AI ate the bull market

The short version

For about a decade, “risk-on” meant that easy money would chase risky stuff: tech stocks, crypto, leveraged bets. Bitcoin and the Nasdaq tended to rise and fall together because the same speculators were piling into both.

In 2025 that broke. The marginal “risk dollar” — the next dollar of speculative money — started showing up overwhelmingly in a small basket of AI stocks (NVIDIA, Microsoft, Google, and a handful of others), in prediction markets like Kalshi, and in private secondaries like SpaceX. It stopped showing up in Bitcoin.

Below: seven indicators, each measuring this from a different angle. They all tell the same story.

Bitcoin and Nasdaq used to move together. Now they don't.

This is the simplest version of the whole story. We line up Bitcoin and QQQ (the Nasdaq-100 ETF) and pretend you bought $100 of each on January 1, 2018. The chart below shows what those investments would be worth over time.

For seven years the orange (Bitcoin) and blue (Nasdaq) lines danced together — same rallies, same drawdowns, same general shape. Starting in 2025 the blue line keeps climbing while the orange line goes flat. They aren't the same trade anymore.

The “does Bitcoin still care about stocks?” number is collapsing.

In finance, “beta” is a one-number answer to “when stocks go up, how much does Bitcoin go up?”. Beta of 1 means it tracks the market move for move. Beta of 2 means it moves twice as much. Beta of 0 means Bitcoin and the market are unrelated.

We measured Bitcoin's beta against the riskiest part of the stock market (high-beta tech stocks vs the broader market). For most of the last 8 years that number sat above 1 — Bitcoin was a leveraged bet on stock-market risk appetite. In March 2026 it was around 3.3. Today it's 0.74×. That's a roughly 80% collapse in three months. Bitcoin still cares about the stock market, but a lot less than it used to.

Bitcoin vs. Nasdaq: a 1-in-100 weakness reading.

Sometimes the simplest measure is the best one. We took the ratio of Bitcoin's price to QQQ's price and asked: where does this ratio sit in its own three-year history? A reading of 100% means “Bitcoin has never been stronger vs Nasdaq recently.” 0% means “Bitcoin has never been weaker.”

Today the reading is 0th percentile. That's near zero. In the entire three-year window, Bitcoin has rarely been weaker against the Nasdaq than it is right now. A year ago this number was 96. Six months ago it was 31. Now it's 1.

Speculative dollars are flowing into stocks, not crypto.

One way to spot “where is new money going?” is to compare two things: stablecoin supply (a proxy for cash sitting on crypto exchanges, ready to deploy) vs Nasdaq prices (which rise partly because money is being shoveled into ETFs and retirement accounts).

We subtract one from the other (each measured as a 90-day percentage change). When the line is positive, crypto is absorbing more new money than stocks. When it's negative, stocks are. The line has been negative for about two months now, and the latest reading is -24.3%. That's the most lopsided in favor of equity rails the data has shown all year.

Why is this happening?

The crypto-specific charts above tell us that something has changed. They don't tell us why. For that, we look at what's happening in equities. Three observations explain a lot.

The stock market has become extraordinarily narrow.

The S&P 500 is normally diversified across 500 companies. But today the index is being carried by maybe 7 of them — the “Magnificent 7” AI giants. We measure this by comparing the regular S&P (SPY, where bigger companies count more) against an equal-weighted version (RSP, where every company counts the same).

When SPY rises faster than RSP, it means a handful of giant companies are doing all the work. Today this measure sits in the 73th percentile of its three-year range. Every major equity unwind in modern history (2000, 2022) started from a concentration extreme. Concentration alone doesn't cause a crash — it can persist for years — but an index priced on the success of 7 stories has fewer independent legs than one priced on 500.

Speculators are leveraged up, but consumers are not happy.

Margin debt is money that investors borrow against their portfolios to buy more stocks. When margin debt grows fast, it's a sign that confidence is high and speculation is hot. Right now it's growing about +22.9% per year — not a record, but firmly expansionary.

But here's where it gets weird. Normally, when speculation runs hot, consumer confidence is also high — everyone is feeling good about the economy. Today, the University of Michigan's consumer sentiment reading is at 49.8. That's a multi-decade low. It's the kind of number you usually see during a recession, not during a stock-market rally.

What does it all mean?

Put the pieces together and you get a coherent — and slightly uncomfortable — picture of 2026:

• A small number of AI giants are running the entire stock market. (Concentration high.)
• Speculators are using leverage to bet on them. (Margin debt growing 20% a year.)
• The broader economy is not enjoying the party. (Consumer sentiment at recession levels.)
• Crypto, which used to absorb a chunk of risk-on speculation, is getting almost nothing. (Beta collapsed, BTC/Nasdaq at 1st percentile, stable flow turning negative.)

This is not the classic “everyone is euphoric and stocks are at all-time highs” market peak. It's narrower and stranger. Narrow rallies without broad participation historically unwind faster than broad ones, because they depend on a tiny number of narratives holding up — in this case, mostly AI capex economics.

For our regime classifier specifically, this argues for a tempered read. The macro panel sees risk-on (easy money, healthy credit). But the indicators in this post argue the macro panel is picking up something that the broader economy doesn't share and that crypto isn't absorbing. Wiring them into the composite would pull our headline reading back toward neutral — which is probably the right call.