Engineering the Bitcoin Gold Rush: An Interview with Yifu Guo, Creator of the First Purpose-Built Miner
Yifo Guo, co-founder of ASIC builder Avalon. Photos of Guo are by the author.
A month after it reached a new all-time high, the rollercoaster ride that is bitcoin continues to thrill and confound after a series of events helped propel the virtual currency to stratospheric new heights, more than doubling its market value with the digital currency now trading at over $70.
Over in Europe, the threat of financial Armageddon gave citizens new reason to consider the viability of cyberpunk alt-money. As Cypriot officials put 100 euro limits on withdrawals, the tiny Mediterranean island will soon welcome its first bitcoin ATM .
Here in the U.S. Silicon Valley venture rich-guys continued to bet on the promise of this burgeoning internet-of-finance disruption, while the department of the U.S. Treasure that combats money laundering, FinCEN, for the first time, issued guidance on decentralized virtual currencies. tacitly giving bitcoin the legal thumbs up but also paving the way for impending regulations.
And quietly in the background, a company called Avalon shipped the first ASIC-based bitcoin miners. custom-built rigs with specially-designed chips for efficiently printing the market’s hottest commodity, ushering in what can be considered the internet’s first gold rush. (Someone recently paid $20,000 for a $1,500 miner from batch two on eBay. At the time of this writing, another auction has a batch two Avalon miner going for over $19,000.)
After opening up its third batch of 600 miners for sale yesterday, customers from around the world from countries like Argentina, the UK, and even Egypt (although the majority of orders came from the big three of the U.S. Russia, and China) made sure Avalon’s units sold out in fifteen minutes. We had the chance to sit down with Avalon’s founder, Yifu Guo to talk bitcoin, mining, and the future.
Yifu Guo with an Avalon ASIC bitcoin miners at BitInstant's offices in New York
Motherboard: So how and when did you get involved with bitcoin?
Guo: It was early 2011 when some article about bitcoin popped up in my RSS feed. I don’t remember the exact article but I remember thinking at the time that this was the stupidest thing ever. It would never work. But what kept my attention was that it was open source and after a few days of thought and further research, I concluded that this was legit.
What did you do from there?
At first I just got involved with the community. I didn’t do anything presentable per se, but I participated in some feature discussions, became a regular in the forums. One of my first projects was this website that showed verified physical locations that accepted bitcoins called Bitcoin Navigator. It's still up (sometimes).
What was the community like back then?
The community was very. immature, I’d say. It was mostly tons of speculation. There was nothing really meaningful you could do back then. Mining was the central focus because of the profit incentive. And so a lot of people were just mining these things and so was I. The two long term goals then were to build services around bitcoin while also getting people to accept the concept.
Most of the core developers are idealists, much like the community now. A lot of the people in the community are definitely in it for the money, to get in early and watch the price rise. That’s essentially why it blew up very quickly (in July of 2011). There was no volume, but there was a ton of speculative hype and so a bubble formed.
This time around, there’s still a ton of interest, but there’s also more volume. Every day since reaching the new all-time high, more people are getting into bitcoin. They realize it’s not going to go away. And the people who believed from the beginning never left.
When did things start getting more serious?
I think things were always kind of serious. I would say it really took a turn in September in 2011, when the first bitcoin conference convened in New York City. I went, and all these important bitcoin insiders showed up, like Gavin Andreeson, Jeff Garzick, the entire core dev crew basically, as well as the people running real world bitcoin services. Not [bitcoin creator] Satoshi Nakamoto obviously, but a lot of early people showed up. There was this electric buzz, people talking about the projects they were working on, imagining this bitcoin-powered future. It was incredibly exciting.
"The Holy Trinity of bitcoin is the hardware support, the meshnet for powering the infrastructure free of outside control, and then bitcoin itself."
