Advanced Computing in the Age of AI | Tuesday, April 23, 2024

CryptoSuper 500 – Keeping Track of the Supercomputers that Mine Money 

Cryptocurrencies and the associated blockchain decentralized ledger technologies have grown in importance as a new asset class. They have fueled an emerging crypto economy which has become an important driver of technology development and investment decisions globally. Supercomputers, which do the heavy lifting for AI, science, and engineering, are increasingly also safeguarding digital money.

Decentralized Trust

In traditional banking, the validity of a transaction relies on establishing a single system of record. A bank is trusted to approve a transaction. Cryptocurrencies achieve the same goal in the opposite way. All transactions are broadcast to all willing computers, the more the better, and decentralized consensus is used to validate and approve a transaction. For this consensus to be trustworthy, those who participate in validating transactions are incentivized to act honestly, and are penalized for misbehavior. This carrot-and-stick approach removes the necessity for a central arbiter and enables new ways to govern the currency in a decentralized manner.

Proof of Work

The two most robust consensus algorithm based on security and value are Proof-of-Work (PoW) followed by Proof-of-Stake.

In PoW, the proverbial “carrot” for validators is comprised of transaction fees and new coins that are allocated to (mined in) each transaction block. The “stick” is the computationally expensive puzzle that must be solved before a participant is allowed to validate transactions. Because the longest chain of transaction blocks is the official record, a would-be hijacker must do better than tamper with just one transaction block. To have a chance of advancing an invalid transaction the miscreant must consistently solve this puzzle ahead of all others for several blocks. That is nearly impossible and extremely costly. It is economically more advantageous to pursue the carrot as an honest player than to risk facing the stick.

PoW algorithms tend to be more secure, but are so computationally intensive as to require supercomputing levels of resources.

CryptoSupers

Unlike traditional supercomputers engaged in scientific or analytics applications, CryptoSuper excel at cryptographic calculations, specifically some variant of cryptographic hashing functions. While the metric for traditional supercomputers is Petaflops or Exaflops (floating point operations per second), the figure of merit for CryptoSuper is Petahashes per second or Exahashes per second.

Traditional supercomputers extend their capability by clustering multiple systems into a massively parallel resource typically within a single data center. CryptoSupers also pool disparate resources to enhance their chances of winning block rewards.

The CryptoSuper 500 List

Supercomputers represent an important capability, both nationally and globally, so it makes sense to track them and analyze their evolution. Now in its fourth edition, the CryptoSuper 500 list was modeled after existing efforts that track the most powerful supercomputers for science, engineering, and AI (TOP500), the most energy-efficient supercomputers (Green500), the best performing systems for graph analytics (Graph500), and the most powerful systems in terms of input/output and storage (IO500) already exist and provide  In many ways, CryptoSuper 500 and similar lists are a barometer of important national capabilities and provide great insight into government and industrial policies.

CryptoSuper 500 is released at about the same as the TOP500 list. Like some of these lists, it does not yet include 500 entries. Perhaps someday, it will reach 500 or more players, but for now we focus on the top 50.

The Fourth Edition of CryptoSuper 500

Industry data shows that there are now over 5,500 cryptocurrencies in various degrees of circulation and use. Many of them do not use Proof-of-Work to achieve consensus, the computationally intensive approach that we favor and which qualifies it for inclusion in the CryptoSuper 500 list.

This is the first list after a Bitcoin “halving” in May, an event that takes place about every four years and reduces block rewards by a factor of two. It makes Bitcoin mining more challenging and puts upward pressure on transaction fees. As expected, this edition of CryptoSuper 500 shows that Bitcoin halving has impacted the aggregate annual economic value of crypto production. That metric now stands at approximately $5 billion, down by about a third. If Bitcoin prices move higher, as the long-term trend has demonstrated so far, it is reasonable to expect a recovery. The cost of electricity and transaction costs will also play critical roles in such a recovery.

China has the largest share of the top 50 cryptocurrency supercomputer-class mining pools.

The annual economic value (AEV) of the top 5 pools in the current list compared to the last list (published in November 2019) are as follows:

  • F2Pool, a Global pool, with an AEV of $832 million compared to $1.05 billion in November 2019
  • Poolin in the US and China with an AEV of $614 million compared to $896 million previously
  • Antpool based in China with an AEV of $492 million compared to $768 million on the prior list
  • BTC.com which is now a global pool but started in China with projected AEV of $442 million compared to $1.12 billion previously
  • SparkPool based in China with an AEV of $370 million. SparkPool came in at #10 with $234 million on the prior list

The eight coins mined with PoW protocols, and with significant value, that are included in this analysis are:

  • Bitcoin, Ethereum, Bitcoin Cash, Litecoin, Bitcoin SV, Monero, Ethereum Classic, and Zcash

The original Bitcoin continues to be the leader among all cryptocurrencies. There are over a million mining rigs around the world just for Bitcoin, adding up to a total of over 105 Exahashes per second of crypto supercomputing power presently.

Stephen Perrenod and Shahin Khan are partners at OrionX, a technology industry consulting company.

Please note that OrionX is not endorsing any particular product, company, or organization with this article.

EnterpriseAI