AI gold rush draws crypto firms away from Bitcoin

Companies that once filled warehouses with computers to earn Bitcoin – a process known as mining – are now using that computing power for AI.

Bitcoin mining companies invested large sums on giant banks of powerful computers as the price of Bitcoin rocketed in recent years.

But those mining companies have faced falling rewards since the value of the cryptocurrency has slumped from its peak in October 2025.

Now many are pivoting to AI and signing deals with companies such as Anthropic – who are spending billions of dollars on the infrastructure needed to keep up in the AI race.

Like AI systems, Bitcoin relies on vast networks of powerful computers housed in data centres.

Because Bitcoin operates without a central authority, these computers verify transactions and are rewarded with newly-created digital coins.

But the rewards have reduced and the value of coins has dropped since last year.

One bitcoin was worth about $124,000 (£91,000) at its peak in October 2025, but has since fallen sharply.

More recently, it has rallied to around $80,000 – meaning it is up almost 30% so far in August.

But, for companies that have already made the change, even that may not be enough to get them to return to the crypto industry, because the switch of use – once made – is expensive to undo.

Industry analysts say Bitcoin mining companies have been pivoting to AI because they have years of experience in finding cheap electricity and efficiently running large data centres.

TerraWulf, Ionic Digital, Core Scientific, Iris Energy, Bitdeer, Riot Platforms and Hut 8 are just some of the companies increasingly diverting investment and infrastructure from bitcoin mining towards AI.

Riot Platforms signed a $9bn, 20-year compute deal with Anthropic earlier this month.

Company names and websites are changing to reflect the shift too.

Applied Blockchain has become Applied Digital.

TerraWulf’s website once described the firm as an “infrastructure-focused bitcoin mining company”.

It now says it focuses on “next-generation AI and high-performance computing”.

Enegix, which opened a huge Bitcoin mining site in Kazakhstan to great fanfare in 2020, is now pivoting too.

“Today, we are moving confidently towards artificial intelligence and planning the gradual alignment of our energy and infrastructure capabilities, both in Kazakhstan and elsewhere, towards the development of AI infrastructure,” said the company’s chief executive Yerbolsyn Sarsenov.

Enegix says it is in active discussions with AI and HPC (high-performance computing) companies and planning to transform a “significant” portion of the business to AI.

But refitting crypto mines for AI is expensive and in some cases the companies have had to sell some of their Bitcoin holdings to fund the switch.

Wolfie Zhao from The Energy Mag – a publication that rebranded from The Miner Mag to reflect the trend – predicts the industry pivot will continue even as the price of Bitcoin has risen in recent days.

“We expect to see many public miners continue winding down their Bitcoin mining hardware in the coming quarters,” he said.

Asked if the shift away from Bitcoin mining might affect the security or stability of the cryptocurrency, Zhao said he hoped more miners will enter the market as conditions improve.

But he predicted many of the big players would not easily be able to switch back to Bitcoin.

“Once that multi-gigawatt power infrastructure has been retrofitted to AI or HPC colocation, there is no turning back,” he said.

“You can unplug from the Bitcoin network any time but signing a GPU colocation lease for 10 or 20 years means steady revenue and a commitment to keep the infrastructure up for the tenants.”

Bitdeer, which claims to be the largest Bitcoin miner in the world, just announced a 16-year deal to provide compute for Anthropic.

But the company’s chief strategy officer, Haris Basit, said Bitdeer will continue mining Bitcoin, believing many operators would ultimately pursue a dual-purpose model.

“Bitcoin mining is particularly well suited to that model because it is flexible and interruptible, while AI workloads can provide longer-duration contracted revenues,” he said.

Sign up for our Tech Decoded newsletter to follow the world’s top tech stories and trends. Outside the UK? Sign up here.