Could ASICs give bitcoin mining its PC moment?

Personal computing thrived as an open system. Mining can, too.

The relatively short history of bitcoin mining is, in a lot of ways, a common story across mature industries–one that begins with general purpose hardware and moves to increasingly specialized machines. At each choke point, the ‘what’ becomes more powerful–but also more expensive and harder to acquire–and the ‘who’ becomes a smaller and smaller pool of participants.

While it may be a common story, it’s not the only story–not by a long shot. In fact, the opposite is also true. In the early days of personal computing, for instance, it was access to hardware and an open ecosystem that made possible the kind of explosive innovation that has given us much of today’s world as we know it. As we think about what the future of bitcoin mining holds, it can be helpful to look backwards–both at the evolution of mining and the example of parallel industries who have wrestled with their own forces of centralization.

From CPU to GPU, solo mining to pools and farms

In the early days, all it took to mine bitcoin was a PC. The same general purpose CPU that ran your circa-2009 word processor and graphic design program and web browser and music player could also very effectively mine bitcoin. And since pretty much anyone could get their hands on a CPU-having computer, this meant that pretty much anyone could be a bitcoin miner. For a while, that was true, even though “pretty much anyone” translated to a relatively small group of hobbyists and tinkerers.

As more people mined bitcoin and hashrate grew, mining itself became more difficult–an algorithm-enforced feature of the bitcoin network since its inception. 

The natural response to increased difficulty was to point more powerful machines at the problem. Initially, that meant graphics processing units (GPUs)–a more specialized processor perhaps best-known in gaming and other graphics-heavy applications. Widespread GPU usage continued to push up network difficulty–which made mining harder still for CPU users. Some quit mining altogether, while others banded together in an early form of today’s mining pools. Around the same time, mining started to see early versions of mining farms, running racks upon racks of graphics cards to mine for profit.

ASICs centralize mining via hardware, supply chain limitations

While GPUs reigned supreme for a time, application-specific integrated circuits (ASICs)–powerful, purpose-designed chips made specifically for bitcoin mining–offered massive efficiency gains and have ultimately supplanted GPUs and other less-specialized mining hardware. 

But these efficiency gains have come at the cost of centralization in mining itself.

Where CPUs and GPUs are widely available, ASICs are not, for a few practical reasons. The first is engineering complexity. From design and testing, to fabrication and packaging, the process of producing ASICs often starts years before any chip leaves the foundry. Just tapeout, the last stage before production, is itself extremely expensive and potentially precarious–one mistake can cost tens of millions of dollars. ASIC production is hugely capital intensive up front, which limits the field of who can even consider getting into production pretty drastically. Even some companies whose names are synonymous with silicon have passed on chip-making for bitcoin mining–likely due to a combination of cost, complexity, and lack of conviction in the industry’s future.

Assuming a company can afford to fund an ASIC program, ASIC designers, too, are a rare talent pool in very limited supply, and the entire supply chain is highly specialized. Add to all of this the fact that only a handful of chip fabrication plants exist–the fact that, even if you design an amazing chip, production relies on your ability to capture some of an advanced fab’s very limited capacity–and it becomes clear very quickly just how centralized mining hardware supply is. 

Bitcoin mining at a crossroads

Today, bitcoin mining sits at a kind of crossroads, between an industry centralized in the hands of a few, and one that is open and competitive and decentralized–a version that is arguably much more aligned with the spirit of bitcoin itself. The key to future decentralization, as it has been in the past, is at the chip level–diversifying supply, and putting the tools for innovation in the hands of the many rather than the hands of the few. Take, for instance, the PC:

In the 1970s, the first CPU chip hit the market. About as big as a fingernail, it packed about the same computational punch as computers that once filled entire rooms just a few years prior. It also laid the groundwork for the modern digital economy, making the PC itself possible and providing the spark to too many technological advances to list. With this tiny chip came the transformation of computational power from something that only large companies and governments could harness, to something that individuals and tinkerers and would-be inventors the world over could put to countless new uses. 

PCs thrived not because suppliers limited access to hardware in a closed system, but because hardware was accessible in an open system. Many components were made available and drivers and operating systems were made open source. All of that together meant fertile ground for innovation–for multiple manufacturers to overclock and tinker and compete. And while some PC vendors took a different approach–namely vertical integration optimized for control and limiting flexibility–very few of them exist today, and the ones that do rarely find themselves at the forefront of performance or innovation.

For future decentralization, make core tech open and accessible.

In mining, it’s not hard to see the choice before us: between limited access to hardware and continued centralization, or open access to hardware, decentralization, and ultimately innovation. At Proto, we believe that bitcoin is best served by the latter–by putting the right tools into the hands of builders and developers and tinkerers–and that open innovation ultimately gets us a better, fairer, and more resilient network.