Andreessen Horowitz Launches $1.1B Machine Age Fund to Accelerate AI Infrastructure Buildout

Andreessen Horowitz Commits $1.1 Billion to Physical Infrastructure for AI

Andreessen Horowitz has closed a $1.1 billion Machine Age Fund dedicated to accelerating the physical buildout of AI1

1

2

2

. The fund marks a strategic shift for the venture capital firm, traditionally known for software investments, now betting that hardware has become the binding constraint on AI advancement. Five senior partners—Ben Horowitz, Martin Casado, Raghu Raghuram, David Ulevitch, and David George—put their names to the launch, signaling how central the firm considers this thesis2

2

.

The Machine Age Fund will invest in AI infrastructure spanning computer chips, memory, networking, storage, data centers, robotics, and home AI appliances3

3

. This represents new capital rather than a carve-out from the firm's $15 billion venture fund raised earlier this year5

5

. According to Raghu Raghuram, who previously served as CEO of VMware, the decision to raise a separate fund reflects the unique needs of physical infrastructure investments, including different underwriting requirements and larger initial capital commitments5

5

.

Hardware Emerges as Critical Bottleneck in AI Scaling

The fund addresses an urgent reality: every layer of the AI stack is hitting supply chain bottlenecks and the limits of physics and computer science3

3

. Compute density per rack increased 28-fold from an H100 rack to a Rubin rack, while rack power moved from roughly 5-10 kilowatts to 100-250 kilowatts and is expected to reach 1 megawatt within three years2

2

3

3

. Individual data centers are scaling from tens of megawatts to hundreds, with some campuses approaching gigawatt scale2

2

.

Raghuram emphasized the severity of AI computing capacity constraints: "If you walk around San Francisco, talk to any company: They'll tell you that the only thing holding them back is a lack of AI computing capacity"5

5

. This demand pressure was echoed by Nvidia CEO Jensen Huang, who stated during an earnings call that demand for the company's chips is growing beyond 70% in fiscal year 2028 but remains constrained by supply5

5

.

Triple-Digit Growth Demands Reshape Hardware Industry

The physical layer of AI faces unprecedented scaling challenges. Both the volume of AI work and the compute each unit consumes are rising by orders of magnitude2

2

. The hardware industry supply side typically grows 20-30% annually, far short of the triple-digit growth needed to catch up with demand3

3

4

4

. Andreessen Horowitz argues this will change quickly as the bottleneck has moved from models and software into the physical world2

2

.

Specific areas requiring innovation include faster and more efficient systems, cheaper and higher-bandwidth memory across the memory hierarchy, faster and more scalable interconnects between nodes and systems, and power-efficient edge devices for AI to explore and interact with the world1

1

3

3

. The buildout also requires substantial cooling, materials, electrical, and real estate infrastructure to support these systems3

3

.

Hardware Deal Flow Surges to 20% of Pipeline

Andreessen Horowitz's deal flow reflects this shift. Hardware startups grew from a marginal share to over 20% of deals the firm now sees2

2

3

3

. This represents a fifth of the firm's pipeline—the point at which traditional objections about hardware taking longer, costing more, and scaling worse than software become less decisive2

2

.

The firm has already backed multiple AI infrastructure startups including Unconventional AI, Nexthop, Volta, Atoms, and Mind Robotics2

2

3

3

. Earlier investments include Skydio (series A in 2016), SpaceX, Anduril (first check in 2019), and Waymo (2020 raise)3

3

. In 2025, the firm backed Heron Power Inc., which develops solid-state transformers for data centers using silicon carbide chips instead of traditional metal coils submerged in insulating liquid4

4

.

Venture Capital Expertise Aligned with Hardware Demands

The team behind the Machine Age Fund brings deep hardware and data center expertise. Guido Appenzeller previously served as CTO for Intel's Data Center Group, while Raghu Raghuram and Martin Casado spent decades in the data center space with system software requiring deep hardware partnerships3

3

. Shangda Xu and David George have led investments across the AI infrastructure stack from silicon and networking to large-scale systems and compute platforms3

3

.

For chip companies specifically, Raghuram noted that investments could be even larger than model makers at inception: "These companies have to turn the design into a prototype of a chip or networking system. And that takes quite a bit of money"5

5

. Despite larger capital needs, he pointed to timely exits like AI startup Groq's reported $20 billion licensing deal with Nvidia and Cerebras Systems' public debut as evidence of viable returns5

5

.

Broader Venture Landscape Shifts Toward AI Infrastructure

Andreessen Horowitz joins other major venture capital firms ramping up AI investment. Kleiner Perkins raised $3.5 billion across two new funds in March, with one vehicle focusing on early-stage startups4

4

. OpenAI backer Thrive Capital secured $10 billion a month earlier4

4

. The Machine Age Fund will primarily invest in early-stage startups, with some capital reserved for more mature companies5

5

.

Watch for how memory and interconnect improvements emerge as critical focus areas. A rack full of accelerators that cannot be fed data fast enough becomes an expensive heat generator, and the industry has spent two years discovering this represents the actual limit more often than anticipated2

2

. Real estate, power distribution, and cooling infrastructure—historically not venture categories—may see increased attention as 63% of new capacity now goes somewhere other than the five established hubs2

2

.