The Atlantic reports that scaling AI now depends on concrete, copper, and megawatts more than code. Data centers, electricity grids, construction equipment, and industrial suppliers have become the bottleneck. Caterpillar's stock surge signals the shift: companies supplying the physical backbone for server campuses are cashing in alongside Nvidia, OpenAI, and the cloud giants.
This reorders who holds power in tech. If AI depends on land permits, grid capacity, and equipment that only a handful of suppliers can provide, then utilities, industrial manufacturers, and data-center landlords become gatekeepers. A startup with a brilliant model still cannot scale without someone else's power infrastructure and turbines. That dependency shifts negotiating leverage away from Silicon Valley.
Winners: industrial equipment makers, energy producers, data-center operators, and incumbent tech firms with balance sheets to spend at scale. Losers: bootstrapped startups, communities absorbing power and water tradeoffs, and the 20-year-old myth that talent and code alone are enough to compete.
Silicon Valley has lost its biggest advantage. For two decades, the region sold the story that software floats above the physical world. AI just demolished that. The next moat is procurement. Your breakthrough model means nothing without grid access, and grid access means dealing with Caterpillar, not just Sequoia.
Filed to the Technology desk · Jul 18