The rally in AI and semiconductor names has revived a familiar question: is this a durable growth cycle, or the late stage of a bubble. Dan Niles argues the market looks closer to 1997 internet infrastructure expansion than the 1999 blowoff phase. Paul Tudor Jones says the AI bull run is roughly 50% to 60% complete and could last one to two more years. Michael Burry, by contrast, warns that current conditions resemble the period just before the 2000 dot-com collapse.
The author of the source piece does not try to settle that debate. He takes a narrower question instead: if an investor believes a bubble may be forming, how should they short it without getting run over by the move first.
Why shorting a bubble can be harder than spotting one
The article argues that the main problem is not simply being wrong. Even a correct thesis can fail if the trade is entered too early and gets squeezed. In a bubble, prices can rise in a way that makes short exposure expand rapidly as the asset keeps climbing. A long investor who sells can step aside. A short seller who exits still has to buy back the position, and that pressure rises with every leg higher.
The author points to cases such as the 2008 Porsche-Volkswagen squeeze and the GameStop episode. In very high-volatility setups, options can also become so expensive that a pure directional bet loses much of its appeal. From that, the piece develops three approaches: find a “wedge,” short the “victims,” or wait for confirmation.
The “wedge” is an external force that can break the valuation story
A wedge is not a direct short on the hottest asset. It is the outside factor that can puncture the pricing framework supporting the move. In the article, rising inflation and higher rates are presented as the leading candidates. Assets priced on distant future cash flows are especially sensitive to discount rates. If those rates move up, valuations built on optimistic projections for later years can compress fast.
The author says this setup has one major advantage: timing does not need to be perfect. The bubble does not have to burst immediately. It may only need to stop accelerating for a quarter before highly levered assets start to crack. He mentions Canadian banks trading around 3 times book value and broader credit markets, including private credit, as areas worth watching under that framework.
The weaker neighbor may be a better short than the bubble itself
The second approach is to short the “victims” rather than the center of the frenzy. These are the companies or assets tied to the same environment but carrying much more fragile balance sheets. Evergrande is used as the clearest example. In that case, the better target was not necessarily the entire Chinese banking system. It was the heavily levered developer that depended on pre-sales and could break even if property demand only slowed modestly.
The article also looks back at airlines before the pandemic and financial stocks in 2007 and 2008. They were not always the bubble itself, but they offered highly asymmetric downside once conditions turned. The author’s point is simple: do not step in front of an exponential move if there is a related asset that can fall faster and farther.
Waiting for confirmation is the hardest part
The third method relies on discipline more than conviction. The author says many traders fail on the short side because they enter before the market actually turns. Instead of fighting every leg higher, he prefers to wait for a visible break in trend, the kind of support failure that suddenly gets shared across trading circles because the chart no longer looks healthy.
He also highlights two things to monitor beyond the chart itself: changes in rates and inflation, and shifts in correlation. If an asset suddenly becomes sensitive to a macro factor it had previously ignored, that can signal the market’s old playbook is starting to break down.
His own trades: more SPX and HYG shorts, plus Treasury put spreads
In the practical section, the author says that in the early hours of May 13 Beijing time, before the selloff deepened, he had already put on some hedges. He then increased short exposure to the S&P 500 by 5% and to high-yield bonds through HYG by 10%, while also buying short-dated put spreads. He did not short semiconductors, saying the core demand picture was still intact and the uptrend had not broken.
Instead, he added bond shorts through U.S. Treasury put spreads and sold 5% of his Canadian bank holdings. If the trend line holds and the market rebounds, he treats that as a limited cost for protection. If the trend fails, the cash and hedge positions would leave room to size into more specific short ideas later.
The piece does not end with a firm call that the market is already in a bubble. Its main message is narrower and more tactical: do not automatically short the asset going vertical. Find the wedge, identify the likely victims, and wait for confirmation before pressing the trade.