Andreessen Horowitz announced on Aug. 28 that it has raised an $1.1 billion fund dedicated to what it calls the Machine Age: physical infrastructure for artificial intelligence. The fund will invest across chips, memory, data centers, robotics, cooling, power and real estate, a departure from the firm’s long-standing bet that software would dominate the technology economy.
The fund’s name is deliberate. The firm’s founders built their reputations on the thesis that software eats the world, and their early bets on internet companies made them one of Silicon Valley’s most influential venture firms. The Machine Age fund acknowledges that the next phase of the industry’s growth will be built in factories, data centers and power plants rather than in code alone.
Partners at the firm said the fund will invest at all stages, from early startups designing specialized chips to later-stage companies building data center capacity. The firm has already made AI infrastructure investments through its general funds, and the dedicated fund allows it to write larger checks and move faster in a market where capital requirements are enormous.
The timing reflects the economics of the AI buildout. Training and running large models requires computing infrastructure at a scale the industry has never seen, and the companies selling that infrastructure have become the AI boom’s biggest winners. Nvidia’s dominance of AI chips has made hardware the most profitable part of the technology industry, and venture firms that once avoided capital-intensive businesses are following the money.
The fund also signals a shift in how Silicon Valley thinks about risk. Hardware startups require more capital, take longer to mature and fail more expensively than software companies. But the returns available to the winners, as Nvidia and the data center builders have shown, are large enough that a growing number of investors are willing to accept the risk.
a16z’s partners said the fund will focus on what they call the physical constraints of AI: power, cooling, land and manufacturing capacity. AI data centers consume electricity at rates that strain local grids, and the companies that solve those problems, from power generation to liquid cooling, have become essential to the industry’s growth. The fund is structured to capture value across that stack.
The firm has already backed companies in the space, including data center operators and chip startups, and the new fund gives those efforts a dedicated mandate. Its size, $1.1 billion, is large by venture standards but small relative to the capital needs of the companies it targets, and the firm is expected to co-invest with other funds and strategic partners.
The move has implications for the broader venture industry. Other firms have created AI-focused funds, but most have concentrated on software applications, which are cheaper to build and faster to scale. a16z’s bet on physical infrastructure is a bet that the industry’s most durable value will accrue to whoever controls the means of production, a view that echoes the firm’s earliest investing principles.
There are reasons for caution. Hardware cycles are long, and the AI infrastructure market could consolidate around a few dominant players, leaving late-stage investors with concentrated exposure. Power and real estate investments carry regulatory and environmental risks that software startups rarely face. The fund’s partners said they are prepared for those realities, and that the scale of the opportunity justifies the complexity.
The announcement also reflects a broader reordering of the venture industry. The largest AI companies have become so valuable that they can fund their own growth, and venture firms are searching for the next layer of opportunity. Infrastructure, once considered the unglamorous foundation of the industry, has become the prize.
The fund’s structure reflects the realities of infrastructure investing. Hardware companies need capital in tranches, often years before products ship, and the fund is designed to follow companies from early rounds through later rounds without losing ownership. The firm has also built teams with operating experience in semiconductors, manufacturing and energy, recognizing that diligence on a chip startup requires different expertise than diligence on a software company. Partners said the fund will make a limited number of large bets rather than spread capital widely, a structure suited to a market where the winners absorb most of the value.
For the founders who will pitch the fund, the message is clear: the firm that built its name on software is now writing checks for chips, power plants and robots. The Machine Age fund is a bet on the physical world, and on the belief that the AI industry’s next fortunes will be made where the machines are built. Whether that bet pays off will depend on power, land, chips and the patience of limited partners, but the firm’s partners are convinced the returns will justify the wait.
