Bitcoin Miners Shift Toward AI and High-Performance Computing
As Bitcoin miners turn their attention to artificial intelligence and high-performance computing (HPC), many expect their main competition to come from major cloud providers like Amazon, Google, and Microsoft.
However, Frank Holmes, the executive chairman of HIVE Digital Technologies, suggests the situation is more intricate and costly than it appears.
Holmes points out that the shift from Bitcoin mining to HPC isn’t just gradual; it’s quite revolutionary.
He noted, “An ASIC chip Bitcoin mining data center costs around $1 million per megawatt of electricity. If you switch to Tier 3 HPC, your infrastructure expenses jump to $10 million per megawatt. That’s a tenfold increase.”
Costs associated with hardware are merely a portion of the hurdles. With the addition of Nvidia’s advanced chips, expenses can escalate significantly.
“NVIDIA chips will add an additional $25 million to that cost,” he stated, referencing estimates from the University of Cambridge which suggest that a cutting-edge HPC facility can cost over 30 times more than conventional Bitcoin mining setups.
The operational intricacies also rise dramatically. Unlike typical mining sites, HPC data centers need multiple redundant internet connections, backup fuel systems, advanced cooling mechanisms, and sophisticated power management systems.
“When you acquire an ASIC chip from Bitmain, it takes about six hours to unpack, connect, and start generating cash flow,” Holmes elaborated. “In contrast, with the H200s, it requires six weeks to develop a brain.”
As miners transition into AI computing, Holmes argues that establishing partnerships with hyperscalers will be crucial for maintaining reliability and accessing revenue streams.
He highlighted industry collaborations like Core Scientific’s with Coreweave, while others seek alliances with Microsoft, Amazon, and Oracle.
“For Bitcoin miners, partnering with other hyperscalers is essential,” Holmes conveyed. “CoreWeave seems to be the quickest. Oracle provides a lot of flexibility.”
Conversely, he identified Microsoft as being particularly challenging to collaborate with.
Beyond private enterprises, Holmes sees rising global military expenditures as a catalyst for demand in advanced computing infrastructure. Technologies such as autonomous drones and robotics require substantial data processing capabilities.
“All these technologies necessitate data centers,” he remarked, adding that he believes governments are seeking long-term contracts, which is prompting a broader expansion cycle in the industry.
Holmes maintains a positive long-term outlook despite occasional market downturns. He explained that the merging of Bitcoin infrastructure, AI computing, and increasing global digital needs reflects a broader, ongoing trend.
“Buy the push and hold on for a while,” he advised. “This isn’t limited to Bitcoin mining; the entire AI and technology sector is experiencing a sustained supercycle.”
