The rapid surge in interest around artificial intelligence and the subsequent need for computing power has left everyone wanting in on the action- including Bitcoin miners

Numerous Bitcoin mining companies have announced moves into the AI computing space, spurring interest and intrigue from the private and public markets.

The big picture: transitioning from Bitcoin mining to powering AI computing is not a straightforward endeavour.

• The process of mining Bitcoin requires ASIC chips. ASIC chips are not able to be repurposed to anything to do with AI
• However, the infrastructure of large Bitcoin mining operations have many of the foundational pieces of AI compute, such as low cost power, cooling systems and security
• There is still a global shortage of GPUs for AI compute, with large chip providers such as Nvida (NASDAQ: NVDA) citing many months of back orders

Who is making moves:

• Applied Digital (NASDAQ: APLD) entered into a strategic partnership with Hewlett Packard to deliver AI cloud computing
• Iris Energy (NASDAQ: IREN) announced plans to “revitalize” their high performance computing data centres because of the surge in interest for AI
• Hut8 (NASDAQ: HUT) has invested in numerous data centres that can power AI compute, marking their first strategic investment away from solely Bitcoin mining
• HIVE Digital Technologies (NASDAQ: HIVE) changed their name from “HIVE Blockchain” and announced their intent to focus on the “evolving opportunity” with Nvida GPUs

The fine print:

• The true boom in AI computing relies on AI startups who are raising venture funding to train large language models or generative AI tools, but don’t want to purchase their own GPUs, because it’s either too expensive or because the wait time for GPUs is simply too long
• Ultimately, startups- especially venture backed ones- are renting computing power to advance product and technology development
• If the boom cycle for AI startups dwindles, there will likely be a large correction in the valuations of high performance computing power for AI
• Much of the pricing power for GPUs is driven by a large dislocation between supply and demand- if that changes, so will the economics of selling high performance computing

‍