The ASIC Mining Race and Where It is Going
People who truly understand cryptocurrencies know that mining is not just a way to make money but an essential and fundamental piece of any cryptocurrency. Mining is the mechanism that secures the cryptocurrency blockchain and continuously extends it. Cryptocurrency Mining resembles the role of a central bank in any sovereign economy. It is the process where new money is minted. The main difference is that in mining, printing new money is costly and is regulated by the network rules. In sovereign economies, printing new money is cheap and happens at the will of politicians and central bank officials.
For these reasons, many VCs started to recognize the importance of being network contributors, e.g., as miners in PoW cryptocurrencies or stakers/validators in PoS currencies. Erik Torenberg wrote an excellent piece about the transition from mining 1.0 to mining 2.0. Apparently, investing and financially supporting the projects you believe in is not enough anymore. Active participation is the new model of crypto investing. I have realized early on that projects need much more than capital to be successful. Projects need investors that are technically involved and actively supporting the network. That is why I formed a deeply technical research-focused investment fund. Historically, venture capital firms competed based on reputation and capital. Moving forward, crypto funds will be cooperating based on technical expertise and active network contribution.
As this philosophy is getting traction, I’m planning to publish a number of articles discussing the state of mining/staking in different networks. In this article, I am discussing the recent developments in the ASIC mining hardware for PoW networks focusing on the Bitcoin mining hardware. The article discusses why the competition in this specific area is difficult, who are the new players in the space and how this competition affects the decentralization aspect of Bitcoin and other PoW networks.
Brief History
People who understand the importance of mining in Bitcoin started to mine since the inception of Bitcoin. The initial goal was to keep the Bitcoin network active and secure. Eventually, when Bitcoin started to have increasing value, the Bitcoin mining and the associated financial incentives created a new economy. Miners wanted more efficient mining hardware to have a bigger reward for the same amount of energy consumed. Consequently, Bitcoin mining moved from CPUs to GPUs, then to FPGAs (Field Programmable Gate Arrays). Finally, it moved to the most performant and specialized form of hardware, the Application Specific Integrated Chip (ASIC).
ASICs are very efficient because, compared to other options, they cannot do anything except a single specific task, e.g., solving a specific hashing algorithm. CPUs, GPUs, and even FPGAs are all hardware chips that can be programmed to perform a wide range of tasks. They differ in the scope of how general these tasks are. CPUs can run any software. GPUs require specific code and are only effective when used to run a highly parallel code. FPGAs are a step further because actually they are used to design a custom hardware instead of just write some software code.
Why Are There Only a Few Mining ASIC Companies?
ASICs require a fully-custom hardware design to achieve this level of performance. The performance of an ASIC chip depends entirely on two factors:
- How skillful is the chip design
- The technology “node” used to fabricate the chips i.e. the smallest feature that could be fabricated in the chip.
Improving any of these factors require a much longer time in development and testing compared to software development. However, chip design is a more open field for competition than the manufacturing technology part. Realizing a new technology node requires new chip foundries which are very large capital investments. Therefore, this area is dominated by a few Semiconductor industry players like Intel, Samsung, and Taiwan’s TSMC. The big names in the mining industry, like Bitmain, usually don’t fabricate their own chips but instead, use the services of one or more of the major semiconductor industry players. For example, TSMC is the leading chip manufacturing partner for Bitmain.
Current Situation of The Bitcoin ASIC Mining Race
It is possible to improve the performance of mining ASICs by achieving a better chip design and using the same technology node. For example, this is the case for the new Whatsminer M10 Bitcoin miner. The M10 miner uses the same 16 nm technology as Bitmain’s S9i miner. The M10 miner uses 315 ASIC chips to generate 33 TH/s hashing rate leading to about 100 MH/s per chip. The S9i uses 189 chips to generate 14.5 TH/s hashing rate leading only to about 77 MH/s per chip. Comparing the per chip power consumption of both miners reveal similar power consumption of about 7.25 Watt/chip. While the cost/chip for the M10 miner is currently about 3 times that of the S9i, this comparison is just an indication that ASIC chip performance could be improved without migrating to a new technology node.
However, Shifting from a technology node to the following, e.g. moving from 28nm to 16nm, can lead to real performance breakthroughs. An example of that is the performance difference between Bitmain’s S7 and S9 mining chips. However, these breakthroughs are not easy to happen. For such a technology transfer to occur it needs to be economically sound. It is not enough to be just able to make chips using the new technology but it is necessary to be able to do that with a high yield. Yield is defined as the number of working chips coming from each manufacturing run or per semiconductor wafer. Low yield means a higher cost per chip which could remove the economic advantage over older chip generations. For such reasons, many chip manufacturers, like Intel, are struggling with the 10 nm and 7 nm technology nodes.
While some companies like TSMC can make simpler chips at 10 nm or even 7 nm, the economics of these systems are not suitable yet for large-scale production. The cost of the chips produced using these technologies could be higher than older technologies because of the low yield. However, a number of mining hardware companies have released products based on these technologies.
For example, the Hong Kong Based Innosilicon are currently selling their T2 Turbo SHA256 miner utilizing 10 nm mining ASIC chips fabricated by Samsung. The performance of this miner is comparable to that of the Whatsminer M10 in terms of the hash rate per watt. In addition, Japanese GMO miner has launched the B2 SHA256 miner in June 2018 based on 7 nm technology. The B2 miner was followed by the B3 updated version that could achieve 33 TH/s hash rate. Samsung has also announced in April 2018 its involvement in the fabrication of 10 nm mining chips that were speculated to be used in the Halong’s Dragonmint T1 miner.
Where is The ASIC Race Going?
While the discussions in this article focused on the developments in Bitcoin (SHA256) mining ASICs. The mining race is much bigger. All the companies that have entered the Bitcoin mining race will expand to other hashing algorithms to maximize their profit and market share. So far, Bitmain is the largest supplier of ASIC miners for other hashing algorithms like Scrypt, Ethhash, and Equihash
In addition, with a lot of cash being poured into the race, it is expected that the number of mining hardware companies will only grow. I even expect that some of the large semiconductor companies like AMD, Nvidia, and Samsung to enter the race. AMD, for example, has already recognized the importance of the mining hardware sector. Some of their GPU production partners have started selling GPU cards specifically for mining that doesn’t include the additional connectors required for gaming. It won’t be a surprise if they direct some of their teams’ design capabilities to design and fabricate ASIC mining chips.
How is The Mining Race Affecting Decentralization?
The heat of ASIC mining race has proven to be very useful for the ecosystem so far. With multiple mining ASIC companies, the distributed nature of cryptocurrencies is being reinforced. As the production of ASIC miners is now distributed between multiple players, it is not possible to have a monopoly where the manufacturer sets the prices to be the only winner.
The competition will also drive down prices and deliver more energy efficient miners. Regular users would be able to obtain relatively cheap miners which would further enhance the network decentralization and security. Additionally, manufacturers will have a bigger incentive to hit the market faster with their most recent technology to have a bigger market share. consequently, secret mining would not be very lucrative setting a fairer playing field for all network contributors.
In conclusion, the increasing interest in network contribution in general and in ASIC mining manufacturing in specific are very positive signs for the ecosystem. These are indications that cryptocurrencies are moving out of the speculation phase into the value creation phase. Building distributed networks with real distributed network participation are the base layer needed before moving into real-life use cases.
Disclaimer: Information in this article is for educational purposes only and cannot be taken as investment advice.
Note: A previous version of this article mentioned a mining ASICs company called Bittech. References to that company were removed as their miners have not been tested by users yet.
