The Bitcoin-Mining Arms Race Heats Up
Joel Flickinger’s two-bedroom home in the hills above Oakland, Calif., hums with custom-built computing gear. Just inside the front door, in a room anyone else might use as a den, he’s placed a desk next to a fireplace that supports a massive monitor, with cables snaking right and left toward two computers, each about the size of a case of beer. Flickinger has spent more than $20,000 on these rigs and on a slower model that runs from the basement. They operate continuously, cranking out enough heat to warm the house and racking up $400 a month in electric bills. There isn’t much by way of décor, other than handwritten inspirational Post-it notes:
“I make money easily,” one reads.
“Money flows to me.”
“I am a money magnet.”
Flickinger, 37, a software engineer and IT consultant by trade, doesn’t leave the house much these days. He’s a full-time Bitcoin miner.
Bitcoin is the digital currency that thrills nerds, inspires libertarians, and incites the passions of economists who debate the value of money made from nothing but ones and zeroes. Devotees watch the fluctuations of Bitcoin’s price with a fanaticism typically reserved for college football scores. Alternative currency startups are being lavishly funded by venture capitalists while visionaries gush about the world-changing possibilities of money free from government control. Silicon Valley is the natural center for Bitcoin mania. An advocacy group named Arisebitcoin recently put up 40 billboards around the Bay Area with messages such as: “The Revolution has started … where do you stand?”
As with an actual precious metal, Bitcoins are in limited supply—they must be “mined.” Unlike with precious metals, this mining is done purely by computer. Miners set their machines to run a series of complex calculations that tally up and certify all the transactions of other Bitcoin holders around the world. If the miner’s computers complete these calculations and solve a complex mathematical puzzle before anyone else, he earns about 25 Bitcoins as payment. It’s a nice haul: With the price of each Bitcoin nosing up near $1,000, that’s $25,000 for 10 minutes or so of work. For the moment at least, miners are the rare grunts who can also get rich.
Over the past six months the price of a Bitcoin has shot up, dived, shot up again—and kept on rising, making Bitcoin mining one of the most frenzied corners in technology. Engineers are racing to design and build unique chips that crunch Bitcoin algorithms at high speeds. Thousands of entrepreneurs like Flickinger are betting that their ability to harness these complex computer systems will make them rich—and maybe create a new world order along the way.
Flickinger is part of a group called Give Me Coins that pools together its computing power to have a better shot at verifying transactions and solving the cryptographic puzzles first. He spends much of his time chatting online and monitoring his account on Give-Me-Coins.com, which shows the group’s progress and his share in the work. His setup currently accounts for about 10 percent of its total computing muscle.
One of the first things he does in the morning is check the temperature of his mining rigs. Processing chips calculate faster when they’re hot, but not too hot—the optimal temperature is around 73.5C (164.3F). If it goes much higher, the machine can overheat and malfunction, but Flickinger likes to push the limit. Occasionally, he says, he stuffs the air holes of his machines with paper to bring up the temperature. “There is serious money in this,” he says, noting that he’s earned 100 Bitcoins over the past few months from mining and other transactions. He estimates his two fastest computers will earn him $150,000 each this year. “It takes up a lot of time, but I have no kids. I have no life. I have a cat.”
The Bitcoin system was introduced in 2008 by a shadowy figure who went by the name of Satoshi Nakamoto. To this day no one knows if Nakamoto is a man or a woman or some sort of cabal, though few deny the ingenuity of the creation. Digital currencies had been devised before—DigiCash and Bit Gold, for example—and never took off. Nakamoto started with the best ideas behind these earlier currencies, added a few of his own, and wove it all together with technical elegance. The biggest achievement was solving a long-standing problem of borderless digital money: With no government oversight or central database to track transactions, how do you prevent fraud?
