Even if you’re not a gadget geek, you likely know whether your laptop is powered by an Intel chip or one from a competitor like AMD. The sticker plastered next to your keyboard won’t let you forget. But even if you know your Ryzens from your Ice Lakes, you probably don’t put much thought into who makes the memory chips that store your data and keep your laptop and smartphone working. There’s a decent chance at least one of your gadgets includes memory made by a company called Micron Technology.
Boise, Idaho-based Micron is one of only three outfits that still make DRAM, the chips that provide short-term memory in personal computers, smartphones, tablets, and other devices. Micron, which also sells products under the brand name Crucial, is the smallest of the three with a market share of around 21 percent, but it’s the only one based in the US. Micron also makes another kind of memory, called flash, where it faces more competition and holds about 13 percent of the market.
That’s helped Micron build a pretty good business, thanks to strong demand for memory for everything from smartphones to cloud computing servers. The company generated $20 billion in revenue last year, putting it among the world’s five largest semiconductor companies, ahead of better known companies like Nvidia and Texas Instruments. Micron shares have more than doubled in the past year, giving the company a market value approaching $70 billion.
Not bad for a company few outside the chip industry have heard of, based far outside traditional high-tech hubs.
With success comes challenges, though. Memory chips are considered commodities with little differentiation among them, making competition notoriously brutal, and marked by boom and bust cycles. Micron’s two larger rivals, Samsung and SK Hynix, are part of better-heeled, more diversified Korean conglomerates. Now, China plans to throw its awesome manufacturing capability and the resources of the world’s second-largest economy into the game.
To survive, Micron is counting on continued growth as computing power is built into more devices, from TVs to cars and robots. “We believe artificial intelligence and machine learning and autonomous driving applications will present huge areas of opportunity for Micron,” CEO Sanjay Mehrotra tells WIRED.
Mehrotra was a cofounder of the flash-memory maker SanDisk who joined Micron last year after SanDisk was acquired. Key to his strategy is developing new technology that can distinguish Micron from rivals. Prime example: a new type of memory called 3D Xpoint that Micron developed with Intel that is intended to bridge the gap between DRAM and flash memory.
DRAM, short for dynamic random-access memory, loses most of the data it stores when power is shut off. Flash memory can store data long term, and is much cheaper than DRAM, but it’s slower. So most computing devices use both: DRAM for short-term memory, and flash (which has largely replaced mechanical disk drives) for long-term storage. Micron pitches 3D Xpoint as faster than flash but cheaper than DRAM, which could be a boon for applications that depend on large data sets, such as artificial intelligence. Most important, says Objective Analysis analyst Jim Handy, it’s patented technology that customers won’t be able to get anywhere but Intel and Micron.
From Potatoes to Silicon
Micron was founded in 1978 by three former engineers for Texas-based memory maker Mostek, along with one cofounder’s twin brother. The team spent their early years doing consulting work for Mostek out of the basement of a dentist’s office in Boise.
Micron lost its contract after United Technologies acquired Mostek in 1979, so the team decided to start making its own DRAM chips. That meant competing not only with Mostek, but with Intel, which pioneered DRAMs.
Instead of relocating to Texas or Silicon Valley, the company raised money locally, most notably from potato tycoon J.R. Simplot, who made a fortune selling frozen french fries to McDonalds. Micron’s chips quickly found their way into the first wave of personal computers, such as the Commodore 64, and Micron went public in 1984.
The timing wasn’t great. Japanese companies were flooding the market with cheap chips, leading seven North American semiconductor companies, including Intel, to exit the DRAM business by the end of 1985. In response to Japan’s aggressive moves in the semiconductor industry, the Reagan administration imposed tariffs on Japanese computers, power tools, and televisions. But additional competition from Taiwanese and South Korean companies brought more chips, pushing DRAM prices down.
Analysts say Micron survived largely because of its ability to hone its manufacturing technology, reducing its costs even when chip prices are low. Memory chip prices tend to soar when new technologies are developed, which slows down the manufacturing process and decreases supply, explains IDC analyst Shane Rau. Once the manufacturing kinks are worked out, prices drop as chip makers increase supply.
“The most important thing is that they’re willing to stay in the market even during downturns,” Handy says. “Micron sees downturns as an opportunity to buy competitors for pennies on the dollar.” In 1998, for example, it bought Texas Instruments’ memory business for $800 million; in 2012, it bought Japanese chipmaker Elpida Memory for $2.5 billion.
“One thing is we were willing to take some pretty significant risks and bet big on our technology,” says Micron executive VP of technology development Scott DeBoer. “The Elpida Memory acquisition looks brilliant in hindsight, but at the time it was controversial.”
Snapping up competitors also helped the company improve its technology and its manufacturing processes. “We consolidated all the best memory talent in North America,” DeBoer says.
Micron’s tenacity has paid off, at least for now. There’s a shortage of both flash and DRAM, so prices have been high. With only three significant companies vying for the DRAM market, prices are less volatile. And thanks to the rise of smartphones and other mobile devices, Micron’s sales are no longer dependent on trends in the personal computer business. Thursday, Micron reported record quarterly revenue of $7.4 billion and net income of $3.3 billion.
But Handy warns DRAM is still a cyclical business, and another downturn is inevitable. “When things are bad, people think they will never be good again, and when things are good they think they’ll never be bad again,” he says.
China is the biggest looming threat to the memory industry’s established players. Today, China imports the majority of its chips and exports very few. But the country is trying to change that through through its $20 billion Integrated Circuit Industry Investment Fund. If China can flood the US with cheap chips, it could be the 1980s again for the memory industry.
Competitors like Samsung and SK Hynix can subsidize losses from their memory divisions with sales from other divisions, but Micron has nothing else to fall back on. But DeBoer says its focus on the memory market is also an advantage. “When our backs were against the wall, we had to make memory work because there was nothing to fall back on,” he says.
That’s going to mean developing new memory technologies, such as 3D Xpoint. The new technology comes with a catch: Even if 3D Xpoint is successful, it could cannibalize Micron’s DRAM sales, because applications wouldn’t have to depend on as heavily on short term memory. DeBoer says that’s a risk Micron is willing to take. “We’re a memory company, if something is going to disrupt DRAM, we need to be the ones to do it,” he says.