Bitcoin Mining Stocks vs Bitcoin: Miners Fell 46% While BTC Rose 22% — 5 Numbers That Decide If the AI Pivot Is a Trap or a Bargain

The 46% Gap Nobody Expected

Bitcoin rose 22% over the last three months. Over the same three months, an equal-weight basket of nine publicly traded Bitcoin miners fell 46.2%. It is the widest divorce between Bitcoin and the companies that mine it in the history of the sector — and it happened while Bitcoin itself was going up. If you bought a "Bitcoin miner" stock in August expecting leveraged exposure to Bitcoin, you lost nearly half your money while Bitcoin gained more than a fifth. Here are the five numbers that decide whether the wreckage is a bargain or a trap.

The short version: these companies stopped being Bitcoin miners. In 2026 they became AI landlords — owners of power, land and cooling that they rent to artificial-intelligence companies — and the market now prices them on AI credit, not on Bitcoin. Over the last 126 trading days, the miner basket's correlation with CoreWeave was 0.72; with Bitcoin it was 0.40. CoreWeave alone explains 52% of the miners' daily moves; Bitcoin explains 16%. On October 14, CoreWeave fell 15.5% and the miner basket fell 11% — while Bitcoin fell just 1.8%.

Bitcoin itself traded near $76,000 on November 7, 2026 — about 40% below its $126,156 all-time high. The miners' story is no longer Bitcoin's story. Here is what actually happened, number by number.

Number 1: -46.2% — The Three-Month Divergence

The headline number comes from a return study covering prices through October 30, 2026: Bitcoin up 21.6% over three months, an equal-weight basket of nine AI-pivoting miners — IREN, Cipher, TeraWulf, Core Scientific, Hut 8, MARA, CleanSpark, Bitdeer and Applied Digital — down 46.2%. No earlier period shows a split like it, because no earlier period had miners this detached from mining.

It was not a slow bleed. In the 21 trading days into November 2, Cipher Mining fell 56% and IREN fell 47%, and most of the group lost 30% to 45%. Then on November 3, nine big miners ripped 10% to 17% in a single session — Cipher up 16.7%, Riot up 16.3%, TeraWulf up 14.2% — while Bitcoin rose only about 1%. Two days later, on November 5, the whole index fell again, down nearly 15% by late morning, with IREN down about 18% and Cipher down about 17%. Then on November 6, IREN jumped 20% in a single session. A stock that falls 18% on Thursday and rises 20% on Friday is not trading on Bitcoin.

Number 2: 0.72 vs 0.40 — The Correlation Flip

The second number rewrites how you should think about these stocks. Over the last 126 trading days, the miner basket's daily correlation with CoreWeave was 0.72. With Nvidia it was 0.53. With Bitcoin it was 0.40. The same flip shows up in the longer history: the correlation of the WGMI miners fund with Bitcoin fell from 0.67 in 2023 to 0.44 in 2026, while its correlation with chip stocks rose from 0.34 to 0.69. Coinbase and Strategy, by contrast, still trade like Bitcoin — which is why they belong in a different mental bucket.

In practice, the miners now fall when AI credit wobbles, even when Bitcoin is fine. The repricing of AI debt was the single biggest driver of the autumn sell-off: when CoreWeave dropped 15.5% on October 14, the miner basket dropped 11% the same day while Bitcoin slipped 1.8%. The miners' risk is no longer "Bitcoin goes down." It is "the AI tenant can't pay." That is a credit risk, a customer-concentration risk and a construction risk, all dressed up in a Bitcoin costume.

This is also why the old rule of thumb — miners are a leveraged play on Bitcoin — is now dangerous. It was true when mining revenue was the business. In 2026, mining revenue is the shrinking part of the business. CoinShares estimated in March 2026 that listed miners could derive as much as 70% of their revenues from AI by the end of the year. The full return study behind the correlation numbers is published by TTM Financial.

