Chip Stocks Are In a Bubble. Here’s More Good News.

Tech stocks aren’t in a bubble, but chip stocks are another story. (SeongJoon Cho/Bloomberg)

Key Points

• The current tech sector does not meet bubble criteria, as its two-year excess return to the S&P 500 peaked at 69%.
• Semiconductors, a tech subsector, do meet bubble criteria, with a five-year return of 470%, but this is half of the 900% level it reached in March 2000.
• On Oct. 30, 17.9% of the S&P 500 market cap was in a bubble, compared with 31.5% at the March 2000 peak.

By at least one measure, tech stocks aren’t in a bubble. The same can’t be said about semiconductors—but even that’s not as bad as it sounds.

The tech sector’s huge rally in recent years led many investors to worry about a bubble even before the November weakness. Even Nvidia CEO Jensen Huang has mentioned the B-word.

All the bubble talk led Ned Davis Research Chief Thematic Strategist Pat Tschosik to go looking for evidence of one, using the following criteria: Stocks had to have two years of price returns of more than 100% and beat the S&P 500’s return by more than 100% plus have a five-year return of more than 50%.

Using this definition, only two S&P 500 sectors going back 50 years have been in bubbles: energy in 1980 and tech in 1999-2000. Notably, the current tech sector returns don’t make the cut because its two-year excess return to the S&P 500 hasn’t exceeded 100%; instead tech’s two-year excess return to the S&P 500 peaked at 69% in early January.

That said, not all of tech is off the hook—among subsectors, semiconductors do pass the bubble test. “Semiconductors, famously known for boom and bust cycles, have accounted for three of the nine industry groups that met the bubble criteria, going back to 1991,” writes Tschosik.

Yet that’s not as bad as it seems. Back in their March 2000 dot-com peak, the five-year S&P Semiconductor returns exceeded 900%; by contrast, today the five-year return is 470%, just over half of that level. It’s true that’s very high on a historical basis but “there is precedence for S&P 500 Semiconductor returns heading even higher,” he writes. “While we can use this bubble criteria to identify industries and stocks at risk, it does nothing to tell you about when an industry will peak or burst (if it is truly a bubble).”

Although this definition was developed for top-down level analysis, applying it to S&P 500 stocks show that, on Oct. 30—the current cycle’s peak—5.8% of the index’s components also fell into bubble territory. That list is dominated by the usual artificial intelligence suspects including Nvidia, Broadcom , and KLA Corp., Lam Research , Micron Technology , Monolithic Power, and Advanced Micro Devices .

Nonetheless, that, too, is below the 9.2% of S&P 500 stocks that were at this level in 2000. Or put another way, on Oct. 30, “17.9% of S&P 500 market cap was in a bubble,” Tschosik writes. “At the March 2000 cycle peak, a record 31.5% of S&P 500 market cap met the bubble criteria, implying far more S&P 500 risk than the current environment.”

That leaves him with two main conclusions. The first is that whenever the “AI trade turns south, avoiding the datacenter-related semiconductor stocks could help portfolio managers outperform.” That echoes other strategists who think that the next wave of AI winners won’t be like the current one.

“Second, it feels too early to call the ‘bubble’ peak given how much investors have been worried about it,” notes Tschosik. “Still, we are late enough in the current semiconductor cycle to be on high alert for a major peak.”

Again, he isn’t the only one to argue that while we may be in a bubble, it’s nowhere near popping yet.

Write to Teresa Rivas at teresa.rivas@barrons.com