Bitcoin Mining Capex Surges as AI Push Outruns Revenue 15:1

Public Bitcoin miners are pouring large sums into artificial intelligence and high-performance computing (HPC) infrastructure as part of a broader push to diversify beyond pure mining revenue. But new data compiled by BlocksBridge Consulting suggests the transition is still dominated by upfront capital spending, with returns lagging far behind.

In its latest Miner Weekly newsletter, BlocksBridge reports that a group of 15 Bitcoin miners and AI data-center companies spent a combined $30.7 billion on capital assets in their most recent 2026 reporting periods. That figure is 42.6% higher than the $21.53 billion these companies spent over all of 2025. The figures help quantify just how expensive it is to build capacity for AI workloads—often in parallel with continuing mining operations.

Key takeaways

• $30.7B: Total capital asset spending by 15 Bitcoin miners and AI data-center companies in their latest 2026 reporting periods, per BlocksBridge.
• Capex far exceeds AI/HPC revenue: Nine comparable miners spent $5.11B on capex in the first half of 2026 while reporting only $341.2M in directly reported AI/HPC revenue.
• Revenue growth is accelerating: AI/HPC revenue from those nine miners rose to $205.8M in Q2 2026, up 52% quarter-on-quarter.
• Pivot requires more than power and land: BlocksBridge highlights the need for substations, buildings, cooling, networking, and often GPUs.
• Industry funds are reframing the thesis: CoinShares rebranded its strategy ETF to include companies supplying digital power beyond mining alone.

Capex surge highlights the cost of scaling AI-ready capacity

AI and data centers have been widely discussed as diversification paths for Bitcoin mining companies facing a challenging industry backdrop. BlocksBridge’s analysis adds a granular cost lens to that narrative, showing how quickly capital needs expand when miners attempt to convert existing infrastructure advantages into AI-ready computing environments.

According to BlocksBridge, spending was calculated based on cash purchases and allocations to hardware, property, equipment, and other productive assets—after taking into account proceeds and refunds from asset sales. Even with those adjustments, the gap between investment and revenue remains large.

Among Bitcoin miners specifically, the mismatch looks particularly stark. BlocksBridge identifies nine comparable miners that collectively spent $5.11 billion on capital assets during the first half of 2026, generating just $341.2 million in directly reported AI and HPC revenue. That equates to roughly a 15-to-1 capex-to-revenue ratio for the period covered.

Q2 revenue growth suggests demand is building, even if profits lag

While the early spending burden is clear, BlocksBridge also reports signs that AI and HPC revenue is gaining momentum. For the same group of nine miners, total AI and HPC revenue increased to $205.8 million in the second quarter—a 52% quarter-on-quarter rise.

BlocksBridge notes that companies including Core Scientific, TeraWulf, and Bitdeer were among those reporting gains tied to their AI/HPC efforts. The acceleration matters because it indicates the investments are beginning to translate into recognizable business performance, even if the scale of capex still overwhelms what is currently booked as revenue.

For investors and analysts, the immediate implication is that the diversification story is shifting from “planned buildout” to “commercialization,” but with significant timing risk. The cost is already on the balance sheet or cash-flow path; the payoff appears to be arriving later and in uneven increments across companies.

From mining advantage to AI infrastructure: what still must be built

BlocksBridge frames the pivot challenge in practical terms. While miners may have initial advantages—such as access to power contracts and available land—those assets do not automatically become AI-capable capacity. In its reporting, BlocksBridge says that converting such advantages into AI-ready infrastructure typically requires additional components, including substations, buildings, cooling systems, networking equipment, and—depending on the business model—GPUs.

This matters because it clarifies why AI/HPC commercialization can be slower than headline narratives imply. Mining operations can often run with relatively straightforward operational continuity, but AI workloads involve different infrastructure requirements and more intensive engineering to achieve reliability, scalability, and performance.

BlocksBridge also leaves open a key question for the near term: whether any broader improvement in Bitcoin’s price environment will reduce financial pressure on companies still operating large mining fleets. Bitcoin’s price moves can help sentiment and—depending on each firm’s leverage and hedging—may influence how much runway companies have while AI projects ramp.

Earlier this week, Bitcoin rose more than 13% and moved back above $72,000, following a statement by the US Treasury that it would at least double the maximum size of its long-term bond buybacks to $4 billion per operation. That decision was described as aiming to improve liquidity in the Treasury market, initially pushing yields lower and boosting risk appetite.

ETF strategy shifts mirror the broader “digital power” rebrand

In parallel with the infrastructure buildout, parts of the investment industry are adjusting how they package exposure. CoinShares, this week, announced changes to the way its industry-tracking ETF is positioned and branded.

The fund is now called the CoinShares Bitcoin Mining and Digital Power ETF (WGMI). CoinShares reports that the ETF has $222.4 million in assets under management, and that it draws from a broader set of businesses than a pure mining basket. Its “universe includes 29 holdings” spanning bitcoin miners, data center operators, AI semiconductors, power generation, and HPC companies, which CoinShares describes as “the businesses powering the digital economy.” The fund’s details are listed on CoinShares’ site: https://coinshares.com/us/etf/wgmi/.

For market participants, the ETF shift signals that investors are increasingly seeking exposure to the infrastructure layer around compute—not only the economics of mining blocks. Still, BlocksBridge’s capex-to-revenue figures emphasize that this infrastructure layer is currently expensive to build. The critical test will be whether rising AI/HPC revenue can eventually narrow the investment gap as projects move from construction into sustained operating contracts.

Over the next few reporting cycles, readers should focus on whether the revenue ramp continues for individual miners and whether capex intensity begins to cool relative to AI/HPC income. The data already shows acceleration in Q2, but the core uncertainty remains timing: how long it takes for heavy infrastructure spend to convert into durable, scalable returns.