Bitcoin miners racing to repurpose their infrastructure for artificial intelligence need an estimated $50 billion in near-term capital, according to a new note from VanEck. The asset manager also pegged the long-term requirement at $221 billion — a figure that underscores just how capital-intensive the shift from mining blocks to hosting GPUs really is.
The funding gap
VanEck's analysts arrived at the $50 billion shortfall by modeling the costs of retrofitting data centers, securing high-performance computing hardware, and meeting the uptime guarantees AI customers demand. That's on top of the usual capital expenditures miners already faced after this year's halving cut block rewards in half, squeezing margins across the sector.
The $221 billion long-term estimate covers a full build-out scenario where a significant portion of Bitcoin's hashrate capacity converts to AI hosting over the next several years. Neither figure includes financing costs, which could add billions more if interest rates stay elevated.
Delivery gap
VanEck also flagged a worrying gap between announced AI hosting deals and actual hardware deployment. Only about 25% of leased capacity has been physically delivered so far, according to the note. That means three-quarters of the revenue miners have projected from these contracts is still contingent on equipment arriving, installation going smoothly, and customers not walking away during delays.
The AI hosting business demands a fundamentally different operating model than Bitcoin mining. Miners are used to running ASICs that churn through hashes 24/7 with minimal customer interaction. AI hosting requires negotiating service-level agreements with enterprise clients, maintaining near-constant uptime, planning hardware generations years in advance, and arranging different kinds of financing. It's a different game.
Narrative and reality
Mining stocks have rallied this year as companies announced AI partnerships and data center plans — sometimes before the first GPU was installed. VanEck's note acknowledges the narrative power: shares can run on AI buzz even if the business model hasn't yet produced durable revenue. But the funding gap and low delivery rate suggest that the market may be pricing in execution that hasn't happened yet.
The next few quarters will test whether miners can close that gap. Those that have already secured financing and delivered hardware will have a clearer story to tell. For the rest, the gap between announcement and reality remains wide — and investors are starting to look at the numbers.
