The AI Rally is Narrowing: Why Market Internals Look Eerily Like the 2000 Dot-Com Peak
Table of Contents
A Record High Built on a Narrow Base
The S&P 500 recently closed at a record high, but the victory is superficial. Beneath the surface of the headline index, a stark divergence has emerged: while the index is climbing, the number of stocks actually participating in the rally is shrinking. This phenomenon, known in trading circles as ‘poor breadth,’ is creating a market structure that bears a striking resemblance to the peak of the dot-com bubble in March 2000.
On a critical trading day at the end of May, only 20 members of the S&P 500 hit new all-time highs. Even more telling was the composition of that group; just seven of those 20 companies had no direct connection to artificial intelligence. The rest were the primary beneficiaries of the current generative AI gold rush, suggesting that the broader market is being dragged upward by a tiny cluster of high-flyers rather than a healthy, diversified recovery.
The Ghost of March 2000
The historical parallel is not accidental. Michael Hartnett, a prominent strategist at Bank of America, noted that the exact same pattern occurred at the absolute zenith of the internet bubble 24 years ago. In March 2000, the market saw a similarly narrow spike where only 20 stocks were hitting record highs just before the collapse.
While the underlying technology—LLMs and neural networks versus the early days of commercial web browsers—is vastly different, the psychology of the trade remains identical. Speculative price action is currently concentrated in the ‘picks and shovels’ of the AI era: semiconductors and memory chip manufacturers. The surge has been violent and concentrated, with Micron Technology jumping 88% in a single month, SK Hynix rising 81%, Samsung climbing 44%, and AMD soaring 46%.
This concentration has pushed the Nasdaq Composite into its most aggressive two-month stretch in over two decades, spanning April and May. However, when the gains are this concentrated, the index becomes fragile. If the few stocks propping up the market stumble, there is no broad-based support to catch the fall.
Analyzing the Market Internals
To understand why analysts are nervous, one has to look at the ‘advance-decline’ lines—the ratio of stocks rising versus those falling. These lines peaked in late March but have been trending downward since mid-April, a classic bearish signal indicating that the rally is losing steam among the majority of companies.
Technical analysis from Oppenheimer’s Ari Wald suggests that market internals have lagged significantly since the initial April surge. This is corroborated by data from BCA Research, which found that as of May 20, only about 55% of S&P 500 constituents were trading above their 200-day moving average. In a healthy bull market, that number is typically much higher.
BCA strategists, led by Arthur Budaghyan, argue that while the headline indexes look strong, this narrowness is a sign of “underlying stock market vulnerability.” Essentially, the market is flying on one engine while the other three are stalled.
The Pivot to Defense
The catalyst for a potential correction, according to Hartnett, will likely not be a failure of the technology itself, but the macro-economic environment. He points toward central banks and the persistence of high interest rates as the likely triggers that will pop the bubble.
Given the current trajectory, Hartnett is advising clients to transition to a defensive posture. His recommended “post-bubble roadmap,” based on market behavior since 1929, involves pivoting toward long bonds and sectors that were dramatically underperformed during the final months of the bubble. For investors currently riding the AI wave, the warning is clear: the higher the peak, the more dangerous the descent when the narrow base finally gives way.
