Here's a sentence you probably didn't expect to read in 2025: investors are dumping software stocks and piling into chemicals, freight companies, and farm equipment manufacturers. Yet that's exactly what's happening.

Software and AI-adjacent stocks stumbled right out of the gate this year, and the selling accelerated in February. The culprit? Fresh anxiety that artificial intelligence might actually disrupt demand for software products instead of supercharging it. Turns out, when AI tools can automate what your software does, that's not necessarily bullish for your subscription revenue.

The concerns hit a new pitch after announcements from Anthropic's Claude Cowork and Google's Genie 3 forced investors to confront an uncomfortable question: who actually benefits from the next phase of AI? The answers weren't reassuring for software bulls.

But here's what's interesting—this hasn't triggered widespread market panic. Instead, investors are simply rotating their capital elsewhere, and Goldman Sachs equity strategist Ben Snider says the destination is remarkably clear: sectors that offer insulation from AI disruption, not exposure to it.

That's a crucial distinction. Last year, AI exposure was treated like rocket fuel for any stock lucky enough to have it. This year, it might be more like kryptonite—at least if you're selling software that an AI agent could replace.

When the Dow Outperforms Silicon Valley

The rotation is showing up in some striking ways. Cyclical and consumer-linked industries have been rallying while software names get hammered. During one of software's worst weeks since the 2022 interest rate panic, the Dow Jones Industrial Average—that collection of industrials, transportation companies, and decidedly old-economy names—was marching toward all-time highs.

According to Goldman Sachs, the market is organizing itself around one defining characteristic: insulation from AI-driven productivity risk. In a note released Friday, the investment bank analyzed which industries offer the greatest protection from AI disruption. Their methodology? They measured which sectors have the weakest correlation with iShares Tech-Expanded Software Sector ETF (IGV)—essentially identifying the anti-software trades.

What they found tells you everything about where capital is flowing. The industries at the top of the list share a common thread: their revenues come from physical assets, real-world demand, and cyclical activity. Not software licensing, not data monetization, not subscription pricing power.

The AI-Insulated Portfolio

Here's what the escape routes look like, according to Goldman's analysis. Passenger airlines showed the weakest correlation to software at just 0.1, with U.S. Global Jets ETF (JETS) and Delta Air Lines (DAL) representing the space. Air freight and logistics came in at 0.2 correlation, captured by iShares Transportation Average ETF (IYT) and United Parcel Service (UPS).

Retail REITs and trading companies each registered 0.2 correlations as well. For retail property exposure, investors are looking at iShares U.S. Real Estate ETF (IYR) and Simon Property Group (SPG). The trading and distribution angle is represented by Industrial Select Sector SPDR Fund (XLI) and Fastenal Co. (FAST).

At the 0.3 correlation level, you'll find cargo ground transportation with iShares Transportation Average ETF (IYT) and FedEx Corp. (FDX). Commodity chemicals show up here too, tracked by Materials Select Sector SPDR Fund (XLB) and LyondellBasell Industries (LYB). Oil and gas refining lands at 0.3 correlation, accessible through VanEck Oil Refiners ETF (CRAK) and Marathon Petroleum (MPC).

The beverages sector appears at this level as well—distillers and vintners tracked by Invesco Dynamic Food & Beverage ETF (PBJ) and Diageo plc (DEO). Construction machinery weighs in at 0.3 correlation with Industrial Select Sector SPDR Fund (XLI) and Caterpillar Inc. (CAT). Diversified banks and food retail both register 0.3 correlations, represented by Financial Select Sector SPDR Fund (XLF) with JPMorgan Chase (JPM), and Consumer Staples Select Sector SPDR Fund (XLP) with Kroger Co. (KR).

Moving to 0.4 correlation territory, diversified chemicals lead the pack with Materials Select Sector SPDR Fund (XLB) and Dow Inc. (DOW). Oil and gas drilling shows similar numbers through SPDR S&P Oil & Gas Exploration & Production ETF (XOP) and Halliburton Co. (HAL). Building products register here with iShares U.S. Home Construction ETF (ITB) and Vulcan Materials (VMC).

Regional banks hit 0.4 correlation, accessible via SPDR S&P Regional Banking ETF (KRE) and PNC Financial Services (PNC). Integrated oil and gas companies like Exxon Mobil (XOM) show the same correlation, tracked by Energy Select Sector SPDR Fund (XLE). Hotel and resort REITs round out this tier with AdvisorShares Hotel ETF (BEDZ) and Marriott International (MAR).

At the highest correlation levels analyzed—0.5, still notably weak compared to software itself—you find steel represented by VanEck Steel ETF (SLX) and Nucor Corp. (NUE). Agricultural and farm machinery hits this level with First Trust Indxx Global Agriculture ETF (FTAG) and Deere & Co. (DE). Rail transportation rounds out the list at 0.5 correlation, captured by iShares Transportation Average ETF (IYT) and Union Pacific (UNP).

What It All Means

The pattern is unmistakable. These are businesses where AI can't easily eat your lunch. You need actual freight trains to move containers across the country. You need real steel to build infrastructure. You need physical airplanes staffed with human pilots to transport passengers. These aren't industries where a clever large language model can suddenly automate away half your workforce or replace your entire product with a better AI agent.

It's a dramatic shift from 2024's narrative, when simply having "AI" somewhere in your investor presentation was enough to catch a bid. Now, the market is making a more nuanced calculation about which businesses benefit from AI and which ones might get disrupted by it. Software companies that once seemed like obvious AI winners are being repriced as potential AI victims.

The Dow's strength amid software weakness tells the story perfectly. All those "boring" industrial, manufacturing, and transportation names suddenly look attractive when investors are worried about digital disruption. It's not that these companies are immune to technological change—it's that their business models are anchored in physical reality in ways that AI can't simply virtualize away.

Whether this rotation has legs or represents temporary panic selling in software is an open question. But for now, Goldman's data suggests that investors have a clear playbook: if you're nervous about AI disruption, buy the stuff that AI can't disrupt. Chemicals, not cloud platforms. Railroads, not SaaS subscriptions. Steel beams, not software licenses.

Who would have thought that in 2025, the flight to safety would lead investors straight into freight logistics and oil refining?