Number 3: $9.7 Billion and $28 Billion — The Two Deals That Define the Sector

Two contracts now define the mining sector more than any hashrate chart. The first is IREN's $9.7 billion Microsoft contract covering four 50-megawatt phases at its Childress site in Texas. The second is Riot Platforms' 15-year, roughly $28 billion lease with Anthropic covering 910 megawatts at its Corsicana, Texas campus. Add Core Scientific's roughly $10.2 billion CoreWeave colocation deal and Hut 8's $7 billion, 15-year AI lease at River Bend, and the picture is clear: the seven miners whose filings could be checked have about 3.9 gigawatts of AI and HPC capacity under contract.

Here is the catch: much of this rent starts in 2027 and 2028. The market is paying today for buildings that bill later. The tenant has to stay solvent, the build-out has to finish on time, and the power has to actually arrive. When a report suggested a delay to Microsoft's timeline, IREN fell about 18% on November 5 — hours before it reported fiscal first-quarter results — and dragged the entire PRISM Crypto Mining Index down nearly 15%, as PRISM MarketView reported. One customer, one delay rumor, one-fifth of the share price. That is concentration risk in its purest form.

Then IREN reported — analysts had expected revenue of about $221.7 million and an adjusted loss of $0.67 per share — and the next day, November 6, IREN closed at $21.64, up 20% on the session. The whiplash tells you who trades this stock now: AI-infrastructure speculators reacting to contract news and delay rumors in real time.

Riot's story is steadier but carries the same lesson. Its Anthropic lease is the biggest single AI deal in the sector, and Chardan Capital raised its price target on Riot from $27.50 to $32.00 on November 5, keeping a Buy rating. Riot closed November 6 at $16.71, up 13.2% on the day — yet still down 30.7% over six months. Even the sector's best AI deal cannot fully offset the drawdown, because the whole group got re-rated at once.

Number 4: -60% — What the "Pure" Miners Actually Lost

Strip out the AI darlings and the damage is worse. MARA Holdings closed November 6 at $6.35 — down 38.7% in one month, down 50% in six months, down 60% in one year. Its 52-week range is $6.37 to $18.69, so it spent the week plumbing its yearly low. CleanSpark closed at $6.16 — down 46.5% in one month, down 55.9% in six months, down 60.4% in one year. IREN, despite the 20% Friday bounce, closed at $21.64 — down 44.1% in one month, down 61.9% in six months, down 65.3% in one year, with a beta of 4.45. Side-by-side data is on MarketBeat's CLSK–RIOT–MARA comparison.

The table below uses November 6 closing prices. These stocks did not behave like Bitcoin at any horizon.

Company	Price (Nov 6 close)	1 month	6 months	1 year	Market cap
IREN	$21.64	-44.1%	-61.9%	-65.3%	$9.9B
Riot Platforms (RIOT)	$16.71	-9.9%	-30.7%	-1.8%	$5.1B
MARA Holdings (MARA)	$6.35	-38.7%	-50.0%	-60.0%	$2.8B
CleanSpark (CLSK)	$6.16	-46.5%	-55.9%	-60.4%	$2.1B
Bitcoin (BTC)	~$76,000	—	—	~ -25%	~$1.5T

Two things stand out. First, Riot held up far better — down just 1.8% in a year — because the market credits its Anthropic lease as real, contracted revenue rather than a hope. Second, MARA and CleanSpark, the two most "pure" miners, both lost about 60% in a year in which Bitcoin fell only about a quarter from its highs. The pure miners were supposed to be the leveraged Bitcoin bet. Instead they were the worst way to own the Bitcoin story in 2026.

There is also a mining-economics reason the pure miners are hurting. Hashprice — what one petahash per second of mining earns per day — is about $36, after a record low near $27.60 in June 2026, versus close to $300 a day in 2021. Miners earned $1.215 billion in October, but hashprice fell 17.5%, and the 2028 halving will cut the block reward in half again. That is the structural squeeze behind the AI pivot. If you want the full picture on whether mining still pays, read our guide on whether Bitcoin mining is profitable in 2026 and the November update on the hashprice squeeze.

