Analysis
Andreessen Horowitz has raised $1.1 billion for the Machine Age Fund, the firm's first vehicle built specifically around AI hardware and physical infrastructure rather than software or applications. The fund will invest in AI processors, memory chips, networking equipment, data storage, robotics platforms, data centers and home AI appliances -- what general partner Raghu Raghuram summarized as everything "within the four walls of the data center," TechCrunch reported.
General partners Martin Casado and Raghu Raghuram will run the strategy across both early- and growth-stage companies, with five named partners on the launch: Ben Horowitz, Casado, Raghuram, David Ulevitch and David George. Both lead GPs spent multiple decades in data-center system software, requiring deep hardware partnership -- a background the firm is positioning as differentiated underwriting for a category most software-focused VCs are not built to evaluate.
The thesis
Casado's framing is blunt: "Every time we have one of these technical epochs, it puts pressure on the infrastructure, but none of us have ever seen it this dramatic." The firm points to a 28-fold jump in compute density per rack between Nvidia's H100 and its newer Rubin generation as evidence that hardware supply chains -- which historically grow 20% to 30% annually -- structurally cannot keep pace with triple-digit AI demand growth. Hardware now accounts for more than 20% of a16z's overall deal flow, up from a much smaller share before the AI buildout accelerated.
The competitive and market context
This lands the same week Nvidia posted a $96.2 billion quarter and guided to $108 billion for Q3, the same week Marvell and Rubrik both beat estimates and got punished by investors anyway, and the same week a16z's own portfolio companies -- CoreWeave among them -- are racing to build the data-center capacity Nvidia's chips need somewhere to run.
A16z is not alone chasing this thesis: specialist infrastructure funds and strategics including Nvidia's own venture arm have been circling the same physical layer, but a $1.1 billion generalist-VC fund dedicated to it is a scale signal few peers have matched.
The counterweight
Hardware investing carries a capital-intensity and time-to-liquidity profile that looks nothing like a16z's software-era playbook -- fabs, data centers and robotics platforms take years and hundreds of millions of dollars to reach revenue, a very different risk curve than a SaaS seed round. And a fund this size launching now, at what may be the peak of AI infrastructure enthusiasm, is exactly the kind of late-cycle capital formation that looks prescient in a boom and reckless in a correction; a16z is betting the buildout has years left to run, not quarters.
For founders in chips, networking, robotics or data-center infrastructure, a dedicated $1.1 billion pool with two GPs who actually understand hardware supply chains is a meaningfully different diligence conversation than pitching a generalist partner who's never built a rack.
