Bitcoin Miners Sink $30.7 Billion into AI, Reap Just $341 Million in Returns

Bitcoin miners are betting billions on artificial intelligence and high-performance computing. The returns? Not even close — yet.

In the first half of 2026, nine public miners pulled in $341 million from AI and HPC operations. Sounds decent until you see the other side of the ledger: those same companies, along with a broader group of 15 Bitcoin miners and AI data-center firms, spent over $30.7 billion on capital assets during their most recent reporting periods. That’s a 15-to-1 capex-to-revenue ratio, and it’s the number everyone in the sector is quietly sweating over.

$30.7 Billion Spent, $341 Million Back

The $30.7 billion figure comes from BlocksBridge Consulting, which tracked capital expenditures across 15 Bitcoin miners and AI data-center companies through 2026. That’s a 42.6% jump from the $21.53 billion spent across the same group in 2025. The spending covers hardware, property, and other productive assets, net of any sales or refunds — so it’s not inflated by paper moves.

And it’s not like revenue is standing still. The nine companies in the AI and HPC cohort saw a 52% jump in revenue from those segments in the second quarter alone, hitting $205.8 million for Q2. Core Scientific, TeraWulf, and Bitdeer were among the names reporting gains. But 52% growth on a small base is still a small number when you’re burning through capital at that pace. The math doesn’t get comfortable anytime soon.

Not yet, anyway.

Bitcoin’s Price Surge Adds Breathing Room

Bitcoin climbing past $72,000 — up more than 13% recently — gives miners running traditional operations a bit of cushion. That move was partly driven by the US Treasury’s decision to ramp up long-term bond buybacks, which initially pushed yields lower and nudged investors toward riskier assets. Classic risk-on behavior. Miners still holding significant Bitcoin exposure probably felt some relief, at least temporarily.

But the price bump doesn’t fix the structural problem. Pivoting from Bitcoin mining to AI-ready infrastructure is expensive in ways that go beyond buying a few GPUs. Facilities need substations. They need serious cooling systems. Networking equipment has to be upgraded. Some business models require GPU integration on top of all that. These aren’t small line items — they’re the kind of capital commitments that take years to pay off, if they pay off at all.

See also: Bitcoin ETFs Surge $517 Million as $2.7 Billion in Short Bets Vanish

So miners are basically running two businesses at once: the old one, which lives and dies with Bitcoin’s price, and the new one, which is burning cash waiting for AI revenue to scale.

CoinShares Repositions, Industry Follows

CoinShares made a telling move in response to all this. The firm rebranded its industry-tracking ETF to pull in a wider range of digital economy assets — AI semiconductors, power generation, HPC companies. It’s a pretty clear signal about where institutional money thinks the sector is heading. The ETF’s updated scope isn’t just about Bitcoin miners anymore; it’s about capturing the broader infrastructure build-out that AI demands.

That kind of repositioning probably makes sense given the numbers. If you’re tracking an industry where companies are spending $30.7 billion and generating $341 million in the revenue segment they’re pivoting toward, you want exposure to the whole chain — not just the miners who may or may not survive the transition.

The power contracts and land holdings that miners already own do give them a real edge, at least on paper. Data centers need power. Miners have it, or at least have the contracts for it. That’s not nothing. But having the land and the power contract is maybe 20% of what it takes to run an AI-capable facility. The rest — the substations, the cooling, the networking — still has to be built.

More context: Bitcoin Surges to $69,749 as $1.48 Billion in Shorts Liquidated in One Hour

Unclear how many of the 15 companies tracked by BlocksBridge will actually complete that build-out at scale. Some will. Some probably won’t.

The 52% quarter-over-quarter revenue growth in AI and HPC is the number bulls keep pointing to. And it’s real. But at $205.8 million for Q2, even doubling that figure every quarter for the next year barely dents the capital already deployed. The gap between what’s been spent and what’s coming back is still enormous — and it’s probably going to stay that way for a while.

Core Scientific, TeraWulf, and Bitdeer reported gains. The rest of the details on individual company performance weren’t broken out in the BlocksBridge data.