Number 5: $1.22M vs $2.15M — The Megawatt Math That Decides Everything

The fifth number is the one the companies use to justify the pivot. One megawatt filled with modern miners earns about $2,150 a day, or roughly $0.8 million a year before power — and it moves with Bitcoin. Leased to AI, the same megawatt earns far more, and it is contracted: TeraWulf's Fluidstack deal works out to about $1.85 million per megawatt per year for 10 years, and Cipher's AWS deal to about $1.22 million per megawatt per year for 15 years. The AI rent is roughly one-and-a-half to two times the mining revenue, locked in for a decade or more, and it does not fall when Bitcoin falls. On paper, the pivot is obviously right.

But "on paper" is doing a lot of work. The AI rent only exists once the data center is built, the power is connected and the tenant moves in — and the market will punish any hint of delay with an 18% one-day drop, as IREN learned on November 5. Mining revenue starts the day the machines are plugged in. AI revenue starts years later. The companies are crossing a bridge between those two realities, and the share prices are pricing the bridge, not the destination. There is a second wrinkle: the AI tenants are themselves leveraged. CoreWeave's revenue runs at about $10.3 billion a year, but its debt-fueled growth model is exactly what the market repriced in October. The miners have swapped Bitcoin volatility for AI-credit volatility.

What Most People Get Wrong About Miner Stocks

Myth 1: "Miner stocks are a leveraged way to buy Bitcoin." That was true in 2021 and false in 2026. The correlation with Bitcoin has fallen to 0.40 while the correlation with CoreWeave has risen to 0.72. A leveraged Bitcoin bet that falls 46% while Bitcoin rises 22% is not leverage — it is a different asset. If you want Bitcoin exposure, a spot Bitcoin ETF is a cleaner tool than a miner stock that now trades on AI credit.

Myth 2: "The AI pivot makes miners safer." It makes them different, not safer. A 15-year lease is only as good as the tenant, and the tenant's credit is the new single point of failure. IREN's 18% drop on a Microsoft delay report is the proof: one customer, one rumor, one-fifth of the value gone.

Myth 3: "The 10–17% bounce on November 3 means the bottom is in." The group has bounced hard before and then fallen further. In the 21 trading days before that bounce, most miners lost 30% to 45%. A one-day 10% rally inside a 46% three-month decline is a relief rally until proven otherwise. Watch the next CoreWeave move, not the next Bitcoin move, for the signal.

Myth 4: "All miners are the same trade now." They are not. Riot is down 1.8% in a year; MARA and CleanSpark are down 60%. The spread is the market separating companies with signed, long-dated AI revenue from companies still mostly exposed to hashprice.

Myth 5: "Mining is dead, so the stocks are worthless." Also wrong. Mining still earns $1.215 billion a month across the network, and the 2028 halving is two years away. The miners that survive the transition own something genuinely scarce — permitted power capacity — and power is the bottleneck for both AI and mining. The question is whether the companies can survive the bridge to contracted AI rent without diluting shareholders into the ground.

Bull, Base and Bear: Three Scenarios for Miner Stocks

Bull case — the bridge holds. The AI leases start billing in 2027 and 2028 as promised, Microsoft and Anthropic pay on time, and Bitcoin stabilizes above $75,000. In that world, Riot's $28 billion Anthropic lease and IREN's $9.7 billion Microsoft deal get re-rated from "hope" to "contracted cash flow," and the group recovers a large part of the 46% decline.

Base case — choppy, stock-specific, Bitcoin-irrelevant. The group trades sideways with violent 10–20% swings on contract news and AI-credit headlines, while Bitcoin does its own thing between $68,000 and $80,000. Riot and TeraWulf hold up; MARA and CleanSpark keep bleeding on hashprice unless Bitcoin rallies hard. This is the most likely path, and it means stock-picking matters more than market-timing.

Bear case — the AI credit repricing continues. CoreWeave or another big AI tenant stumbles, a major lease gets delayed or renegotiated, and the market decides the 3.9 gigawatts of contracted capacity is worth far less than the valuations assume. The miners, with betas above 4, fall two to three times as fast as the market. If you own these stocks, this is the scenario your position size has to survive.

Checklist: 5 Questions Before You Buy Any Miner Stock

If you are considering a miner stock after the 46% fall, ask these five questions first. A "no" on any of them is a reason to wait.

