DATA COMPUTE

Josie Doan/PitchBook News

Chips & Datacenters

VC may dominate AI computing deals, but PE supplies the capital

While VCs are funding the experimentation, PE firms are financing the infrastructure.
DATA COMPUTE

Josie Doan/PitchBook News

VC investment in advanced computing accounted for 86% of deal count in Q4, yet only 16% of deal value, well below its historical share of 29%, according to PitchBook’s latest Advanced Computing PE and VC Trends report.

The rest came from PE investors.

The trend reflects the growing scale of the AI infrastructure buildout, highlighting the need for massive investments not just in software and large language models, but also in data centers, chip manufacturing and networking equipment. These assets require billions to operate, making them more attractive to PE firms that have the capital and appropriate return expectations for the endeavor.

VC investments in advanced computing, which includes AI chips, data centers and quantum computing, reached $53.5 billion in Q4, a five-year high.

There were 293 deals, a little less than the previous quarter’s 304. Of those, 251 were VC deals, with PE making up the remaining 42 and accounting for $44.9 billion in deal value.

Data centers, in particular, are driving the surge in funding, with PE investors putting up large checks in the segment. Investment in data centers grew to $48.1 billion in Q4, also a five-year high and a significant uptick from last quarter’s $13.2 billion.

There was a string of mega-deals. In October, a group of investors, including BlackRock and Abu Dhabi’s sovereign wealth fund MGX, announced a $40 billion acquisition of Aligned Data Centers, an operator of data centers designed to support AI workloads. In November, Blackstone invested $1.2 billion in Wolf Summit Energy, a power plant under construction to meet the demand from data centers.

At the same time, VCs continue to fund companies building specialized computing services on a smaller scale. Lambda, for instance, raised $1.5 billion in November and operates a few facilities as an AI GPU cloud services provider.

These trends are shaping the market into a two-tiered landscape: one where VC fuels experimentation and services, and the other where PE provides the massive amounts of capital needed to build physical infrastructure.

It’s also the surest sign yet that AI is becoming an infrastructure asset class of its own that private market investors all want a piece of.

Learn about our editorial standards.