Here's a funny thing about artificial intelligence: everyone thought it would be a rising tide that lifted all tech boats. Instead, it might be redrawing the entire map of what makes a winning stock in today's market.
Over the last six months, something interesting has been happening. Investors haven't been throwing money at anything with "AI" in the name. Instead, they've been quietly rewarding industries that have one thing in common: they're not that worried about AI taking their jobs. Or more precisely, their business models aren't built on jobs that AI can easily automate.
The result is a growing performance gap that's pretty counterintuitive at first glance. On one side, you have asset-heavy, goods-producing sectors—the companies with rigs, mines, and factories. On the other, you have asset-light, service-oriented industries—the companies with lots of people in offices. And right now, the market is favoring the former in a way that goes beyond your typical economic cycle explanations.
A new analysis from Goldman Sachs suggests this isn't random. It's the market starting to price in a fundamental question: which companies get disrupted by AI, and which ones get to keep doing what they're doing?
The Labor Exposure Screen: Who's Got Skin in the Game?
Goldman's analysts did something clever. They created a company-level metric to estimate exposure to AI automation. They looked at job functions, overlaid them with task-level measures of what AI can actually do, and then combined that with each firm's labor-cost-to-revenue ratio. What you get is a forward-looking gauge of how vulnerable a company's wage bill might be to getting AI-ed away.
The industries that came out looking most exposed read like a who's who of the modern service economy: software, communication and professional services, consumer services, banks and financial services, media and entertainment. These are businesses where people—and their salaries—make up a big chunk of the cost structure.
On the other end of the spectrum, you have sectors that look… well, more old economy. Retail goods, energy, equity REITs, tobacco, autos, food and beverage, materials, utilities, and semiconductors. Their labor-cost share relative to revenue is significantly lower, which means they're less directly exposed to the risk that AI starts doing their employees' jobs.
This is the double-edged sword of AI that the market is starting to price. Yes, AI can boost productivity. But it can also compress margins or displace white-collar labor in ways that create real uncertainty about future cost structures and earnings visibility. And right now, the market appears to be treating high labor exposure as a risk factor. It's not just about who can use AI; it's about who might get used by AI.
The Asset Heavy Advantage: Moats You Can Touch
Goldman's second screen looks at something more tangible: asset intensity. They define this as assets (excluding cash and intangibles) relative to revenues. In plain English: how much stuff do you own versus how much money you make?
Industries that require significant physical capital—the rigs, mines, machinery, and infrastructure of the world—tend to have higher barriers to entry. You can't just code up an oil field. They also have lower near-term automation risk because, well, robots still need to be bolted to something.
What's interesting is that in recent months, these asset-heavy businesses have sharply outperformed their asset-light peers in a way that macro conditions alone don't fully explain. The divergence has been particularly pronounced within technology, media, and telecom (TMT), where names linked to physical infrastructure have fared far better than software-centric firms. Goods-producing companies have also outperformed services firms, reinforcing the same theme: the market likes stuff you can kick.
The Sector Scoreboard: Winners and Losers
Let's look at the numbers. Year to date through February 25, the capital-intensive sectors are leading the pack:
Meanwhile, the asset-light and service-oriented sectors have been lagging:
The divergence becomes even more striking when you zoom in to the industry level.
Top performers look like they're from a different investing era:
Underperformers, on the other hand, read like a list of modern service economy darlings:
The contrast couldn't be clearer: hard-asset, commodity-linked, and capital-intensive industries are leading, while software, internet, and financial services are lagging. It's not that investors don't like technology—it's that they're being picky about what kind of technology exposure they want.
What's Really Going On Here?
So what is this divide signaling? Investors appear to be making a fundamental differentiation between companies anchored by tangible assets—things you can see, touch, and that are hard to replicate—and those heavily dependent on human capital that might be vulnerable to AI disruption.
Think about it this way: if AI materially reduces the economic value of certain white-collar functions (and let's be honest, a lot of white-collar work is just moving information from one place to another), then service-heavy industries could face margin compression and competitive disruption. Their moats might not be as deep as they thought.
Meanwhile, asset-heavy businesses might benefit from structural advantages that are harder to replicate with code. You can't AI your way into owning a copper mine or an oil field. The barriers to entry are physical, not just intellectual.
The big question, of course, is whether this rotation proves temporary or structural. Is this just a moment in the market, or is it the beginning of a longer-term revaluation of what makes a business resilient in an AI-driven world?
For now, the message from the market is clear: AI isn't just driving a technology narrative. It's reshaping sector leadership in real time, creating a visible divide between what the market sees as winners and losers in this new environment. And somewhat surprisingly, the winners right now look a lot like the old economy.
