Andreessen Horowitz Raises $1.1 Billion for AI Hardware Investments
Andreessen Horowitz has launched a "Machine Age" fund with $1.1 billion raised, aimed at accelerating the physical buildout of AI infrastructure. The fund will focus on hardware, a notable departure for a firm whose track record has been built on software investments. TechCrunch
The fund's investment mandate covers the tangible infrastructure underpinning AI: semiconductors, memory components, data centers, and robotics. a16z described AI as the "strongest tool ever developed for solving problems and bestowing abundance" and characterized its advancement as a "social and national imperative." TechCrunch
This is not a16z's first capital commitment to AI infrastructure. In 2024, the firm established a dedicated $1.25 billion fund for AI infrastructure bets. Bloomberg In January 2026, a16z raised more than $15 billion across its funds, its largest-ever funding haul, with plans to deploy capital across multiple industries. Bloomberg The Machine Age fund, announced seven months later, adds a hardware-specific vehicle to that broader capital pool.
The pivot toward physical assets reflects where the bottlenecks in AI actually sit. Training and inference, the two core computational workloads of AI, are constrained by GPU supply, memory bandwidth, power delivery to data centers, and cooling capacity. Software can orchestrate compute, but it cannot manufacture it. A fund targeting chips, memory, data centers, and robots is, in effect, placing bets on the supply chain that determines how fast AI capability can scale.
The language a16z uses, framing AI advancement as a "social and national imperative," aligns the firm's investment thesis with a geopolitical argument. That framing positions domestic AI infrastructure buildout as a matter of competitive necessity, which, if it resonates with policymakers, could shape regulatory tailwinds for the portfolio companies this fund backs.
The decision to stand up a separate, hardware-focused fund also carries a structural logic. Hardware investments differ from software deals in capital intensity, timeline to exit, and technical diligence requirements. Semiconductors and data centers demand deep supply-chain expertise and patient capital cycles measured in years, not quarters. A dedicated fund with dedicated leadership can build that competence without diluting the firm's software-focused teams.
In my view, the more interesting signal is what the fund implies about a16z's assessment of where value accrues in the AI stack. The application layer is crowded; foundation models are capital-intensive but increasingly commoditized; the physical layer, from advanced packaging to data-center power systems to humanoid robotics, remains underbuilt relative to projected demand. A $1.1 billion vehicle is modest by the standards of the capital flowing into AI model training, but it is directed at a layer where fewer competitors have the patience or the operational expertise to deploy effectively.
The robotics inclusion is worth noting. Industrial and humanoid robotics have attracted renewed attention as AI models capable of real-world perception and control mature. Placing robotics alongside chips and data centers in a single fund suggests a16z sees these as parts of a single buildout problem: compute, the facilities that house it, and the machines that extend AI into the physical world.
The cumulative trajectory is clear from the numbers. A $1.25 billion AI infrastructure fund in 2024, a $15 billion raise in January 2026, and now a $1.1 billion hardware-specific fund in August 2026. a16z is systematically expanding its footprint in AI's physical layer. Whether that translates into returns will depend on execution, but the capital allocation strategy reveals a firm betting that the next phase of AI is as much about concrete, silicon, and steel as it is about code.