It was at this time that I started getting to know a lot of the core community. Not the speculators and miners, but people who were actually trying to build out this ecosystem around this idea. I had this sort of epiphany. What I realized was that for true bitcoin adoption, we needed hardware. Tons of software was being developed but no hardware. So I set out to make an Android-powered tablet back then, a simple, cheap tool merchants could use as a point-of-sales. That project is still ongoing, by the way.
Soon after, Occupy Wall Street happened. I knew someone who was doing tech ops and I wanted to help out, since bitcoin and Occupy are, in a way, philosophically aligned. I met someone working on this meshnet project. which was essentially this apparatus that would provide everyone in the area with internet access.
It was then that I realized that this was going to be the Holy Trinity, if you will, of bitcoin, where you have the hardware support, the meshnet for transferring information and powering the infrastructure and services free of outside control, and then bitcoin itself.
Avalon's ASIC assemblyline in Shenzhen, China, via Avalon
This was around the point that I became fully immersed in the bitcoin community. The next conference I attended, in San Francisco, was in March (there was also one in London in late 2012) and by that point, I had come to terms with the fact that this was the future. By then, I had been working on this tablet prototype for a few months.
A month later, in April, I finally had the opportunity to go to China to take the next step and figure out logistically what I had to do to get this thing manufactured. It was there that I ended up meeting one of my future co-founders. That, along with being on the ground in Shenzhen would eventually allow us to fast track our ASIC production when the time came.
"We can’t have a monopoly in mining. It's not very good for decentralization."
From an ASIC perspective, things got serious in around June or July when Butterfly Labs first released their ASIC lineup, which was a big deal. They promised to ship in October, and in a matter of days took in over a million dollars, That, on the one hand, was exciting because bitcoin is progressing technologically, but on the other hand, we were aware of the potential risks. We can’t have a monopoly in mining. It's not very good for decentralization.
So this is when we decided we had to step up and this is when developing an ASIC suddenly became a real discussion. In September 2012, we announced that we would offer 300 units if we could get the crowdfunding to do it. We announced this fully knowing that our competitor would deliver the next month, but that was fine since we simply wanted to get the chips and the technology out there in the wild.
We never intended for this to be a real business. But then lo and behold, Butterfly Labs announced that they were delayed. October came and went, November came and went, December came and went and they still hadn’t shipped. Meanwhile, we’re chugging along.
Then in late January we tried to deliver. Well, we tried. It was almost Chinese New Year and nothing happens during Chinese New Year. Fast forward to now, we’re done shipping the first few hundred units, so we’re 90 percent done with batch one.
This is an obvious question, but why aren’t you mining your own chips?
Like I said, the current reality is vastly different from what we originally had planned. We always figured we would be late to the party. Butterfly Labs and bASIC (another ASIC developer that ultimately failed to launch) were supposed to launch in succession. But due to various complications and delays, it didn’t work out that way.
Avalon 2 ASIC Bitcoin Miner
If they had delivered on their promises, they would have shipped months before our first batch. We wanted to start selling chips so people could make their own units, providing a hedge from a single entity becoming too powerful, and then move onto a new project. That was our main goal. We wanted to prevent this potential monopoly. As it turned out, we became the monopoly we tried to prevent.
via Avalon
You aren’t doing any mining at all?
Nope. Fun fact: none of the Avalon team have their own mining units (outside of test units).
If this was going to be a temporary project, how much longer will you keep the business going?
At this point, we’re going to keep making them until we run out of funding, now that bitcoin mining will continue to be “a thing” rather than not. Of course, everything changes when our competitors eventually deliver. And then we will see what the situation is and go from there. The short term plan is to push for generation two chips as soon as we obtain the necessary funds.
Batch one was $1,300 and batch two was $1,500. At a cost of 75 BTC, batch three units effectively cost
$5000. What gives?
Well, one of cofounders commented on that, and the main thing is trust, which, since the beginning of the bitcoin movement, has always been the most valuable resource above all else. The first batch could have been a potential scam, so we had to price it at $1,300 (which is the same price that Butterfly Labs has been offering).
