- Cryptomining operations are increasingly looking to offer AI compute
- They have access to key resources like power and data center facilities
- But they need more networking and AI expertise to be successful
There’s about to be a flood of new blood in the cloud world. Companies originally formed to fuel cryptocurrency mining operations are aiming to stake a claim in the AI market and capitalize on the 21st century’s digital gold rush. And they just so happen to have a glut of a key resource other data center operators are scrambling to secure: power.
“They have access to electricity, which is an even harder commodity to come by than Nvidia chips,” J. Gold Associates founder Jack Gold told Fierce.
AvidThink Founder Roy Chua made the same point, noting in an interview that multi-gigawatt power pipelines have a major appeal in an energy constrained market like AI computing.
Case in point: CoreWeave’s recently announced deal to buy Core Scientific. Sure, CoreWeave (which itself was originally a cryptomining operation before pivoting to cloud computing in 2019) will get Core Scientific’s infrastructure as part of the transaction. But Chua noted it’s also getting power – and lots of it.
Core Scientific will give CoreWeave access to 1.3 GW of power across its data center footprint, with another 1 GW+ of incremental power available.
The great crypto migration
CoreWeave is perhaps the best example of the crypto to AI migration (or, if you prefer, hedge) that’s happening. But while it has found an unusual degree of success, it is far from the only one to be making this shift.
Hut 8, for example, has advertised its newly built 205 MW Vega data center campus as perfect for AI training workloads. It also recently purchased nearly 600 acres in Louisiana as part of a plan to build 430 MW of AI data center capacity.
Elsewhere, Singapore-based Bitdeer launched an AI-cloud offering in late 2023 and followed up with an AI training platform in mid-2024. Last month, it noted discussions to build “large scale sites in the U.S.” for AI and other high-performance compute workloads are underway.
Meanwhile, Bitcoin-fueled Riot Platforms just hired a new Chief Data Center Officer to focus on building compute facilities for hyperscale and enterprise clients.
Chua and Gold said that beyond power, there are a few other things that crypto players like these bring to the table. For starters, they bring existing facilities and experience running data centers. While their networking, cooling infrastructure and chips might need an upgrade, the existence of physical facility is a big deal given new data centers can take several years to construct.
They also have experience with fundraising, which is an important skill in a market that is extremely capital intensive.
Sure, they might also need to recalibrate operations to meet the more stringent standards required for AI workloads, but that expertise can be hired in, Chua said.
Consolidation or collapse?
Chua said crypto converts would most likely be in competition with traditional cloud compute providers, including hyperscalers, rather than colocation providers. But at least for the time being, there’s more than enough demand to go around.
Gold told Fierce that for now, AI is easy money. The big question, though, is how long that will be the case. Additionally, Chua said rising cryptocurrency prices (for example, Bitcoin hitting $120,000) and regulatory favorability could change the calculus for those mulling a migration. Some, he said, might just hedge, finding profitability in both markets.
It’s hard to predict exactly how things will play out given how quickly the sands are shifting in both the crypto and AI realms. But there seem to be two likely outcomes of the current trends: consolidation or collapse.
Gold said consolidation usually happens once a market matures a bit more, usually three to five years in (as it did in the cloud market, for instance). Thus, he wouldn’t expect it in the near term, but consolidation will likely follow over the next few years. Chua agreed.
“The reality is consolidation will happen,” Chua said. Those, like CoreWeave, who have already made the “rig to rack” leap will likely go out and look for assets to fill in gaps in their portfolio. And growing their scale will help them raise more money, which in turn can be used to fuel more consolidation.
But there’s also the possibility that the AI bubble eventually bursts, both analysts said.
On one hand, the market could simply end up with a glut of supply that finally surpasses demand. But it’s also possible that the demand side could end up being the problem.
Chua noted that if the industry isn’t careful, AI could essentially end up cannibalizing its own customer base.
Essentially, if AI replaces too many jobs and people can’t find new occupations at similar pay levels, they will cut non-essential spending. That doesn’t just mean on direct AI products like ChatGPT or Gemini subscriptions but also on other AI-enabled services across industries. And that wouldn't just be a problem for crypto converts but the economy as a whole.
“As we grow this, we need to ensure everyone grows together” lest the demand side of the AI equation suddenly disappears, Chua concluded.