One key part of the AI trade has decoupled from the rest. Why that is, and whether it can come back
Industrials began 2026 trading in tandem with chipmakers — the posterchildren of the artificial intelligence trade. That’s now far from the case. The S & P 500 industrials sector ETF (XLI) is up just 8.8% this year through Thursday’s close, while the VanEck Semiconductor ETF (SMH) has rallied 64%. A look at their year-to-date performance shows how closely the two traded until mid-April. Since April 14, XLI is down nearly 3%, while SMH has soared 33%. Several drivers contributed to this divergence, including elevated valuations for industrials, investors turning cautious on data center-linked names along with declines in defense stocks. However, this underperformance may present an opportunity to get into industrials. “It’s a pretty good risk-reward,” Nicholas Colas, co-founder at DataTrek Research, told CNBC on Thursday. He added that industrials should “do as well as the market” or outperform going forward. Triple threat to industrials A challenge for the sector is its valuation. Industrials trades at 23 times forward earnings, a premium to the S & P 500’s multiple of 19, FactSet data shows. The sector also commands a premium to its 10-year average of around 20. The valuations become even more extreme when looking at individual names. Caterpillar, for example, trades at a forward PE of 28, while Nvidia — the stock at the center of the AI bull story — has a ratio of 18. Citi believes industrials’ high valuations reflect “elevated growth expectations” relative to its historical valuation, the bank said in a Sept. 14 note. Another source of industrial’s underperformance is recent weakness in defense stocks, a core group influencing the sector’s strength, according to DataTrek. The iShares U.S. Aerospace & Defense ETF (ITA) is down 10% over the past three months. The moves comes as traders assess a volatile geopolitical outlook. President Donald Trump said this week a deal to end the war with Iran could be reached after the November midterm elections. To boot, investors have also grown more cautious about the data center construction recently as projects face greater scrutiny over energy costs, power requirements and effects on local communities . The pushback has weighed on companies expected to supply the equipment as well as construction services needed for the buildout. Caterpillar, is down 23% over the past three months while other core holdings such as GE Vernova and GE Aerospace are down 12% and 13.7%, respectively. That underperformance is notable because industrials have emerged as an extension of the AI trade. The S & P 500 Industrials sector has doubled since the end of 2022, adding $2.6 trillion in market value, according to Melius Research’s Sept. 11 report on the sector. However, 57% of that increase came from companies exposed to AI and aerospace markets – both areas that have recently come under pressure. A shift in investor positioning has also amplified the decline. Bank of America analysts said industrials recorded its “biggest outflows,” where four-week average flows had reached record lows, analysts said in an early September research note to clients. The bank noted that industrials had become one of the most “crowded” and “expensive” sectors, leaving the group vulnerable. However, there’s light at the end of the tunnel for investors. In fact, they could see marked move higher in the near future. A comeback in the cards? Colas believes renewed strength in industrials should come from upcoming third quarter earnings and forward guidance from companies in the sector, which will come next month. He believes if industrial companies show evidence of strong backlogs and reliable earnings streams, “a valuation premium is warranted.” Industrials have historically been “a fairly low valuation group,” said Colas, who has been covering the sector for 40 years and is familiar with how it trades. He explained that, throughout the years, their growth has come from businesses becoming more global, while maintaining relatively fixed costs. However, the AI trade has led to a “burst of energy” within the sector. Colas predicts industrials, particularly those tied to the AI theme, will get an “ever increasing multiple” because the group now has “a base of business that is so large they could actually move the needle and create a much more stable stream of earnings over the long run.” Moreover, DataTrek expects some of the pressure around data center development to subside. “The very visible pushback on AI data center construction, which has clearly hurt sentiment on the group, should fade once U.S. midterm elections are over,” the firm said, noting that long-term demand remains intact, and is “worth a look” at current oversold levels. The firm considers that over the long-term industrials are supported by the “ongoing durability of the U.S. economy,” which has been faring well as evidenced by strong earnings, consistent spending and a solid labor market. While the industrials sector may not be moving lockstep with the AI trade at the moment, the AI buildout, and the companies supply it is far from finished and companies in the industrials sector could be poised to benefit. Ways to play For investors looking to add exposure to industrials, they can do so via the XLI ETF or through individual companies analysts expect will do well. CNBC Pro screened the XLI for stocks that have buy ratings from at least 55% of analyst covering it and have an average 12-month price target that implies upside of 35% or more. Among the stocks that made the list are Axon Enterprise , Boeing and logistics giant C.H. Robinson . Both Boeing and C.H. Robinson are expected to rally more than 36% over the next 12 months. Axon’s average price target, meanwhile, signals a gain of more than 62%. Howmet Aerospace and United Airlines also made the list. “While we would not be surprised by some continued near-term stock volatility given a combination of factors such as elevated/rising long-term interest rates, questions regarding the pacing of AI spend, especially ahead of US mid-terms, and typical seasonal trading patterns, we do believe AI/data center-related tailwinds will remain resilient, and given solid signs of shorter-cycle momentum continuing to build, we remain constructive in our outlook for Multi-Industrials heading into late ’26,” Citi analyst Andrew Kaplowitz wrote to clients.