• 1. How much revenue is contracted AI rent, and when does it start? Riot's $28 billion Anthropic lease and IREN's $9.7 billion Microsoft deal are real, dated contracts. Vague "AI pipeline" talk is not. If the rent starts in 2028, you are buying a construction project, not a cash flow.
• 2. Who is the tenant, and can they pay? One customer is one point of failure. Check the tenant's credit, not just the headline contract value.
• 3. What is the beta? IREN's beta is 4.45, MARA's is 5.42, CleanSpark's is 3.85, Riot's is 4.15. These stocks move four to five times as much as the market. Size the position for a 20% down day — November just delivered two in one week.
• 4. What happens if Bitcoin falls to $68,000? The pure miners' economics break first: hashprice is already near record lows and the halving is coming. Know the Bitcoin price at which your miner stops making money on mining at all.
• 5. Would you rather own Bitcoin itself? If your thesis is "Bitcoin goes up," own Bitcoin — or read our honest guide on whether Bitcoin is a good investment in 2026 first. Miner stocks are now a hybrid of both theses and a clean version of neither. And if you do buy Bitcoin directly, a plan beats a feeling: our guide to dollar-cost averaging into Bitcoin covers the approach that has historically beaten lump-sum timing, and our piece on what six past Bitcoin crashes did next puts the current 40% drawdown in context.

FAQs

Why did Bitcoin mining stocks fall 46% while Bitcoin rose 22%?

Because the miners stopped trading like Bitcoin. Over the last three months the miner basket's correlation with CoreWeave was 0.72 and with Bitcoin only 0.40. Investors now price miners on AI data-center contracts and AI credit. When AI credit repriced in October, the miners fell 11% while Bitcoin fell just 1.8%.

Which Bitcoin mining stock performed best in 2026?

Riot Platforms, down just 1.8% over one year to November 6, 2026, thanks largely to its 15-year, roughly $28 billion Anthropic lease covering 910 megawatts. By contrast, MARA and CleanSpark each lost about 60% over the same year, and IREN lost 65.3% despite a 20% single-day bounce on November 6.

Are Bitcoin miner stocks a good buy after the crash?

It depends on the individual company, not the group. The "bargain" case rests on contracted AI revenue starting to bill in 2027–2028. The trap case is that AI credit keeps repricing, leases get delayed, and the pure miners keep bleeding on record-low hashprice. Use the five-question checklist above before buying anything.

What is hashprice, and why does it matter for miners?

Hashprice is what one petahash per second earns per day — currently about $36, after a record low near $27.60 in June 2026, versus close to $300 a day in 2021. When hashprice falls, the least efficient operators get squeezed out. The 2028 halving will cut the block reward in half again — a major reason miners are pivoting to AI rent.

Do mining stocks still follow the Bitcoin price?

Barely. The correlation of the WGMI miners fund with Bitcoin fell from 0.67 in 2023 to 0.44 in 2026, while its correlation with chip stocks rose from 0.34 to 0.69. Bitcoin explains only about 16% of the miners' daily moves; CoreWeave alone explains 52%.

What should I watch next for miner stocks?

Three things: the November 10 CPI print and the December 9 FOMC meeting; any news on the big AI leases, especially IREN's Microsoft build-out and Riot's Anthropic project; and MARA's Q3 earnings, expected around November 9. A delay headline on any major AI contract has moved these stocks 15–20% in a day.

Conclusion

Bitcoin mining stocks fell 46% in three months while Bitcoin rose 22%, and the gap is not a mistake the market will quickly correct. It is a repricing: these companies are now AI landlords with Bitcoin side-businesses, trading on AI credit, tenant risk and construction timelines. IREN's week — down 18% on a Microsoft delay report, up 20% the next day — is the new normal. The opportunity is real for companies with signed, long-dated AI revenue, but that revenue bills in 2027 and 2028, not today — and the bridge between now and then is where the 46% went. Before you buy the dip, decide which thesis you actually hold — Bitcoin, or AI infrastructure — and buy the cleanest version of it instead. The miners, for now, are neither.

This article is for information and education only, not financial advice. Prices cited are November 6, 2026 closes unless stated otherwise. Do your own research before investing.