Nakamoto’s solution has two important parts. The first is a public ledger, shared across the Internet, which is a way to validate all Bitcoin transactions and eliminate counterfeiting. Every time someone makes a purchase or exchanges Bitcoin for another currency, that transaction is broadcast to anyone running open-source Bitcoin software. Some users run a highly specialized version of the software, which brings us to the second part of Nakamoto’s solution: mining. Bitcoin miners are, in a sense, both maintainers of the public ledger and expanders of the virtual money supply. One more twist: The Bitcoin system makes sure the money supply expands gradually. The maximum number of minable Bitcoins worldwide is currently 25 every 10 minutes. It’s like a worldwide math competition that resets six times an hour. There are a total of 21 million possible Bitcoins; about half are currently in circulation. The last will be mined in about 2140 at the earliest.
Nakamato’s puzzles—akin to searches in a digital haystack for fiendishly long numbers preceded by strings of zeroes—were easy to solve at first. Miners used standard laptops and desktops to complete calculations and earn their Bitcoins. The currency was mined and traded in small volumes and relative obscurity. It garnered attention as a popular method of payment on the online black market Silk Road, a bazaar for illegal drugs, and for the enthusiasm of high-profile backers such as Cameron and Tyler Winklevoss, the twin brothers better known as Mark Zuckerberg’s legal adversaries. It’s accepted as payment by online dating service OKCupid, at a cafe in the Netherlands, and by Richard Branson’s Virgin Galactic. There’s now a Bitcoin ATM in Vancouver and another one planned for Hong Kong.
As the currency has gathered momentum, miners have piled in. But, because Nakamoto’s puzzles are designed to get more difficult over time, solving them requires ever-escalating computing capacity. It’s an ingenious trick, like a Rubik’s cube that gets more complicated as more people try to solve it. By 2010 the puzzles grew too difficult for ordinary desktops. Miners needed souped-up systems with fast graphics processors, typically used for playing high-performance video games or conducting intense scientific research, or they needed to distribute the problem among dozens of computers at once.
It’s not clear whether Bitcoin’s inventor, who disappeared from the Internet in mid-2010, was prophetic enough to imagine what happened last year. The value of Bitcoin exploded, with the price of an individual coin jumping from $100 in July to $200 in October to more than $1,000 as of this month. Economist Paul Krugman, among many other critics wearing the dour goggles of history, argued that it was little more than the latest speculative mania, no less ridiculous than the Dutch fever for tulips in the 17th century. But the miners showed up anyway. And an industry of microprocessor manufacturers has sprung up to serve them. No matter what the fate of Bitcoin, one consequence of the race to mine it will probably be faster chips.
In July 2013, an anonymous geek going by the name Bitfury proposed a chip design tailored to the problem of mining Bitcoins. Little is known about Bitfury; collaborators say he’s a brainy Ukrainian now living in Russia who taught himself microprocessor engineering and designed his chip by hand at his kitchen table. Dave Carlson, a serial entrepreneur from Seattle, says he first met Bitfury in an online forum devoted to Bitcoin. Backed by a Polish investor, Carlson and Bitfury started a company called MegaBigPower, selling computers based on Bitfury’s chips for as much as $11,000 each. Carlson says he’s never actually seen his partner in person, but they’ve formed a frenetic, virtual working relationship. “We had a Skype session that never closed for two months,” says Carlson. “It was like a war room.”
Carlson also operates what he says is North America’s largest Bitcoin mine, an array of Bitfury-designed rigs in a warehouse. Inside the facility, which is in the Pacific Northwest—he asks that the exact location remain undisclosed—row after row of Bitcoin computers sit on cheap metal racks he bought at Home Depot. Industrial fans on the floor keep the machines cool. Carlson says he generates 100 to 300 Bitcoins a day, and he’s hanging on to them because he believes their value will only increase.
Carlson plans to sell sophisticated Bitfury arrays to other Bitcoin enthusiasts, charging upwards of $1 million for a mine that can mint $150,000 in Bitcoin a day. He needs the money. He’s still recovering financially from his previous venture, an advertising firm that collapsed, and is so strapped for conventional money he may lose his house. Still, he’s not willing to cash out of his mined currency. “I am riding this Bitcoin wave,” he says.
