• Bitcoin dropped to around $76,000 as AI-linked shares sold off, reversing the bounce that followed the midterms.
  • Roughly $1bn of leveraged crypto positions were closed out automatically, which speeds a fall up rather than starting it.
  • CoinDesk is our only account of the day’s trading and we haven’t seen a primary source, so treat the precise figures as provisional.

What moved bitcoin this week had very little to do with bitcoin. Shares in the big AI names fell hard, the wider US market went with them, and bitcoin slid to around $76,000, handing back the bounce that followed the midterm results.

Bitcoin is still widely described as something that moves on its own, away from the stock market and away from whatever is worrying fund managers that week. On days like this one it clearly doesn’t. If you hold bitcoin and you also hold a tech-heavy index fund, or a pension weighted towards large US technology companies, there’s a reasonable chance you’re holding one bet twice. That’s a useful thing to know before a fall rather than during one.

Why bitcoin and tech shares fall together now

The link isn’t mystical. It comes down to who the marginal buyer is.

Since spot bitcoin arrived, buying bitcoin has meant pressing the same button you press to buy a share. That brought in a category of money that thinks in portfolio buckets: a slice for bonds, a slice for equities, a slice for the higher-risk end of things. Bitcoin sits in that last bucket alongside unprofitable tech and anything else with a story attached. When a manager decides to carry less risk, the whole bucket gets trimmed at once, and nobody stops to consider bitcoin’s monetary properties on the way.

Bitcoin also trades every hour of every day, while the stock market does not. That makes it the convenient thing to sell when a position needs reducing in a hurry, which tends to exaggerate its share of the move.

This correlation isn’t a permanent feature. It has loosened before, usually when something specific to crypto has been driving prices. It just happens to be tight at the moment, and it has been tight for most of the period since the ETFs opened the door.

What a billion dollars of liquidations actually means

Plenty of people trade crypto with borrowed money. You put up a deposit, the exchange lends you the rest, and your gains and losses are multiplied accordingly. The arrangement comes with a condition: if the price moves far enough against you, the exchange closes the position itself to protect the loan. You don’t get a phone call.

Those forced closures are sales. Sales push the price down. A lower price triggers the next set of forced closures, and so on down the ladder. That is the mechanism behind a figure like $1bn in a session, and it’s why a fall that began in US equity markets can land harder in crypto than in the thing that caused it.

A billion dollars is a real wave rather than a historic one. The market has had multi-billion-dollar days. The number is worth reading as a measure of how much leverage was sitting there, not as a measure of how many people decided bitcoin was finished.

The miners are inside the AI trade too

Focused detail of a modern server rack with blue LED indicators in a data center.
Server racks in a data centre: several bitcoin miners now rent out capacity like this to AI firms, tying them to the same trade as tech shares. Photo by panumas nikhomkhai on Pexels.

The overlap isn’t only about investor sentiment. It’s increasingly physical. IREN, one of the listed , reported bitcoin mining revenue down 58% while its AI cloud revenue rose 182% to $200m in the past quarter, and it plans to exit mining entirely by the end of the year, according to The Block.

Warehouses, power contracts and high-end chips serve both businesses. Several of the companies that ordinary investors think of as bitcoin exposure are now substantially AI infrastructure companies, which means a bad day for AI valuations is a bad day for them twice over.

One session, and what it is worth

We’ve made this point before about ETF flow headlines and it applies just as neatly to a price drop: change the window and you change the story. A fall to $76,000 reads as a rout over a day, as a wobble or a drawdown over a month, and as something else again over a year. We’re not going to attach figures to those longer windows here, because the reporting we have covers the session rather than the trend.

CoinDesk also quotes a trader at Wintermute, the market maker, taking the more optimistic reading, on the basis that crypto held up better than the AI names did. We haven’t seen the full remarks, so we’re not going to reconstruct the argument from a headline.

What to watch

Whether the correlation holds once AI shares find a floor. If bitcoin keeps sliding after tech stabilises, something else is going on and the AI explanation stops working.

ETF flows over the next couple of weeks, because they show whether institutions were actually selling or whether this was mostly leverage unwinding on thin weekday liquidity. Those are different situations with the same price chart.

And the miners. IREN is the clearest case so far, but if more of them complete the move into AI hosting, the listed companies people use as a proxy for bitcoin will stop being one. More on the day’s trading across the fifteen coins we track sits in Markets.