No, this piece is not about the TV reality show that has a gruff lot of hopeful entrepreneurs blindly bidding for the contents of abandoned storage lockers. With hyper-accelerating technology creating data at an exponential rate it is getting far too big to physically store.
In 2017, more than $80 billion was spent on data centers across the country, often in the remotest areas imaginable. Bend, OR, rural West Virginia, or dusty, sunbaked Sparks, NV. Yes, they're all there. And you know what the biggest headache for the managements of many tech companies is today? A severe shortage of cost-effective data storage and the skyrocketing electric power bills to power them.
During my lifetime, storage has evolved from 1-inch magnetic tape on huge reels to highly unreliable 5 ¼ inch floppy disks, then 3-inch discs, and later to compact discs. The solid-state storage on silicon chips that hit the market six years ago was a dream, as it was cheap, highly portable, and lightning fast. Boot-up time shrank from minutes to seconds. The only problem was the heat and sitting on it when you forgot those ultra-slim designs on the sofa.
Moore's law, which has storage doubling every 18 months while the cost halves, has proved faithful to the bitter end. The problem now is that the end is near, as the size of an electron becoming too big to pass through a gate is increasingly a limiting factor. As of 2016, the world needed 44 gigabytes of storage per day. According to the International Data Corporation, that figure will explode to 460 billion gigabytes by 2025. That's a mere seven years. That's when the global datasphere will reach 160 trillion gigabytes, or 160 zettabytes. It all sounds like something out of an Isaac Asimov science fiction novel. You can double that figure again when Google's Project Loon brings online the planet's 5 million residents currently missing from the Internet.
In the meantime, companies are making fortunes on the build-out. Some $50 billion has to be spent this year just to keep even with the burgeoning storage demand. And guess what? Thanks to rocketing demand from electric cars and Artificial Intelligence (A.I.), memory grade silicon is expected to run out by 2040. All I can say is, “Better pray for DNA.”
Deoxyribonucleic acid has long been the Holy Grail for data storage. There is no reason why it shouldn't work. After all, you and I are the product of the most dynamic data storage system known to man. All of the information needed to replicate ourselves is found in 3 trillion base pairs occupying every single cell in the human body.
To give you some idea of the immense scalability of DNA consider this. One exabyte of data storage using conventional silicon would weigh 320 metric tonnes. The same amount of information in DNA would occupy five cubic centimeters weighing five grams, or 0.18 ounces! And here is the big advantage of DNA. Conventional silicon permits only two programing choices, “0” or “1”. Even with just that we have been able to achieve incredible gains in computing over the past 50 years.
DNA is made of four difference bases – adenine, cytosine, guanine, and thymine – which allow four squared possible combinations, or 16. The power demands are immeasurably small and it runs cool. Also known as NAM, or nucleic acid memory, it has already burst out of the realm of science fiction. Microsoft Research (MSFT), the University of Washington, and (IBM) have all gotten it to work on a limited basis. So far, retrieval is the biggest problem, something we ourselves do trillions of times a day, every day without thinking about it.
DNA is organic, requires no silicon, and can replicate itself into infinity at zero cost. The information can last tens of thousands of years. Indeed, scientists were recently able to reconstruct the DNA from Neanderthals who lived in caves in Spain 27,000 years ago. Yes, you can now clone your own Neanderthal. Gardening work maybe? Low-waged assembly line workers? Soldiers? Traders? I think I already know some. Look for that tell-tale supraorbital brow (click here for details).
But I diverge. If you want to make money, like tomorrow, instead of in a decade, there are still a few possibilities on the storage front.
If you want to take a flyer on the ongoing data storage build-out you might look at Las Vegas-based Switch (SWCH). The company IPO'd in October and has since seen its shares drop by 32%, which is normal for these small tech companies.
A much cleaner and safer play is Cisco Systems (CSCO), one of my favorite lagging old technology companies. After all, everyone needs Cisco routers on an industrial scale.
