DEFINITION of Application-Specific Integrated Circuit (ASIC)

An application-specific integrated circuit (ASIC) is a piece of computer hardware that is designed for one particular use, such as running the hash algorithms used in bitcoin or litecoin mining.

More on Application-Specific Integrated Circuit (ASIC)

ASICs have been designed to carry out a large array of applications. Smartphones include ASICs designed to make and receive calls, for example. The term was not commonly found outside of niche technology circles, however, until the advent of bitcoin and other cryptocurrencies, when ASICs were built to carry out the calculations involved in mining.

ASICs in Cryptocurrency Mining

The first cryptocurrency, bitcoin, was initially mined using a computer's central processing unit (CPU). The bitcoin whitepaper assumes that mining would be carried out using CPUs, referring to proof of work as "essentially one-CPU-one-vote." (See also, How Does Bitcoin Mining Work?)

This age of techonological naïveté did not last long. By late 2010, less than two years into bitcoin's history, graphical processing units (GPUs) had displaced CPU miners. GPUs, also known as graphics cards, are used to render images and are popular with gamers, meaning that many would-be bitcoin miners already had them, and they were easily obtained by those who didn't. GPUs are capable of processing the hash functions used in bitcoin mining much faster than CPUs.

Mining is a competitive process, in which each node repeats the hash function as quickly as possible in order to maximize their probability of finding a valid hash (one below the target). The difficulty of finding that hash is automatically adjusted by the network, depending on the total amount of hash power. The entry of GPU miners therefore made it all but impossible for a CPU ever to be the first to find a new block and earn the block reward.

The process repeated itself with field-programmable gate arrays and then, in 2013, ASICs. Butterfly Labs was the first to announce a line of bitcoin mining ASICs, followed by the Chinese company ASICMINER. Profitably mining bitcoin with GPUs quickly became impossible. Other bitcoin ASIC manufacturers as of February 2018 include AntMiner, Avalon, Spondoolies and RockMiner. Samsung announced in late January 2018 that it would begin producing ASICs for mining cryptocurrency, but did not provide details.

Bitcoin mining ASICs can only be used to mine bitcoin and other cryptocurrencies using the same SHA256 hash algorithm. Litecoin, which uses Scrypt, followed bitcoin's trajectory from CPU to GPU: litecoin's founder Charlie Lee attributes the cryptocurrency's success in part to bitcoin's switch to ASICs, which created demand among GPU miners for a new coin to mine. Eventually an ASIC was designed to mine Scrypt, and at the time of writing GPU bitcoin mining is no longer profitable. (See also, Bitcoin vs. Litecoin: What's the Difference?)

Anti-ASIC Backlash

Bitcoin's core appeal is its ability to decentralize financial transactions, removing intermediaries such as banks and replacing them with a network of miners, who achieve distributed consensus about the state of the blockchain. The inability to mine profitably using cheap hardware has led to a trend towards re-centralization, however: a few enormous operations employing thousands of ASICs control an outsized share of the network's hash power. 

A number of cryptocurrencies have used different hash algorithms in order to fend off ASICs. Litecoin was often touted as an ASIC-resistant alternative to bitcoin in its early days, but that did not last. Today there are a number of altcoins that cannot be mined using ASICs, but it's not always clear whether their algorithms are in fact ASIC-proof, or potential manufacturers simply don't see an opportunity to profit from producing an ASIC. Zcash, which uses the Equihash algorithm, is one example. 

Some developers have committed themselves to changing their coin's hash function if ASICs are built to mine them. Vertcoin, which bills itself as "the people's coin," uses a memory-intensive algorithm called Lyra2RE and promise to fork the chain to increase the memory requirements if an ASIC is developed.