"If we wanted to maximize our profit, we could charge much more. For us, this has always been an ideal-oriented problem, not a business-oriented one."
We’ve had plenty of opportunities, such as keeping the technology to ourselves and simply mining, but that was never our intention. So how much is that trust worth? We think it’s worth $5000 by itself. Beyond that, the calculation is based on the current difficulty rate so that break-even point occurs in a month. Which, by the way, as an investment vehicle is a fairly unique proposition. It should be mentioned also that the cost of the batch two was also 75 bitcoins. That the exchange rate has moved is out of our control.
And in the end, this was never what we wanted. If we wanted to maximize our profit, we could charge much more. For us, this has always been an ideal-oriented problem, not a business-oriented one. This is one of the primary reasons we aren’t clearly in the black.
We’re lucky that we held onto the second batch of bitcoins because if it weren’t for the appreciation, we would be very, very deep in the red, which ultimately still isn’t that big of a deal. But batch three now gives the project monetary sustainability, especially since we need to purchase an SMT machine (a pick-and-place machine for mounting devices onto circuit boards) to produce our generation two chipset.
Okay, so now you’ve started shipping your ASIC systems, but as more people get the same miners, the machines become less profitable. Does it really make sense to dive in, for such a premium, this late in the game?
The whole risk/return calculation is detached from fiat currency. The loop has been closed. It’s like any other investment, part mathematics, part faith. Difficulty is not going to skyrocket by multiples of fifty or even ten.
No one can predict the future. We don’t know if Butterfly Labs is going to ship or how fast they’re going to ship. In the end, the price is based on a break-even formula of one month. After that, if you don’t think this is a profitable endeavour, then please, for the love of God, don’t buy from us and stop complaining.
Basically, as more people get into mining, everyone makes less money.
Yes. This is the gold rush. It’s first come, first serve. In reality, the first have come and have already been served. Batch one and two was where it was at.
So then what’s your advice for a prospective miner that’s been recently seduced by all the bitcoin headlines? Is it too late to join in on the fun?
[ Points to t-shirt above ] In reality, if you have bitcoins already then, I think buying a miner is kind of cool because it’s not only about making money. If you’re here for short term profits, you should stick to day trading.
Mining is sort of magical. I remember hearing a story at a team meetup in Beijing. There was this miner who had moved from Brazil, and he basically expressed that, sure, this thing is making me some money, but what it has really done is changed my life. He couldn't believe that this was real, mining around the clock in front of his house.
"Eventually you’ll be able to hook it up to APIs where it will automatically order your groceries for you."
This machine liberates people. Suddenly, you can figure out what you want to do as a human being because it takes care of your living expenses. Eventually you’ll be able to hook it up to APIs where it will automatically order your groceries for you. It’s crazy but it’s going to happen.
If ASICs are the technological pinnacle, what’s next for bitcoin and mining?
Of mining’s technological evolution? Yes. In terms of major technological developments, it will be a while. What’s going to happen is that there is going to be a die shrink in order to reduce power consumption and increase speed based on smaller gate sizes. Eventually we’re going to reach the theoretical wall because of physics. The atoms will be too small. I don’t know when that will be hit, but the sooner the better.
If bitcoin is a $1 billion market, and it only takes less than $1 million to secure the network right now, that’s not a lot of money for someone to try and take over the mining scene. The faster the technology progresses, the more secure the network is, because it will be that much harder for a malevolent entity to mess with the system. We want to reach 14 or 10 nanometers as soon as possible. IBM/Intel is playing with 7 nanometers right now. The sooner the better so we’ll never again have this scenario where one company like Avalon essentially controls more theoretical computing power than the entire network’s hash rate. This will never happen again.
So inherent technological security?
Yes. Then we can feasibly protect the network based on predictive production and performance. At some point quantum computing might show up, but what is really next? Nobody knows.
