a way for the company to de-risk its investment into additional capacity, including its $100 billion semiconductor fabrication plant in upstate New York.

"Once you have that thing up and running, you need it to be loaded," Shih said, given the high fixed costs of building a fabrication plant and fitting it with chip-making equipment.

When asked for comment on its strategic customer agreements, Micron pointed to its most recent announcement with General Motors to supply memory and storage platforms.

"These strategic customer agreements are part of Micron's broader approach to strengthening supply continuity across the global semiconductor ecosystem," the company said in a release - adding that they allow for "improving planning visibility, reducing supply variability and helping ensure that critical industries" can reliably get chips.

Micron's new supply deals and the evolution of HBM could shake up and change the traditional memory cycle, Bryson said. But he added it's "still an industry where supply-demand matters, and one of the fundamental truths is you don't know exactly what your customers want."

While AI has created more demand than the industry had been expecting, Bryson said it should be remembered that memory is a cyclical business that is dictated by supply, not demand.

Traditionally, the cyclical peaks in memory would see the market move to discount earnings, he noted. With Micron's customer deals, there's some certainty around cash flow that allows for better planning and less fear of cycles on the industry side, he said, and the company doesn't necessarily have to worry that it will be losing money in the next two years.

In the long term, however, Bryson doesn't think Micron's new customer agreements mean the industry will take on a noncyclical nature. That means an eventual downturn looms on the horizon.

Part of that could be driven by the crop of chip companies, including Cerebras (CBRS) and Nvidia (NVDA), that are putting more memory straight onto an AI chip, Gold noted. That could impact the amount of external memory that chip makers need in the long run.

"That's not going to happen tomorrow," Gold said, "but it does potentially change the equation of when the next downturn comes."

-Britney Nguyen

This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.