Today's AI-fueled market frenzy is even more dangerously inflated than the 1990s internet bubble that ended with a historic crash in 2000.

That's according to Apollo's chief economist Torsten Slok. This time, it's not just hype-it's concentrated, institutional, and sitting atop the S&P 500's biggest names, he adds.

Top-Heavy Trouble

Back in 2000, the top 10 companies in the S&P 500 had a price-to-earnings (P/E) ratio of 25-lofty, but still loosely tethered to fundamentals. Today? That same top 10 group, now dominated by AI darlings like Nvidia Corp (NVDA  ), Microsoft Corp (MSFT  ), and Alphabet Inc (GOOGL  ) (GOOG  ), carries a P/E of 25 once again... but with far more weight.

What's different now, Slok warns, is that the rest of the market isn't anywhere near as stretched. The top-heavy S&P is showing signs of structural imbalance-making the AI bubble both narrower and potentially more dangerous.

While the overall S&P 500 P/E ratio sits at a seemingly reasonable 20, the ex-top-10 group has a P/E of just 20, indicating that the bulk of the valuation risk sits squarely on the backs of a few AI titans. If sentiment cracks, the fallout won't be gradual-it'll be violent.

How To Hedge Or Harvest The AI Mania

For investors seeking to navigate the storm, ETFs provide a practical way to express their views. If you believe the bubble still has room to inflate, the Global X Robotics & Artificial Intelligence ETF (BOTZ  ) gives exposure to a broad range of AI-focused companies.

But if you're feeling the chill from Slok's warning, consider defensive hedges like the Invesco S&P 500 Equal Weight ETF (RSP  ), which dilutes mega-cap exposure, or ProShares Short QQQ (PSQ  ) to bet against the froth.

With valuations this stretched and history offering a sobering mirror, the AI trade might not end in a bang-but it could still blow up.