Fresh miners being tested, via Avalon
If this was never the plan, what’s next for Yifu Guo after Avalon?
Besides ASICS? I want to finish my tablet. I don’t like to build services, I like to build platforms and infrastructure. I want to build a tablet that’s open source. Apple has done a great job of convincing us that technology is expensive when what they’re really selling is a luxury good. Everyone has an iPhone even though it’s essentially a luxury product. They’ve turned the smartphone into jewelry.
Which is fine, but we want something that works but is also cheap. We’re talking really cheap. We want to cut costs tremendously and offer a $100 tablet that is as good, spec-wise, as the Nexus 7, by cutting out all the middlemen. We have access to manufacturing and engineering, and eventually we’ll have our own factories so there will be very little overhead.
What about bitcoin? Where do we go from here?
I think bitcoin still needs more acceptance and the next major milestone is mainstream adoption. Adoption is a natural organic thing that can just happen. Bitcoin’s core features–ease of transfer and production without middlemen–will always give it value.
"How can we take bitcoin’s strengths, its cryptography, its decentralization, and how do we utilize and leverage this model?"
That’s why we don’t really care about its exchange rate. We don’t look at it like a stock, we look at it like technology. It’s a platform. How can we take bitcoin’s strengths, its cryptography, its decentralization, and how do we utilize and leverage this model? Whatever business you’re running right now, add bitcoin to the equation and you’ll see instant profits.
What do you think is the root of all the uncertainty?
People need to get their head out of the sand in terms of thinking of it as a competitor to their local currencies. We no longer live in a localized economy. We live in a globalized economy and bitcoin is designed for a globalized system.
If bitcoin is useful and it has true staying power, what’s the big picture story here? Why does it matter?
Bitcoin is an enabler based around the ideals of peer-to-peer, open source, and decentralization. That’s the future and bitcoin is the vehicle that will take us there. It takes the banks and the humans and the Federal Reserve out of the equation. Without those institutions, we can do things more freely and easily. It allows us to get things done without all of the friction, and whatever we do will be built around these ideals.
I am not a lawyer but the way I understood it, people who mine are not going to have a problem. Mining pools (organizations that allow users to contribute processing power for their share of the bitcoins) might have to deal with certain regulations, such as applying for licenses.
For individual users, it should be okay, since apparently, using bitcoins to pay for goods and services is fine. Once you have the intention of turning that into dollars or some other fiat currency, then problems might arise. Ultimately this is simply guidance. Our motto used to be, ask for forgiveness, not for permission. But recently I came to terms with an even better maxim, since we’re technically not doing anything wrong. Act and defend your position. And our position is bitcoin. We believe in it.
"It’s not about how much bitcoin is worth. The exchange rate is irrelevant. It’s about the concept."
And ultimately, I think the recent guidance is arguably good thing since it finally offers clarity. This kind of verbiage means that certain organizations can now start accepting bitcoins with more confidence knowing that the government has essentially given virtual currencies the official okay. This be a major jumping off point for mainstream adoption. Exchanges have been following the rules anyway knowing that regulation was inevitable so for most companies, there isn’t a whole lot of change.
What could the government conceivably do to bitcoin then? What’s the worst case scenario?
The worst case scenario is that they attempt to destroy bitcoin itself, and there are numerous strategies they could use, which we don’t really have to go into. But the point is, the idea will never die. Even if bitcoin dies, an alternative will arise, one that addresses the vulnerability that was previously exploited. Then you get bitcoin 2.0.
Which again, it’s not about how much bitcoin is worth. The exchange rate is irrelevant. It’s about the concept of a peer-to-peer ledger keeping system, which so far is working swimmingly. And I think the best example is BitTorrent. They’ve spent billions fighting it and failed. BitTorrent is still rampant. Meanwhile, legitimate companies are now finding uses for the technology.
