Semiconductors moving through the production line in 2013. Today’s versions are even smaller. Photo by Gianluca Colla/Bloomberg via Getty Images.
Editor’s Note: MarketMinder does not recommend individual securities. The below are merely examples of a broader theme we wish to highlight.
Moore’s Law got a late birthday present Thursday: several more years (at least) of life. Physicists have long predicted the demise of Intel founder Gordon Moore’s 1965 theory, which held that the semiconductors’ computing power will double every two years as transistors become exponentially smaller. It held true for 50 years, hence its status as “law,” but shrinking transistors has become increasingly costly, and physical barriers have arisen. But a research team led by IBM, Global Foundries, Samsung Electronics and State University of New York forced scientists to stop writing Moore’s Law’s obituary: They showed the world a prototype chip two generations ahead of anything on the market, featuring transistors twice as small. If you were worried moribund innovation and stagnating technology would doom our economy and stocks over time, think again—the potential for society and markets remains as marvelous as ever.
A microprocessor is a really, really, really small integrated circuit. If you ever built a circuit in your high school physics class, think of a microprocessor as an insanely complex, miniscule version of that same principle: It is a series of transistors that conduct electricity, and they work only if your connections are seamless and intact so the current can flow uninterrupted through each component. The first circuits were hand-made, with transistors manually soldered in place and connected with metal wires. That’s fine for a science project but not realistic for computers, which required exponentially more power.[i] The solution was the integrated circuit, developed independently (and roughly simultaneously) by teams at Texas Instruments and Fairchild: All components (transistors and chip) were carved out of the same block of material, with the connectors layered on top. That breakthrough gave us the first semiconductors, paving the way for the PC Revolution. (And pocket calculators, smartphones, tablets, X-ray machines, surgical devices and so much more.)
Building faster chips requires not just more transistors, but also smaller space, limiting the distance the current must travel. That means the transistors themselves must be smaller. Same goes for the pathways and connectors—all big barriers to shrinking chips. There is also the problem of preventing overheating as currents zoom through billions of transistors crammed in a square inch or less. Overcoming these obstacles takes significant research, trial and error—not to mention developments in related production technologies.
The smallest, most powerful chips on the market have transistors 14 nanometers (nm) long—that is 14 billionths of a meter. As CNET notes, this is “7,000 times narrower than a human hair, or alternately, six times wider than a strand of DNA.” The next generation—prototypes in development at Intel and IBM—will feature transistors 10nm long. The IBM group’s prototype leapfrogs that at 7nm: “less than three times the width of a 2.5nm DNA strand.” They’ll be able to fit 20 billion of these jobbers on a wafer the size of a human fingernail.
This is a long way from hitting the market, but the potential it will generate is awesome—the prototype chip is fully 100% more powerful than the current 14nm chips. But the science behind it is even cooler, because it knocked down some of the barriers the “Moore’s Law is dying” crowd cited. The researchers were able to improve the transistors’ conductivity and speed by adding a layer of silicon germanium to the chip “wafer,” allowing them to switch on and off faster. That helps currents flow faster, which in turn makes chips faster. They also solved a problem that has long perplexed hardware developers: The need to etch ever-finer circuitry patterns on the chip. For a decade, chip designers have been limited by existing photolithography technology, which required them to etch fine patterns with a UV beam whose wavelength measured 193nm. Small! But not nearly small enough when you’re trying to cram in circuits nearly 30 times thinner. To borrow CNET’s metaphor, “it’s something like detailed finger-painting with a boxing glove.” IBM’s team solved the problem by using “extreme ultraviolet light” (EUV), whose wavelength is just 13nm.
This opens a whole new frontier. Other teams will no doubt seize the baton, using the technology to cram even more (and even smaller) transistors on a wafer over the next several years. Moore’s Law could realistically run another decade thanks to this breakthrough.
Moore’s Law will face other limits over time—the atomic barrier looms. But that is in the far future, and never underestimate human creativity’s ability to overcome such obstacles. (Even if you can’t split the atom to fit more transistors horizontally, you can always stack them and build up!) For now, we know this: For the next several years, chips will continue getting exponentially smaller and zippier.
This isn’t a Tech story—it is an everything story. Think of all the products that use computer chips today! Yes, your PC, laptop, mobile device and cloud storage—those are the obvious ones. But also: Drones. Cars. X-ray machines. MRIs. Medical devices. Coffeemakers. Videogames. TV sets (and of course DVRs). Thermostats. Watches. Smaller chips could mean amazing advances in microsurgery, cancer detection and treatment and treating hearing and vision impairment. Miniaturization could lead to things as far-fetched as transistors invisible to the naked eye in contact lenses.
Most of what we could speculate about today will sound like flights of fancy—but that isn’t the point. What matters is that technological development and human need and ingenuity always collide in unpredictable ways to create amazing things no one could have imagined. One small step in computing power can be a giant leap for all sectors of the economy, from consumer products to health care and even to finance. Hence just plain owning stocks is your way to capitalize on Moore’s Law. Not just Tech—everything that uses technology.
Forget “secular stagnation,” fretting over “too-low” R&D spending and proclamations of perpetual mediocrity—the 7nm chip smashes all that to pieces. Many grouse about R&D wavering as a percentage of GDP, but IBM’s team spent just $3 billion on this project—a project with near-limitless implications. It doesn’t take much research spending to reap huge societal benefits, both sociological and economic.
This isn’t a buy-and-hold argument, of course—the market cycle is unending, and bear markets will come. But don’t let long-term forecasts of stagnation and gloom scare you off stocks’ long-term potential. They will remain your best way to capitalize on technological advances over time.
So be cheery, raise a glass to Moore’s Law, and have a great week.
[i] For comparison, early digital computers literally weighed tons and held nearly 20,000 vacuum tubes (transistors’ predecessors).