While retail traders debate which AI stock is the next Nvidia, institutional money is doing something unexpected: rotating into the most boring companies on the planet. Railways. Commodity producers. Defence contractors. The kind of businesses that haven’t been exciting since your grandfather’s portfolio.
The logic is disarmingly simple. If AI really does replace most knowledge work, then the companies most at risk are the “asset-light” darlings that dominated the last decade — consulting firms, software companies, financial services. The businesses that can’t be disrupted by a large language model are the ones that move physical things in the physical world. You can’t replace a copper mine with ChatGPT.
Goldman Sachs data shows a clear rotation into “hard asset” equities over the past several months. Defence stocks are up 11.7% year-to-date versus 0.4% for global equities. Energy and commodity names are outperforming. Infrastructure plays are catching bids. It’s the mirror image of the AI hype trade — and the smart money is on the other side.
But this isn’t just an AI fear trade. Three structural tailwinds are converging. First, Europe’s massive rearmament push is creating decade-long demand for defence and heavy industry. Second, the Iran conflict has exposed the fragility of global energy supply chains, pushing investors toward companies with “constrained supply” — assets that can’t be easily replicated. Third, governments worldwide are spending on infrastructure at levels not seen since the post-war era.
The team at Ruffer Investment Company calls it a shift toward owning “constrained sources of supply” as a hedge against inflation and geopolitical shocks. In a world where energy can be weaponized overnight and supply chains can be severed by a single missile, owning hard assets isn’t just a defensive play — it’s a bet on how the world actually works.
This trend has legs precisely because it’s boring. For over a decade, Wall Street told investors that capital-light was superior — low capex, high margins, infinite scalability. That consensus is fracturing. The companies that own real things in the real world might be entering a multi-year rerating cycle. And the best part? They’re still cheap because nobody’s been paying attention.