Eduardo de Castro runs HashFast Technologies, a mining startup founded in early 2013. Sitting in a cafe in Mountain View, Calif., in September, de Castro, too, displays feverish signs: He had a house fire the night before, which he mentions dismissively only after being asked about the bandages on his arms and face. His company is sprinting to design, manufacture, and sell specialized high-performance computers to Bitcoin miners. At the heart of the computers made by HashFast and similar startups such as KnCMiner and Butterfly Labs is a chip known as an ASIC, or an application specific integrated circuit. Unlike the general purpose Intel chip powering a PC or an ARM processor in a smartphone, an ASIC typically performs a single function extremely well. A security company might make an ASIC to speed up the encryption of data, for example. For Bitcoin-mining toolmakers like HashFast, the goal is to make an ASIC that is precisely tuned to crunch Nakamoto’s algorithms. “It would take 70,000 of Intel’s fastest chips to match one of ours,” de Castro says.
A typical company might take a year to 18 months to design and manufacture such a custom chip. De Castro and his partner, Simon Barber, a former engineer at the Palo Alto Research Center, tried to pull it off in a few months. They hired a team of 20 engineers and consultants and hunkered down for weeks in the offices of a local chip design consulting firm. Bills for those services ran into the millions, though HashFast had raised only $600,000 from friends and family. (Several family members balked at investing in what they called “Monopoly money,” de Castro says.) The HashFast partners raised the rest by preselling $15 million worth of mining rigs on the idea alone, without even a prototype. Naturally, the company took payment in Bitcoin.
It was a perilous move. Customers, for example, could pay the equivalent of $5,000 upfront for a HashFast machine. As the price of Bitcoin tripled during the year, they discovered they had paid $15,000. Worse, Bitcoin’s increasing value drew more interest to mining, which made the field even more crowded and lowered the odds of solving the underlying puzzles. De Castro is confident his computers will still be profitable for customers and cheerfully calls his strategy “aiming the car straight at the biggest wall we could find and gunning it.” HashFast promised delivery of its systems by the end of the year and missed its own deadline. De Castro, still optimistic, cites production delays but says they are close to shipping units. Meanwhile, the company faces angry customers, demands for refunds, constant complaints on Bitcoin bulletin boards, and threats of a lawsuit.
In August, Austin (Tex.)-based CoinTerra announced two Bitcoin computer designs, which it dubbed GoldStrike and TerraMiner. So far the company has $20 million in presales. Ravi Iyengar, a veteran chip designer, runs the startup. Iyengar has worked at Intel, Nvidia, Qualcomm, and, most recently, Samsung, where he led a team that developed chips for the company’s phones and tablets. After hearing about Bitcoin, Iyengar quit his job to use his experience to outrace other Bitcoin-mining startups. Iyengar has watched Intel, IBM, Apple, Samsung, and other heavyweights slug it out in the chip business for years. “No arms race in the history of the chip industry even comes close to this,” he says.
The most secretive of the new mining companies is Silicon Valley-based 21e6; its name refers to the scientific notation for 21 million, the maximum number of Bitcoins. According to regulatory filings, the startup raised $5 million in April to build what’s believed to be one of the fastest mining chips in the world. Among investors are the Winklevoss twins; Marc Andreessen and his venture capital firm, Andreessen Horowitz; early Tesla Motors backer Bill Lee; PayPal mafia member David Sacks; and Naval Ravikant, founder of AngelList, a social network for investors and entrepreneurs. (Bloomberg LP, the parent of this magazine, is an investor in Andreessen Horowitz.)
Balaji Srinivasan, a former lecturer of computational biology at Stanford University, is one of the founders of 21e6. He also runs a Bitcoin club at Stanford for students and faculty and last month became a partner at Andreessen Horowitz. Srinivasan made headlines in October when he suggested at an industry forum that Silicon Valley secede from the U.S. He now says he was misquoted and meant only that avid technologists should forge their own Internet-centric communities. “Imagine a society of Inverse Amish that lives nearby, peacefully, in the future,” he wrote recently in a blog post. “A place where Google Glass wearers are normal, where self-driving cars and delivery drones aren’t restricted by law, and where we can experiment with new technologies without causing undue disruption to others.”
