Impinj and Everspin Memory Chips Could Be the Next Big Thing
Memory chips are gaining importance over venerable microprocessors. Two ways to play the trend.
By Tiernan Ray Biography
Barron's
Updated May 20, 2017 1:23 a.m. ET
Pixabay
_____________________
In the world of computer chips, all glory goes to the microprocessor, and especially to the kind that Intel sells, which serves as the brains of your personal computer.
But there are many other kinds of chips in the universe of semiconductors, some increasingly more important.
This magazine argued in a 2015 cover story that memory chips may represent the most important kind of chip in years to come (“ Watch Out Intel, Here Comes Facebook,” Oct. 31, 2015). That’s because tasks that are taking center stage, such as machine learning, artificial intelligence, and the Internet of Things, continue to place greater and greater emphasis on retaining and analyzing vast amounts of data.
We recommended Micron Technology (ticker: MU) in that article. The shares are up more than 70% since then, which probably doesn’t leave a lot of upside at this point. The good news is that there are two other companies worth taking a look at that are incredibly promising, Impinj (PI) and Everspin Technologies MRAM 1.2561274509803921% Everspin Technologies Inc. (MRAM). They are both small companies that recently came public, and a bet on their stocks is not without substantial risk.
But they offer genuine innovations that have been years in the making, real breakthroughs. That could position them at the forefront of the kinds of developments we highlighted in the 2015 story.
The memory chip is, as it sounds, the repository of data. It stores the ones and zeros so the microprocessor can do something with them. For years, the microchip industry has tried to find the perfect balance between DRAM, which is fast but doesn’t retain data, and flash, which is slower and more prone to failure, but which holds data when the power goes out.
So-called nonvolatile memory chips, an in-between solution, are one of the industry’s holy grails. “Nonvolatile memory is a big deal in almost everything you do,” says Carver Mead, the Gordon and Betty Moore professor of engineering and applied science, emeritus, at the California Institute of Technology. “You have to have it,” he tells Barron’s.
Mead was the CEO of Impinj when it was founded 17 years ago and is now an advisor to the CEO, Chris Diorio, a former University of Washington professor for whom Mead served as a mentor.
What they developed is a chip that can operate on fractions of a watt of power. It is built into “tags” that can be attached to any number of objects, such as your luggage when you go on a Delta Air Lines flight. The tag can then be scanned by a wireless reader situated tens of feet away, which retrieves the identifying data on the tag. On a Delta flight, you can use an app to actually see if your stuff is on the plane. No more lost bags, at least in theory.
THE KEY IS THAT THIS PARTICULAR nonvolatile memory doesn’t need its own power source. Instead, it is activated when the wireless signal from the reader strikes the receiving antenna. That means it can be added to a vast amount of objects for pennies per piece—retail items in a warehouse, clothes on a rack, driver’s licenses, and on and on.
Impinj is interesting because it is not a chip company per se, but a systems business. It sells the nonvolatile chips, the radio-frequency circuitry, the technology for the wireless reader, and the software that makes it all work.
Diorio envisions expanding uses as semiconductor technology continues to make circuits smaller and more efficient. “We can put more and more smarts into them,” says Diorio, so that every object will be able to express all kinds of information about its nature, its capabilities, and its contents.
Impinj went public on July 21 of last year. Since that time, it is up 128%. The company is growing fast, with sales expected to rise 30% this year, and it is already profitable. Like most startups, its stock, at $40.96, is pricey, trading for 78 times next year’s projected 53 cents per share in net income. If confidence rises and the multiple gets attached to projected 2019 earnings of 82 cents, you could be looking at a $63 stock.
THE OTHER OUTFIT, EVERSPIN, is not actually that young, even though it went public last year, on Oct. 7. The technology came out of what had been the chip division of Motorola, which was folded into Freescale Semiconductor and later spun out before Freescale was sold to NXP (NXPI).
Everspin’s chip’s speed is close to DRAM, but it keeps files when power is off, like flash, though it is more reliable; flash tends to break down over time.
Today, the company sells a part that is used in industrial equipment requiring small amounts of memory. Newer parts with greater capacity are being designed into data storage systems coming to market later this year or early next. Traditionally, such computers pass memory from DRAM to flash, but there is a risk that data will be lost in between. Extra circuitry and batteries are added to the storage systems to keep the data alive, driving up cost. By replacing DRAM with Everspin’s chips, system makers can save on all that extra stuff.
“For every $10 worth of DRAM it replaces, it also gets rid of $70 of batteries and other stuff,” says Richard Shannon, who follows the company for Craig-Hallum.
As a result, “this is not a commodity chip,” says Rajvindra Gill, with Needham & Co. “This is an application-specific part that is saving companies money on their total system cost.”
“Where it gets really exciting,” says Shannon, is when Everspin can replace the combination of DRAM and flash in mission-critical computing systems such as a database. Those machines have an appetite for far greater amounts of data, which means more-expensive parts from Everspin, boosting both revenue and, presumably, margins.
The company is expected to have only $43 million in revenue this year, and it is losing money. Matthew Ramsay of Canaccord Genuity says the stock is currently worth 14 times projected earnings of 85 cents per share in 2018, or $12, above a recent $9.90. But it could have a multiple of four or five times sales if its newer markets take off, he says, which would put the stock price well above $12.
The whole history of memory-chip technologies is laced with agony and ecstasy, so this is not for the faint of heart. But then, companies with true breakthroughs can be worth the risk.
TIERNAN RAY can be reached at: tiernan.ray@barrons.com, www.blogs.barrons.com/techtraderdaily
Follow @barronstechblog
Like Barron’s on Facebook
Follow Barron’s on Twitter
barrons.com |
