For most of its history, Andreessen Horowitz made its name betting on software — the light, infinitely copyable stuff with fat margins. On Friday it put $1.1 billion behind the opposite bet. The firm's new Machine Age Fund is aimed squarely at the physical layer that artificial intelligence actually runs on: chips, memory, data centres, robots, and the cooling, power, materials and real estate that keep them alive.

It's a striking move for a house built on code, and the firm isn't shy about why. The goal, a16z says, is to "open the throttle and accelerate the physical buildout of AI." In its own words: "We need faster, more efficient systems. We need cheaper and higher-bandwidth memory... faster and more scalable interconnects... power efficient edge devices... and all the cooling, materials, electrical, and real estate build out to support them."

Why the money is moving to the boring layer

The logic is simple arithmetic. Frontier models are only as good as the compute underneath them, and that compute is running into hard physical limits — not clever-algorithm limits, but power, heat, memory bandwidth and land. The scarce resource in AI is no longer a smart idea in a notebook; it's a gigawatt of electricity and a building to spend it in. A fund aimed squarely at chips, interconnects, cooling and power stations is a bet that the next decade of AI value is won or lost in the parts of the stack you can trip over.

a16z frames the whole effort in near-manifesto terms, calling AI "the strongest tool ever developed for solving problems and bestowing abundance" and its advancement a "social and national imperative." Whatever one makes of the rhetoric, the capital is real, and it is one of the largest single commitments by a venture firm to the unglamorous end of the industry.

The buildout has a bill — and opponents

The timing is pointed. Only this week, the physical buildout a16z wants to accelerate was the exact thing UK politicians were fighting about, with the Greens calling to "slam the brakes" on new data centres over the water and energy demands driven by the AI boom. The Machine Age Fund is, in effect, the capital side of that argument: the money betting that the power grids, cooling plants and server farms will get built, fast, whatever the local objections.

That tension — abundance framing on one side, strained grids and drought-year water use on the other — is the real story of AI's next phase. The models get the headlines; the fight is increasingly over the concrete, the copper and the kilowatts. a16z has just put a very large number on which way it thinks that fight goes.