Here's a question that's been keeping investors up at night: has the AI stock rally crossed the line from exciting growth story to full-blown bubble? Wall Street can't seem to agree. Some analysts are waving red flags about excessive enthusiasm, while others insist the fundamentals still make sense.

Alpine Macro, an economic research firm owned by Oxford Economics, took a different approach in its 2026 outlook. Instead of declaring whether we're already in bubble territory, the firm asked a more interesting question: what if the real danger shows up next year, once the Federal Reserve starts cutting rates in earnest?

Why This Doesn't Look Like a Bubble (At Least Not Yet)

Chief global strategist Chen Zhao makes a compelling case that today's market lacks the telltale signs of speculative mania. For starters, investors are skeptical, not euphoric. The fact that everyone keeps asking whether AI is a bubble is itself evidence of caution rather than the blind optimism that typically precedes a crash.

Sentiment surveys back this up. The bull-bear spreads today show relative balance, nothing like the late 1990s when bullishness absolutely dominated market psychology.

"Investor sentiment is neutral. Bull-Bear spreads in the 1990s showed far more bulls than bears," Zhao wrote.

Then there's valuation. Sure, stocks aren't cheap, but they're not completely detached from reality either. Alpine Macro used the Fed model, which compares equity earnings yields with long-term bond yields, and found that the equal-weighted S&P 500 is roughly 25% undervalued. The cap-weighted index looks fairly valued, while the Magnificent Seven trade at a premium that reflects genuine growth expectations rather than pure speculation.

The contrast with the dot-com era is stark. In 2000, growth stocks traded at more than double their bond-implied fair value. Cisco Systems Inc. (CSCO), the biggest stock in the market at the time, carried a triple-digit price-to-earnings ratio. Today, Nvidia Corp. (NVDA) trades at about 30 times forward earnings, supported by profits that are actually growing rapidly.

Capital discipline is another key difference. Alpine Macro sees little evidence of the kind of reckless overinvestment that defined the late 1990s. Major tech companies have largely funded their AI spending through internal cash flow instead of taking on debt. IPO activity remains subdued, earnings are expanding, and markets have shown a willingness to punish companies that spend too aggressively.

"The [AI] investment boom is in its early stages, especially compared with the surge during the internet boom," Alpine Macro said.

The Missing Ingredient: Cheap Money

So if everything looks reasonably healthy, what's the problem? Alpine Macro points to economist Charles Kindleberger's framework for understanding asset bubbles. According to Kindleberger, you typically need three ingredients: a transformative technology, broad access to speculation, and cheap money.

AI clearly checks the first box. Access to speculation through zero-commission trading and derivatives? Check. The only thing missing is cheap money, and that's where 2026 comes in.

"The only missing ingredient is cheap money. But the Fed is gearing up for easing in 2026," Zhao said.

If inflation falls faster than expected, the Federal Reserve could shift toward more aggressive rate cuts in the second half of next year. That would unlock a staggering amount of capital currently sitting on the sidelines. There's roughly $13 trillion parked in money market funds and demand deposits right now, equivalent to about 20% of total U.S. stock market capitalization.

"A massive pool of dry powder is waiting. Money market funds plus demand deposits total $13 trillion—about 20% of U.S. stock market cap," Zhao added. Meaningful rate cuts could push that cash into risk assets.

What do markets expect? According to the CME FedWatch Tool, financial markets are currently pricing in just over two Federal Reserve rate cuts by the end of 2026. Data from Polymarket show similar expectations, with the highest odds assigned to a three-rate-cut scenario at 22%, followed closely by two rate cuts at 20%. Most of that easing is expected in the second half of the year.

The timing matters for another reason: leadership transition. Chair Jerome Powell's term expires in May, and the two candidates President Donald Trump has mentioned most frequently as potential successors, Kevin Hassett and Kevin Warsh, are widely viewed by markets as more dovish than current leadership.

Alpine Macro isn't predicting that an AI bubble is inevitable. What they're saying is that the risk is rising, and the Federal Reserve will likely play the central role in determining whether AI's boom stays grounded in fundamentals or morphs into something more dangerous. It's a useful reminder that sometimes the biggest risks aren't what's happening today, but what could happen once conditions change.