Read more about bitcoin:
A Guide to Bitcoin Mining: Why Someone Bought a $1,500 Bitcoin Miner on eBay for $20,600
The world's first ASIC-based miner, via Avalon
With the price of bitcoins skyrocketing, mining is suddenly big business, so enticingly big that one wannabe miner was willing to pay a 1,333 percent premium to get his (or her) foot in the door of this wildly lucrative bitcoin bonanza. Ladies and gentlemen, welcome to the bitcoin gold rush.
The craziest part? This wasn’t an auction for a physical, working, ready-to-ship bitcoin mining machine from Avalon. which claims to be the first to develop turnkey, bitcoin-specific mining computers for sale. For $20,600 (bidding started at a reasonable $500 ), the lucky winner only received a place in line and the promise that an actual (pre-ordered) miner will be delivered sometime next month. If that sounds ridiculous, well, it’s because it quite possibly is.
But clearly there are bitcoin-savvy folks betting that paying 13 times the price of a machine will actually pay off. How did we arrive at this maniacal juncture? Was it greed? Stupidity? Or simple mathematics? For the full story, we’ll have to start from the top.
How bitcoin mining works
In order to keep a record of everything, bitcoin has a ledger known as the “block chain,” a shared database of all successful transactions. Every transaction that occurs must be broadcast to the bitcoin network and everyone connected to the bitcoin network has a copy of the block chain.
Click to enlarge. How bitcoin works, by Joshua J. Romero, Brandon Palacio & Karlssonwilker Inc.
The purpose of bitcoin miners is to verify these transactions and then add groups of transactions, called blocks, to the block chain. This process occurs roughly every 10 minutes. For every block added, the successful miner receives a certain amount of bitcoins for his troubles, plus transaction fees. The reward started out at 50 bitcoins, but it's cut in half every 4 years. Right now, you get get 25 bitcoins for every block mined.
Anyone can technically become a miner. The software is ready to download. all you have to do is contribute raw computing power, which means your main recurring costs are electricity bills. If you solve the next block, the spoils are yours. The more processing power at your disposal, the greater your mining ability.
But here’s the kicker. Built into the model is a “difficulty” metric, which is recalculated whenever 2016 blocks are added. As the speed of mining goes up (as more processing power is added to the system), the difficulty will increase proportionately to compensate in order to maintain the rate of one block every 10 minutes.
Mining difficulty increases at a predictable rate, via bitcoinX
What this means is that your ability to mine bitcoins isn’t necessarily about absolute computing power, but rather your computing power relative to other mines. Of course, in the end, that still means you’re going to want the most powerful hardware possible if you want to maximize your mining ability. But it also means that if you don’t keep pace, you’re going to be left behind. Which brings us to.
A brief history of mining hardware
Back in 2009, when Satoshi Nakamoto first birthed bitcoin, mining difficulty was relatively low, which meant that anyone could download the software and more or less start mining with only their CPU.
The next logical step was the GPU, dedicated graphics chips usually reserved for gaming. A graphics card from the likes of Nvidia or ATI offered a significant boost over Intel and AMD CPUs. For about $150, you could buy an off-the-shelf graphics card and start a fairly profitable mining business in mid-2011.
Mining profitability over time, via CoinLab
As more miners joined the party, difficulty increased, making the profit to power consumption ratio unpalatable for those used to a higher rate of return. Bitcoin's price collapse in July of 2011 only exacerbated the situation. Even if you believed in the future of bitcoin, if you spent more on your electric bill than you made from mining, you were better off just buying bitcoins.
This initiated the advent of FPGA. or field-programmable gate array, use in mining. That's a mouthful for the technical layman, but all you really need to know is that these add-on cards, which cost in the hundreds of dollars, offered comparable mining performance to GPUs while using way less power. Better energy efficiency meant higher profit margins. Eventually, any self-respecting miner was FPGA-equipped.
