Data centers and crypto: Adapting for the future

The following is a guest post from Shane Neagle, Editor In Chief from The Tokenist.

The steady Bitcoin trickle into the mainstream consciousness since 2009 mainnet launch had many cascading effects. First, it served as a revelatory vehicle by exemplifying the nature of money; why it should be outside of central banking, and why fixed supply is important for the valuation of money.

Second, Bitcoin sparked an entire crypto industry, further making the case for decentralized financial services that eliminate gatekeepers in favor of smart contracts enforced by blockchain networks. As this $2.2 trillion sector develops, banks are further poised to lose their role as trusted intermediaries.

Third, data center infrastructure is becoming more important than ever. Whether home-based or as large mining operations, crypto infrastructure needs reliable high-performance computing resources, storage capacity and memory alongside fast networking to maximally reduce blockchain latency.

In fact, data centers are so critical that an entire knowledge field emerged to balance power requirements, cooling solutions, server density and crypto hosting location. When these factors come together, crypto needs to forge an indelible mark on the data center design itself. Let’s explore how.

The Critical Role of Data Centers in Crypto Infrastructure

In the early days of the internet, broadband connection was rare. This necessitated local resources within businesses and institutions to be used for data storage and management. By the end of 2000s, broadband infrastructure had become sufficiently ubiquitous to start supporting cloud computing.

In other words, data centers were being delocalized into remote, scalable, on-demand server clusters. The ability to eliminate on-premise infrastructure and host data and apps remotely drastically cut upfront capital expenditure. Of course, this ultimately benefited Amazon Web Services (AWS), Microsoft Azure and Google Cloud as the data center triumvirate that powers the bulk of today’s digital landscape.

However, securing blockchain networks exerts an entirely new load layer. Because these digital ledgers facilitate real-time transaction processing, between multiple nodes to verify them, extra CPU, GPU power and RAM is needed to minimize congestion and latency.

And if there is a sudden spike in blockchain network traffic, this too requires resource redundancy. This is why both AI and blockchain-oriented data centers have been transitioning from traditional client-to-server architecture (north to south) to spine-and leaf architecture (east to west).

The spine-leaf approach makes for a non-hierarchical design that allows data to flow horizontally between servers. This is critical for blockchain networks, as each node directly communicates with other nodes without having to go through a congestible central point.

Therefore, a spine-leaf architecture alleviates bottleneck and single point failure potential. Because this mirrors the spirit of crypto decentralization and peer-to-peer (P2P) communication, spine-leaf data centers have become the new standard for blockchain reliability and security.

Energy Consumption and Efficiency Challenges

As blockchain networks need greater compute power to validate transactions and execute smart contracts, so is there greater need for energy consumption. By 2022, blockchain networks have already carved a significant proportion of data center electricity demand.

According to the International Energy Agency (IEA), the data sector servicing the crypto industry globally consumed 460 TWh in 2022, which is forecasted to more than double by 2026.

Image credit: IEA (data network center consumption is excluded)

For comparison, France consumed 447 TWh annually in 2021. These trends clearly point to a reliable source of power, which is why Microsoft saw fit to make a 20-year deal with Constellation Energy to restart Unit 1 nuclear reactor in 2028.

In Europe, the European Commission even designated Small Modular Reactors (SMRs) as “green” to balance decarbonisation efforts with increased electricity demand. But raw power capacity is only the beginning of scaling.

To make crypto-oriented data centers more efficient, they are moving closer to power plants. This is best exemplified by Bitcoin. The primary cryptocurrency uses a proof-of-work algorithm to secure the network, effectively anchoring Bitcoin in the physical world of energy and hardware assets.

This is what ultimately gives Bitcoin its value as decentralized money and global transfer of wealth. In essence, Bitcoin represents digital energy. But because power is lost over long distance electrical transmission, due to copper/aluminum resistance, it would be wasteful to erect crypto data centers just anywhere.

Rather, they should be as close to power plants as possible to minimize transmission loss. Case in point, the New York state power plant bypasses state level network by directly plugging in thousands of servers. Likewise, Ward Roddam, mayor of Rockdale, Texas, recently made the case that Bitcoin mining can revitalize communities by investing to siphon excess energy and stabilize the electric grid with flexible load demand.

“Riot Platforms is building a state-of-the-art facility in Corsicana, which will be one of Navarro County’s largest employers. The mine could bring $1.4 billion in taxable purchases and over $115 million in wages over the next decade.“

Another crypto mining company, TeraWulf, has been building its Nautilus Cryptomine adjacent to the Susquehanna nuclear power plant, now in the hands of Talen Energy. This will be the first zero-carbon, nuclear-powered Bitcoin mining facility.

With a 200 MW capacity, this would be the equivalent to ~160,000 US households’ energy consumption annually.

Adapting Data Center Design for Blockchain Technology

In addition to spatial proximity to reduce transmission loss, data centers servicing blockchain networks need particular mechanical, electrical and plumbing (MEP) requirements. As every PC owner knows, the source of such requirements at large scale comes from heat management.

Continuous solving of cryptography puzzles requires great computing power which generates heat. For many years, air cooling has been the go-to solution to prolong hardware longevity and dissipate heat. Unfortunately, cooling also draws significant energy on top of computing itself.

This is why there is a new trend to rely more on direct-to-chip liquid cooling (immersion cooling) which cuts down on power usage.

₿