a16z Machine Age Fund Commits $1.1 Billion to AI Hardware

Andreessen Horowitz, also known as a16z, has launched the a16z Machine Age Fund, a $1.1 billion investment vehicle focused exclusively on hardware and the physical infrastructure behind artificial intelligence. The fund covers the physical layer of AI, including chips, memory, networking, storage, data centers, robotics, and home AI appliances. The firm said the investment strategy is driven by growing demand for computing infrastructure and constraints across supply chains, power, cooling, and other parts of the AI hardware stack.

Join thousands of readers who receive the latest software reviews, expert comparisons, and industry news delivered straight to their inbox.

a16z Machine Age Fund Targets AI Infrastructure

The a16z Machine Age Fund will invest across multiple layers of the computing infrastructure required to operate AI systems. Andreessen Horowitz said its mandate includes semiconductor technologies, memory, networking, storage, and complete systems such as data centers and robotics. The fund will also consider AI appliances designed for consumers. Five a16z partners are associated with the fund: Ben Horowitz, Martin Casado, Raghu Raghuram, David Ulevitch, and David George.

The investment strategy comes as computing requirements increase across AI workloads. According to a16z, compute density per rack has increased 28 times between Nvidia’s H100 generation and Rubin systems. Rack power requirements have also risen from roughly 5 to 10 kilowatts to between 100 and 250 kilowatts. The firm expects rack power requirements to reach 1 megawatt within three years, while some data center campuses are moving toward gigawatt-scale capacity.

Physical Layer of AI Becomes Investment Focus

The physical layer of AI extends beyond processors and servers. a16z identified memory bandwidth, interconnects, power-efficient edge devices, cooling, materials, electrical infrastructure, and real estate as areas requiring further development. The firm said networking within AI racks is also approaching the limits of copper cabling, while data center power requirements are expanding from tens of megawatts to hundreds of megawatts and, in some cases, gigawatt-scale campuses.

The a16z AI hardware fund builds on a growing share of hardware opportunities in the firm’s investment pipeline. a16z said hardware startups represented more than 20% of its deal flow in recent years, compared with a smaller share previously. The firm has backed companies including Unconventional AI, Nexthop, Volta, Atoms, Heron Power, Mind Robotics, Skydio, SpaceX, Anduril, and Waymo.

AI Hardware Demand Drives Infrastructure Investment

The Andreessen Horowitz Machine Age Fund reflects the firm’s view that AI infrastructure requires investment across the entire computing stack. a16z said the hardware industry has historically grown at around 20% to 30% annually, while current AI demand requires substantially faster expansion. The firm is therefore targeting technologies that can improve memory capacity and bandwidth, increase interconnect performance, reduce power consumption at the edge, and support the physical infrastructure needed by large-scale AI systems.

The fund also expands a16z’s existing hardware investment activity. The firm said it has invested in hardware companies since its early years, including leading Skydio’s Series A in 2016, investing in SpaceX, backing Anduril in 2019, and participating in Waymo’s 2020 financing. The Machine Age Fund formalizes hardware as a dedicated investment focus for the firm while targeting infrastructure across chips, networking, compute systems, power, cooling, and other physical components.

What the Machine Age Fund Means for AI Infrastructure

The a16z $1.1 billion fund places dedicated venture capital behind hardware categories that support AI deployment, from semiconductor and memory technologies to data center power and cooling infrastructure. For AI infrastructure companies, the fund signals continued investor attention toward technologies addressing computing capacity, energy requirements, networking, and physical deployment constraints.