a16z launches $1.1B Machine Age fund for AI hardware

Andreessen Horowitz, a firm built on software bets, has raised a $1.1 billion fund dedicated entirely to the physical infrastructure beneath AI - chips, memory, networking, storage, data centers, robotics, and AI appliances^[2]. a16z frames this as a response to a change in what's scarce: for most of the past decade, capital chased talent and distribution, but the constraint has now moved to the physical stack itself^[4]. That shift shows up in the firm's own numbers - hardware bets, once a negligible sliver of activity, now account for more than 20 percent of a16z's deal flow^[2]. Partners on the a16z Show separately described hardware-focused deals among top founders climbing from roughly 5 percent to over 20 percent of flow in about a year, a figure that lines up with the firm's own disclosure. Five partners - Ben Horowitz, Martin Casado, Raghu Raghuram, David Ulevitch, and David George - put their names to the launch, signaling a firm-wide commitment rather than a single team's side project^[2].

a16z's thesis centers on the idea that AI's bottleneck has moved from the models themselves to the physical stack underneath - chips, memory, power, and interconnects^[1]. The firm's own figures show the scale of that shift: compute density per server rack has grown 28-fold moving from an Nvidia H100-generation rack to a Rubin-generation rack, while rack power draw has climbed from 5-10 kilowatts to a current range of 100-250 kilowatts, with a path toward 1 megawatt within roughly three years^[3][4]. Data center footprints are expanding in parallel, from tens or hundreds of megawatts toward some gigawatt-scale campuses^[3]. Partners speaking on the a16z Show described rack power requirements jumping roughly 70-fold, forcing a shift from AC to DC power and from air to liquid cooling - engineering problems as much as capital ones - and pointed to hyperscaler capital expenditure approaching $1 trillion next year, with GPU supply effectively pre-sold before it is even built.

a16z's core argument for urgency is a growth-rate mismatch: the hardware industry has historically grown 20-30 percent a year, but AI compute demand requires triple-digit growth, a gap the firm says justifies concentrated new capital rather than incremental venture bets^[1]. This is not a first move. In January 2026, a16z had already announced more than $15 billion across new funds, including a $1.7 billion Infrastructure Fund 2 and a $1.18 billion American Dynamism Fund 2, extending a hard-tech thesis the firm first staked out through its original American Dynamism fund backing manufacturing, robotics, space, and defense startups^[5][6]. The Machine Age Fund, closed on August 28, 2026, is the latest and most hardware-specific chapter of that buildout^[1].

Not everyone reads the AI infrastructure race the same way. Coverage of the launch noted questions about whether the stated urgency is overstated, echoing broader AI-bubble concerns that had already trailed a16z's earlier infrastructure bets^[1]. Martin Casado, the general partner leading the firm's infrastructure investing, directly rejected that framing, saying "none of us have ever seen it this dramatic" and arguing that underlying demand for compute is genuine even if some valuations eventually correct^[1]. Public reaction so far has leaned toward treating the fund as a credible read on real infrastructure bottlenecks rather than dismissing it as bubble-chasing, though the fund is barely a day old and the more skeptical case has yet to fully play out.