A mining rig hooked up with 41 Icarus FPGAs, via Xiangfu Liu
The endgame, however, was always going to be the ASIC, an application-specific integrated circuit–in other words, a chip designed from the ground up for the specific purpose of mining bitcoins. The result is a system that is not only incredibly powerful compared to anything else, it’s also exceedingly energy efficient.
ASIC also represents the theoretical limit on the hardware capabilities of mining equipment. Sure, you could keep shrinking the die-size of the chip so that it uses even less power, but even that road eventually ends. It’s simple physics: things can only get so small. Until quantum computing arrives–if it ever does–for bitcoin miners, using ASICs is the way to go.
For a while, the bitcoin ASIC was a pipe dream. Designing and manufacturing your own chip requires significant upfront investment. With bitcoin’s future still uncertain, many figured FPGAs would be the best hardware miners ever got. Then, last summer, a company called Butterfly Labs started taking pre-orders for fully functional ASIC systems for $1,299 and promised to ship in October. Another company, bASIC, started taking pre-orders soon after.
As with most things bitcoin in these early days, the whole ordeal was contrasted by wild enthusiasm and lingering fear, uncertainty, and doubt. The guys from bASIC ended up running with the money (although it appears they have been attempting to give partial refunds). Meanwhile, Butterfly Labs, after numerous delays, has still yet to ship anything, although it’s generally believed that they eventually will.
Avalon's ASIC chips
Only one company followed through. Avalon started taking orders in September, promising delivery sometime in February. That first batch of 300 pre-orders sold out within hours. By the end of January, Avalon shipped their first two units from China just before the new year's festivities, effectively becoming the world’s first ever company to produce an ASIC bitcoin miner. ASICminer, another major developer went online in February as well, although these units were never sold to the public (but you can buy shares in the company). A new era had begun.
Where we are today
Remember, the ability to mine bitcoins is based on relative computing power. As such, whoever got their hands on those first ASIC machines–which are roughly 50 times more powerful than the next best thing–would quite literally print money. That lucky man was Jeff Garzik. who was incidentally pushed to the front of the queue by Avalon for being a core bitcoin developer. It’s an open source project after all. (The other unit went to the Bitcoin Foundation .)
Garzik made back the cost of the $1,299 ASIC bitcoin miner in about a week. The remaining units from batch one were delivered by the end of February.
Having gained some credibility and silenced the trolls, Avalon started accepting orders for batch two, which totaled 600 units at a cost of $1,499 each. Batch two pre-orders sold out within 20 minutes. Those units are scheduled to complete shipping by the first week of April. But while miners in batch two will still do well for themselves, they’ll be doing less well than batch two over time as difficulty inevitably ramps up.
Which finally brings us back to our exuberant eBayer, the one who paid over $20,000 to cut in line and join the other batch two early birds. Is the worm really worth an $18,500 premium? Only time will tell. But if the price of bitcoin continues its meteoric rise, he too, will eventually mine his money back, sooner rather than later. Under current conditions, he'll break even in 50 days, with daily revenues of $434.12, according to BitcoinX. All things considered, not too bad. Granted, it's impossible to know how bitcoin will perform in that time.
The arrival of ASIC-miners, in graph form, via bitcoin.sipa.be
We do know however, that he'll be in for some stiff competition and with it, the reality of diminishing returns as more ASIC units flood the system. Butterfly Labs–rumored to have sold over 20,000 pre-orders as of a month ago–is expected to start shipping in May or June. Still chugging along, Avalon revealed it would start taking orders for batch three in the next few days.
This time around, one of the 600 Avalon miners will cost
75 BTC (the batch three price of the systems will be calculated so that break even point will be 30 days, once the difficulty resets), which comes out to over $5000 with bitcoins trading at
$70.
If that seems pricey–it is nearly five times the price of an identical unit from batch two–it’s still only a fraction of the market value. And really, there are few other businesses whose start-up costs are designed to break even in just a month. Such is the insanity of bitcoin mania.
Read more about bitcoin:
