🔥 Why Bitcoin Miners Are Outperforming BTC by 100% — And What's Next for AI Infrastructure

💡 Market Regime Shift: From Capex Kings to Software Selloff

The first five months of 2025 delivered a concentrated market where companies spending the most on capital expenditures posted the best-performing stocks. Since June 1st, that relationship completely flipped. Putting aside the recent bounce off lows, the companies pouring capital into infrastructure started getting punished.

Bitcoin and crypto fell into the software bucket during this rotation. The semiconductor-to-software ratio became a closely watched metric, and Bitcoin — being open-source software — got caught in the downdraft. The upgrade cycles for open-source projects like Bitcoin move slower than top-down Web2 companies that can force updates. This structural difference, combined with emerging vulnerabilities in projects like Zcash, contributed to software underperformance weighing on Bitcoin and crypto broadly.

The four-year Bitcoin cycle pattern also commanded respect, with market participants positioning accordingly. However, this changing equity market regime since June could create more balance and dispersion in returns, bringing back the ability to harvest alpha from single asset positions rather than pure factor bets.

"I don't think it has to be A works and B doesn't, B works and A doesn't. The market will be more nuanced than that."

⚡ The Bitcoin Miner Transformation: 100% Outperformance Through Strategic Pivots

The node ETF, launched 15 months ago, has managed to outperform Bitcoin by nearly 100 percentage points. This alpha generation came from identifying early that Bitcoin miners were valued at extremely low multiples relative to the megawatts of power they controlled — especially compared to existing data center REITs.

The old Bitcoin mining model was fundamentally broken:

• Dilute shareholders with equity issuance to acquire mining ASICs faster than competitors
• Race to keep up with growing hash rate
• Face revenues falling 50% every four years (the halving)
• Operate a capital-intensive, melting ice cube business model with limited profitability

The new model changes everything. Debt issuance from hyperscalers and data center construction leases has caused the cost of capital for these miners to collapse. They can now fund capital needs through debt markets instead of equity markets. Each successive lease has been signed at better economics, and with lower interest rates, substantial value creation has occurred.

Key tickers leveraging this transition include MARA, Riot, APLD, and WULF — all Bitcoin miners converting existing infrastructure into AI data centers. These companies already built out the raw components needed: power, energy capacity in rural areas, and facility infrastructure.

📊 Leaning Into Volatility: Doubling Down When Leverage Got Flushed

The June regime change hit leveraged players significantly harder than unleveraged strategies. When forced selling from leveraged funds like Situational Awareness created overlap liquidations, it presented an opportunity.

Last Thursday morning at the open, the team executed the largest single-day trade since the fund's launch — exiting almost 10% of lower volatility, lower beta exposures and doubling down on preferred miners. This decisive action came despite having significant position overlap with liquidated funds, because the fundamental story remained intact.

The math was compelling: Some favorite names were pricing in only the value of existing leases, with nothing ascribed to the terminal value of data centers or the pipeline for new leases. This work held even assuming the 10-year Treasury rose another 100 basis points, creating an attractive margin of safety at the lows.

"We haven't seen any fundamental deterioration in the return on capital that the hyperscalers are expecting from this investment. In fact, if you read the transcripts from Amazon and others, those returns are tracking better than their earlier expectations."

Companies that had been renting older GPUs for $2 an hour are now trying to refresh those contracts at significantly higher rates.

🔄 The Optionality Play: When Miners Might Pivot Back to Bitcoin

MARA's CEO Fred Thiel revealed on a recent podcast that if MARA wanted to, they could fit all of their Bitcoin mining machines into the one new facility they just purchased in South Texas. This highlights the optionality available to miners running containerized facilities.

The pivot isn't necessarily permanent. For facilities already repurposed, the economics would need to shift dramatically. For example, CleanSpark would need Bitcoin at $360,000 per coin to justify tearing up their recently signed lease. But for other miners with underutilized facilities, moving ASICs around while preparing additional capex for leasing remains viable.

Names like Bit Deer and MARA retain significant optionality around whether to repurpose facilities or continue mining Bitcoin. There could be a Bitcoin price level where market conversations shift to "when will they pivot back?"

More hash rate leaving for AI applications creates a self-balancing mechanism: as hash rate falls, those who stay connected to the Bitcoin network earn additional profits. This dynamic doesn't present systemic issues to the Bitcoin price — it's a feature, not a bug.

🚂 Railroad Boom vs. AI Boom: Why This Time Is More Sustainable

Critics compare the current AI capex cycle to the railroad boom of the 19th century — a transformative but ultimately capital-destructive bubble. A detailed analysis reveals crucial differences suggesting the AI buildout is more sustainable and could last 20 years (we're currently in year three or four).

Scale Comparison:

• The US put 3% of GDP into rails for almost two decades during the 19th century
• We're only hitting that 3% of GDP level once this year (year five post-GPT)
• Current equity valuations don't discount this continuing for two decades
• Most analysts see capex peaking by 2030

Financing Structure (Critical Difference):

• Railroads: Government-engineered through the Railway Act of 1862. Congress granted hundreds of millions of acres of federal land to railways, but titles wouldn't transfer until the whole network was built. The US Treasury floated construction bonds subordinated to private capital. Railways sold bonds overseas marketed as safe, dependent on land sales for land they didn't even have title to yet — this was a bubble.
• AI Infrastructure: Backed by contracted customer demand. The four largest cloud providers alone (Amazon, Google, Microsoft, Oracle) have more than $2 trillion in contracted backlog. Many deals include customer prepayments and customer-supplied GPUs. Contracts exceed 5 years in duration.

Utility Timeline:

• Railroads: Provided no true utility until East and West coasts connected — a massive coordination problem
• AI Factories: Can start training models and serving inference immediately once chips are trained and proper grid/fiber connections established — no need for synchronized global network

Speculative vs. Contracted Demand:

• Railroads: No one was buying rail tickets in 1870 for routes not completed until 1885. No forward freight market. Pure speculation based on land sales.
• AI Infrastructure: Legitimate purchase orders associated with compute output from these factories. Financing based on multi-year customer commitments, not government handouts.

"My thesis is that the fundamental scarcity here is now atoms, no longer bits."

🌍 From Bits to Atoms: The Reshoring Mega-Trend

Three decades of US de-industrialization occurred because commodities and manufacturing capacity opened globally. COVID exposed the vulnerability of integrated supply chains, creating a reshoring imperative.

The demand shock from AI compounds this challenge: intelligence cannot be served globally. France doesn't want their AI served from US data centers. This means essentially doubling everything and losing economies of scale.

NIMBYism has made building in the US exceptionally difficult, creating a fundamental cap on how quickly capacity can return. Whether the Bitcoin miners were lucky or Bitcoin always anticipated the world would shift back from bits to atoms remains debatable, but the structural advantage is clear.

📉 Why L1 Tokens Remain Underweight: The Corporate Chain Reality

Following the election, when many L1 tokens doubled without acceleration in real-world adoption or viral app emergence, exposure to L1s was reduced across the board. What emerged instead were corporate chains — quasi-permissioned blockchains from Circle, Stripe, and Robinhood that allow public companies to customize user experience and extract economics.

While not in the spirit of open-source development, entities deploying at scale want predictability of the fee stream. Ethereum demonstrated this challenge with highly volatile transaction costs. Corporate chains have captured significant market share.

Until recently, you couldn't buy USDC on Solana in New York. This morning brought headlines of Wells Fargo working on tokenized deposits — on their own custom-built blockchain. Banks and regulated entities don't want to put real capital on open-source chains. When they do, they dilute their interest by supporting three, five, or ten L1s rather than one winner.

This represents both market share loss and a lack of winner-take-all characteristics within remaining players.

💭 What Would Change the L1 Outlook?

Positioning against L1s is crowded — many investors share this perspective. If a Clarity Act somehow passes (though odds are at year-to-date lows), an enormous relief rally in these coins would likely occur.

A proper disclosure regime would reveal:

• Beneficial owners (ending anonymous shilling without ownership disclosure)
• How many tokens labs own versus foundations
• Actual token distribution and control dynamics

This lack of disclosure has made many institutional investors ignore the space entirely. Meaningful regulatory clarity and transparency would need to emerge before getting properly bullish on many L1s.

📊 The Inflation Problem: ETH, Solana, and Near Respond

Ethereum (as of this morning), Solana (earlier this week), and Near have all proposed reducing validator inflation. Ethereum and Solana would reduce inflation paid to validators, while Near proposes a fund generating yield to pay validators. If these proposals pass, inflationary supply for all three L1s would drop to almost zero.

Recent analysis compared average L1 inflation rates to user growth and fee generation — revealing a definite inflation problem across the space. The year-to-date outperformance of semiconductors versus software forced many software companies to reckon with share dilution, and dilution in that space has decreased substantially.

The devil will be in the details, including second-order effects on companies relying on staking revenues (like Bitmine, which touts staking income). But six years after the L1 birth wave, rethinking inflation structures is appropriate.

🎯 Positioning for Q4: Waiting for Volume on Good News

Current positioning remains exposed to the AI infrastructure theme via Bitcoin miners, with optimism that Q4 will bring a bottom to Bitcoin and more market balance. Rather than playing current chop, the preference is buying on good news when accompanied by volume.

For tokenization infrastructure companies and altcoins, the wait continues for:

• Clarity to emerge on rulemaking by the SEC
• Better prices accompanied by actual volume
• Volumes across most crypto tokens remain dismal

The market feels totally apathetic right now, with the equities market and AI trade absorbing all liquidity. But changing conditions, self-balancing mechanisms in Bitcoin mining economics, and the structural advantages of miners-turned-AI-infrastructure plays suggest patient positioning today could capture significant alpha when the rotation eventually occurs.