Bitcoin Miners Weigh AI Conversion at Up to $15 Million per MW

By Tom Whitfield

Publicly listed Bitcoin mining companies signed more than $70 billion in AI and high-performance computing (HPC) agreements from 2025 through early this year, according to CoinShares’ first-quarter report.

The agreement activity comes amid broad local voter support for tighter data-center rules. Cointelegraph reported on November 11 that 40 of 41 local ballot measures tracked by Ballotpedia in the US midterm election favored limits on development.

The count covered local measures, not a single nationwide construction rule. Localities pursued different approaches, giving developers a varied set of approval hurdles before new AI capacity could be built.

What do the ballot measures restrict?

Local measures covered several forms of restriction:

• Bans on new data-center construction.
• Caps on large facilities.
• Temporary moratoriums.
• Requirements for resident approval.

As construction restrictions spread, existing Bitcoin miners’ power contracts, grid access and sites could become stronger alternatives for AI infrastructure developers. The reported analysis treats that advantage as conditional on capital and power supply, rather than a direct conversion of mining capacity into operating AI capacity.

Can miners convert mining sites into AI capacity?

“Bitcoin mining companies have more reason than ever to convert those power agreements and facilities for AI use if they have not already done so,” Jamie Brown, chief executive officer of Global Kiwi Labs, said.

Brown pointed to the strategic case for conversion. CoinShares estimated that building an AI data center costs $8 million to $15 million per megawatt, compared with $700,000 to $1 million for a Bitcoin mining facility.

The spending gap makes conversion a capital-intensive project even for operators that already hold land and power contracts. “Significant additional investment is needed in cooling, networking and reliability infrastructure,” Jordan Knecht, head of strategic integration at Cresus, said.

Knecht’s assessment matters because a connected mining site cannot operate as an AI data center without those upgrades. Power access addresses one requirement; cooling, networking and reliability determine whether the site can support AI computing services.

Could electricity prices and regulation weaken the advantage?

Yuliya Barabash, founder of SBSB Fintech Lawyers, said Bitcoin mining can reduce electricity consumption when power prices rise, while AI data centers require a stable power supply.

That operating difference leaves converted mining sites exposed to electricity-market conditions even when land and grid connections give miners an edge. Mining operators can respond to higher prices by consuming less electricity; an AI conversion requires dependable delivery, so power costs can affect returns.

Electricity costs and regulatory burdens could therefore undermine profitability or delay project timelines.

Knecht said the new Congress may take a closer look at data centers’ electricity costs, the strain they place on the grid and permitting procedures. He also said pressure could build for large data-center operators to absorb infrastructure costs themselves rather than pass them onto consumers.

Those issues extend beyond converted mines. Rules requiring operators to absorb those costs could change project economics even where power access is already in place.

As local approvals become harder to secure, the AI infrastructure market may reward operators with permitted sites and contracted power over developers seeking new connections. The next test is operational: miners must fund conversion, secure stable electricity and satisfy local approval requirements before a grid-connected mining site becomes a running AI data center.