Bitcoin miners mine no more

What happened

Bitcoin miners have had a remarkable run this year, many posting triple-digit gains. While Bitcoin itself has also appreciated, the magnitude of miners’ rally cannot be explained by crypto prices alone. Instead, investors have rewarded those pivoting toward high-performance computing (HPC), the infrastructure backbone of the AI revolution.

HPC and data-center capacity have become the new “picks and shovels” trade for artificial intelligence. With their existing access to cheap power and scalable infrastructure, Bitcoin miners have been quick to capitalize. Yet this pivot has also reshaped how markets value these companies: less as crypto plays, and more as AI infrastructure providers, a second derivative of the broader AI boom.

Impact on our Investment Case

The institutionalization of Bitcoin mining has intensified competition and squeezed profitability. As the network’s hash rate climbed and industrial players entered the space, margins thinned, even as Bitcoin hovered near record highs. The 2024 halving further compressed revenues, while hardware upgrades required continuous capital injections to maintain a mining market share. Dilution became common as miners raced for efficiency and the latest generation of mining chips.

In parallel, in mid-2024, leading actors of the AI revolution began declaring that “unlimited” computer power would be needed. Faced with investor fatigue, Bitcoin miners discovered a new narrative: redeploy their infrastructure toward AI workloads, where every megawatt of energy could quickly generate 2–3× more annual revenue than Bitcoin mining (at the current Bitcoin prices).

Leveraging their expertise in managing complex rigs, cooling systems, and power-intensive operations, miners found a way to diversify their bisiness models. and Investors have rewarded them handsomely, despite the billions in equity, convertible instruments, and debt required to fund the buildout.

Miners disconnected from Bitcoin

After a year of announcements and pilot programs, the HPC transition is now becoming tangible. Facilities once dominated by ASIC rigs are being retrofitted for GPUs and AI clusters. Business models differ between transitioning miners: some operate GPU-as-a-Service platforms (e.g., Iris Energy), while others sub-lease power capacity to hyperscalers like real estate firms (e.g., Hut 8 Mining). Multi-year agreements, often worth billions, are being signed. Early revenue streams from AI workloads are starting to flow in.

Meanwhile, the stock market has gone euphoric. Historically, we tracked the mining sector using a ratio of miners’ market capitalization to their expected Bitcoin rewards over the next four years. This is a proxy for the intrinsic link between miners and crypto economics. This metric long guided how we adjusted our exposure to the segment. Yet it has now become meaningless.

Today, many miners’ valuations are disconnected from their Bitcoin production potential or Bitcoin on the balance sheet, reflecting a shift in how investors perceive them: no longer as Bitcoin proxies, but as speculative bets on AI compute.

Interpreting the multiple expansion

Switching to HPC is undeniably boosting miners’ top lines. That alone explains much of the recent multiple expansion. Yet investors seem to minimize both execution and timing risks.

Most HPC facilities will come online only in 1–3 years. The projects are heavily front-loaded in capital expenditure, with uncertain payback periods. Overcapacity is a real possibility if AI models become more efficient or demand normalizes.

Many data-center operators, including miners now transitioning, rely on 7- to 12-year lease agreements. Who knows what will happen to these actors in 10 years? These contracts seem to secure long-term revenue “visibility”, but they also cap flexibility and future growth potential unless fresh capacity is added.

Comparing miners’ valuations with pure-play HPC peers like CoreWeave shows how stretched some multiples have become. For miners, price-to-sales ratios increasingly seem correlated to planned capacity expansions. As projects are delivered and planned capacity peaks, the base effect should reverse, likely bringing P/S ratios back toward normalized levels.

Correlation with Bitcoin breaks down

Perhaps most strikingly, the 60-day correlation between listed miners and Bitcoin prices has fallen to record lows. This is not what we seek in our blockchain strategy.

Our objective is to capture beta exposure to digital assets, not simply to power infrastructure or AI capex cycles. In a world where tangible assets dominate narratives, we view cryptocurrencies (and by extension, blockchain-related stocks) as a new class of real assets in their own right.

Companies whose performance has become unmoored from cryptos' price action no longer serve that purpose within our portfolio construction framework.

Our Takeaway

Bitcoin miners have evolved into infrastructure plays. History reminds us that while infrastructure often underpins progress, it seldom delivers consistent shareholder returns for early movers. The railroads of the 1800s, the fiber-optic builders of the 1990s, and now the HPC data-center pioneers all share a similar pattern: massive upfront capex and broad long-term societal benefit, although the pioneers rarely endure long enough to reap the rewards.

We recognize the diversification benefits of HPC. Yet within our blockchain mandate, we remain focused on miners that view AI as complementary rather than core. Those continuing to invest in Bitcoin capacity stay truer to our investment philosophy and are less exposed to the multiple compressions that will inevitably follow once the current enthusiasm fades.

And if Bitcoin ever does reach $1 million, the miners who missed the HPC boom will not feel foolish in hindsight. But they’ll certainly have a great story to tell from another cycle.