Andreessen Horowitz just closed a $1.1 billion fund dedicated entirely to AI hardware infrastructure. They’re calling it the Machine Age Fund, and it’s the first time the firm has raised a vehicle specifically focused on the physical layer of AI — semiconductors, memory chips, networking equipment, storage systems, data centers, robotics, and consumer AI devices.

The announcement came on August 28, 2026. The timing is not accidental.

Why Hardware, Why Now

a16z has been a software-first firm for most of its existence. The fact that they’re now betting $1.1 billion on physical infrastructure tells you something important about where the AI buildout actually stands.

The firm’s position is blunt: AI infrastructure is hitting a wall. Every part of the hardware supply chain is capacity constrained — chips, memory, power, and cooling. General partner Martin Casado put it plainly: “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 numbers back this up. Compute density per rack has jumped 28-fold from an H100 cluster to a Rubin rack. The hardware industry’s usual growth rate of 20 to 30 percent per year cannot keep up with demand growing at triple-digit rates. The gap between what AI developers need and what the supply chain can deliver is widening, not narrowing.

Hardware has quietly grown from a sliver of a16z’s deal flow to over 20 percent of what they’re seeing. This fund formalises that shift and puts a dedicated pool of capital behind it.

What the Fund Will Back

The Machine Age Fund’s investment scope covers the full hardware stack that AI runs on:

• Semiconductor design and fabrication — custom AI chips, inference chips, and the tools to design them faster
• Memory systems — the bottleneck nobody outside the industry talks about, but which determines how much data an AI model can process in real time
• Networking and interconnects — moving data between chips fast enough to keep GPUs fed
• Storage — the foundation for enterprise data that AI models need to be useful
• Data centers — the physical buildings and power infrastructure the whole stack runs inside
• Robotics — physical systems that act on AI decisions in the real world
• Consumer AI devices — hardware built to run AI workloads at the edge, not just in the cloud

This is not a bet on one piece of the puzzle. It is a bet that every layer of the physical AI stack needs capital that traditional tech venture firms have not been equipped to provide.

Why This Matters for Businesses Deploying AI Now

For business leaders evaluating AI investments, the Machine Age Fund is a useful signal to read correctly.

The short-term reality is that compute is expensive and will stay that way. Every major infrastructure provider — Amazon, Google, Microsoft, Nvidia — is announcing multi-billion dollar buildouts, and demand is still outpacing capacity. That means AI compute costs will not fall sharply in the next 12 to 18 months, even as model capabilities continue to improve.

Businesses that are waiting for costs to drop before making their first AI deployment may find the wait extends further than they expected, while competitors that moved earlier have already built operational advantages that are hard to close.

The medium-term signal is more encouraging. Capital at this scale flowing into hardware — a16z’s $1.1 billion alongside AWS and Nvidia’s infrastructure expansion plans, Google’s $84 billion raise, and dozens of chip startups — means the supply constraint is being attacked from multiple directions simultaneously. Costs will come down. Capacity will expand. The question is when, not whether.

The Cloud Advantage for Business AI

There is a practical implication here for businesses deciding how to access AI capabilities.

Companies building or evaluating on-premise AI infrastructure face the direct brunt of this constraint. Hardware lead times for high-end GPU clusters stretch to six months or longer in some configurations. Power and cooling infrastructure in owned facilities adds further complexity.

Cloud-based AI agent deployments are insulated from most of these constraints. The infrastructure problem belongs to the cloud provider, not the business. When a business runs AI agents through a managed cloud service, they get the benefits of whatever compute is available without needing to manage the physical stack or absorb supply chain delays.

This is part of why AI agent deployments through services like Omni Ops can deliver results faster than organisations trying to build their own infrastructure. The complexity of the hardware layer is handled by people whose entire business is managing it at scale.

What This Means for Business

If you are still evaluating AI. The a16z fund is evidence that the hardware constraint is real and being taken seriously at the highest levels of capital allocation. Infrastructure costs will stay elevated in the near term. This makes cloud-based AI agent services more cost-effective relative to self-hosted alternatives, not less, for the next several years.

If you are already deploying AI. The supply chain tightening validates working closely with your current cloud and AI providers on capacity planning. If your workflows are going to scale significantly, having that conversation now — before capacity gets tighter — is worth the time.

If you are curious about what comes next. The fund’s focus on robotics and consumer AI devices points toward where a16z thinks the next wave lands: physical AI that operates in the world, not just in software. Businesses that have already built the workflow and data foundations for AI agents will be better positioned to add physical AI capabilities when they mature.

The Machine Age Fund is $1.1 billion of institutional conviction that AI infrastructure is a solvable problem. The businesses that treat it as a reason to delay are reading the signal backwards.

---

Enterprise DNA helps business owners and operations teams put AI agents to work in their specific context. If you want to understand what an AI agent workforce could look like in your business — without the hardware headache — book a discovery call with our team.