Andreessen Horowitz just closed a $1.1 billion fund dedicated entirely to the hardware that makes artificial intelligence actually work. The firm is calling it the “Machine Age” fund, and the name tells you everything about where a16z thinks the real constraint on AI progress lives: not in the code, but in the physical world.
The fund, which closed on August 28, targets the full AI hardware stack. That means chips, memory, data centers, networking equipment, and robotics, essentially every tangible thing that sits between a large language model and the electricity coming out of the wall.
Why hardware, why now
The calculus has shifted dramatically as AI workloads have grown to a scale that existing infrastructure simply cannot handle. Rack power demand in data centers is projected to surge from 5-10 kW to somewhere between 100-250 kW, with some estimates reaching up to 1 MW per rack.
The team and the track record
The fund is being led by partners with deep roots in hardware and data center management, including Raghu Raghuram, Martin Casado, and David Ulevitch. Raghuram’s name will be familiar to anyone who followed VMware, where he served as CEO. Casado co-founded Nicira, the networking startup that VMware acquired.
The firm already has a portfolio of hardware bets to point to. Skydio, the autonomous drone company, stands out as one of a16z’s earlier hardware successes. More recent commitments include investments in Volta and Mind Robotics, both of which sit squarely in the physical-infrastructure-for-AI category the Machine Age fund is designed to accelerate.
This isn’t a16z’s only move in the AI infrastructure space this year, either. The firm reportedly allocated a separate $1.7 billion exclusively for AI infrastructure investments across its broader fundraising efforts in 2026, making the Machine Age fund one piece of a much larger capital deployment strategy.
The supply chain problem nobody wants to talk about
Building AI infrastructure at the scale the industry is projecting requires sourcing advanced semiconductors, specialized cooling systems, and high-bandwidth networking components, all of which face their own bottlenecks. Chip fabrication capacity remains concentrated in a handful of facilities worldwide. Advanced packaging technology, which is critical for the kind of multi-die architectures that next-generation AI accelerators rely on, has even fewer providers.
The fund also aligns with a16z’s broader “American Dynamism” thesis, which focuses on backing companies that strengthen US industrial and technological capabilities.