Srinivasan says Bitcoin would be useful to the Inverse Amish. In an interview, he calls the currency “the missing piece of the Internet” that will make transactions frictionless. Yet when he’s asked about 21e6, Srinivasan goes silent. A PR representative interrupts to say that the company is in stealth mode. People familiar with 21e6, who asked not to be named because they weren’t authorized to speak on the matter, say the company plans to keep its superfast computers for itself, just as a hedge fund might trade on its own proprietary algorithms. It wants to build the world’s largest Bitcoin mine—a virtual money printing machine.
Supermine projects like 21e6 wrap up all the promise and danger of the libertarian sensibility that underlies Bitcoin. For HashFast and other miners, trying to corner the market in the world’s most successful virtual currency is a Dr. Evil-patting-the-cat move. Bitcoin true believers will tell you they aren’t—or aren’t completely—about the money. They dream of building a system free from the narrow interests of governments or the wealthy, allowing individuals greater freedom to move their capital around, whether it’s to avoid credit card fees, shop anonymously, or evade repressive regimes.
The fear is that an organization with piles of capital and not much idealism can buy enough computational might to corner the market and box out the individual miner. That may already be happening: Websites such as Bitcoin Watch that track the total computing power of miners have started to show large, mysterious spikes in capacity.
Even some Bitcoin entrepreneurs think mining has become a sucker’s game. Fred Ehrsam is a former Goldman Sachs trader and co-founder of Coinbase, a Bitcoin startup making wallet software that allows people to trade and store Bitcoins, and which recently raised $25 million in venture capital. Ehrsam is committed to Bitcoin but pessimistic about underfunded prospectors making any money. “This is very much a fad that is going to die soon, if it’s not even dead already,” he says. But that’s not the same as saying individual mining will end. He suggests that the next generation of miners might run their computers for ideological purposes—to support the currency and be a disruptive force in global finance—even if doing so has become unprofitable.
“Mining was supposed to be a democratized thing, but it’s now only accessible to the elite of the elites,” says Chris Larsen, CEO of Ripple Labs, which has introduced a virtual currency called Ripple. It’s similar to Bitcoin but without the mining. (The company gradually hands out increments of the currency to supporters.) “Hordes of brilliant engineers are raising money for mining equipment that regular folks can’t compete with,” Larsen says.
For idealists not swimming in startup capital, it’s still possible to join the Bitcoin rush, mostly by adding power to a distributed array of processors. Craigslist is full of miners selling their old rigs, and there are peripherals that cost about $250—they look like USB thumb drives, plug into standard PCs, and are mostly ineffective. Online calculator sites like BitcoinX let prospective miners enter processor speed, current Bitcoin exchange rate, electricity costs, and other variables to figure out whether their investment makes any financial sense. Most of these calculators suggest that even miners with older, specialized Bitcoin machines can still make a little money as long as the price for a Bitcoin is above $700. Members of mining groups are rewarded according to the amount of work they contribute.
That’s basically Flickinger’s arrangement with Give Me Coins. The collective has dozens of members scattered around the world and successfully mines Bitcoins every few days. Flickinger says he’s uneasy about the volatility of Bitcoin prices, which dropped 20 percent in the month after Beijing said it was banning Chinese payment processors from working in the currency. It topped $1,000 again in January when the game company Zynga announced it would accept Bitcoins. He’s also tired of the scammers, and there are a lot of those. He recently paid one huckster on Craigslist $14,000 for two rigs that never arrived and now keeps an eye out for similar cons. “If I see anyone doing this kind of thing, I will warn people,” he says. “I’ve been doing my part.” He’s still mining, though, doing his daily server ministrations, redlining his chips above 73.5C, and writing himself inspirational Post-its. “I don’t fully understand how Bitcoin works,” he says. “But then, I don’t get Miley Cyrus either.”