KKR Backs $10B AI Infrastructure Startup Helix, Taps Ex-AWS CEO Adam Selipsky to Lead It

The clearest signal that AI infrastructure has crossed from tech investment into utility-scale infrastructure investment came on April 30, when KKR announced the launch of Helix Digital Infrastructure — a new company backed by more than $10 billion in committed financing — and simultaneously revealed that Adam Selipsky, the former CEO of Amazon Web Services, would serve as its chief executive and chair.

It is a telling combination. Selipsky is not an infrastructure operator or a real estate developer. He is one of the most experienced cloud executives in the world, someone who watched the AI compute boom accelerate from the inside and understands what hyperscalers need from their infrastructure providers. That KKR chose him — rather than a traditional data center operator — to lead a business that will design, build, own, and operate AI physical infrastructure says something about how this asset class is evolving.

What Helix Will Actually Do

Helix Digital Infrastructure is not a REIT or a colocation provider in the conventional sense. Its scope is broader: the company will design and build the entire physical stack required for hyperscale AI deployment, including data centers, power generation assets, electricity transmission, cooling systems, and high-speed connectivity. The company then owns and operates these assets over long periods, generating revenue through long-term take-or-pay contracts with its hyperscaler customers.

This model is closer to what energy companies or telecommunications infrastructure operators do than to traditional data center developers like Equinix or Digital Realty. Helix is betting that hyperscalers want a single sophisticated counterparty who can handle the full complexity of power-constrained AI infrastructure development — not just the building, but the power procurement, the grid interconnection, the permitting, and the long-term operational accountability.

The Power Problem Behind the Business Model

The rationale for Helix is rooted in a constraint that has become the defining bottleneck of the AI expansion: power.

Building GPU clusters is now relatively straightforward. The real chokepoint is finding sites with adequate power, securing grid interconnection agreements (which can take three to five years in many US markets), procuring stable long-term power supply, and managing the thermal loads that come with dense AI compute. Every major hyperscaler has cited power availability as the primary constraint on how fast they can add AI compute capacity.

Private equity firms like KKR have an institutional advantage in navigating long-cycle infrastructure assets of this type. They can take on the multi-year development timelines and the capital intensity of power generation and transmission projects that don’t fit neatly inside a public company’s quarterly reporting structure. They can partner with utilities, negotiate with regulators, and develop power assets that generate returns over 20-year periods — time horizons that are poorly suited to the venture capital structures that fund most AI startups.

The $10 billion in committed financing at launch is substantial, but it is intended as a foundation rather than a ceiling. KKR’s ambition, according to sources familiar with the fund structure, is for Helix to eventually manage assets at a scale closer to $50-100 billion as it deploys capital and brings hyperscaler contracts online. KKR and Energy Capital Partners separately announced a $50 billion AI data center deal structure in recent months, suggesting the firm is approaching AI infrastructure investment as a multi-fund, multi-decade opportunity.

Selipsky’s Mandate — and Why He Said Yes

Adam Selipsky led AWS from June 2021 until early 2024, presiding over the cloud division as it crossed $100 billion in annual revenue and navigated the early AI compute surge. He was widely credited with steadying AWS after Andy Jassy’s elevation to Amazon CEO and with maintaining its market position against an increasingly aggressive Microsoft Azure during the period when OpenAI’s products began driving cloud demand in entirely new directions.

After leaving AWS, Selipsky joined KKR as a senior technology and AI strategy advisor, a role that appears to have been a transitional engagement before the Helix CEO appointment. His public explanation for joining is that Helix addresses the fundamental constraint on AI’s growth — not the models, not the software, not even the chips, but the physical infrastructure that has to exist before any of those other things can scale.

The argument has merit. For every GPU cluster that a hyperscaler wants to add, there has to be a building to put it in, a power line feeding it, a cooling system keeping it from melting, and a fiber connection sending data to and from it. The AI models attract the headlines, but the infrastructure layer is where the scaling constraint actually lives.

The Infrastructure Investment Playbook

KKR’s move is part of a broader private equity and infrastructure fund pivot toward AI-adjacent physical assets. Blackstone, in a widely noted posture earlier in 2026, announced a $150 billion target for AI infrastructure investment. Brookfield, Stonepeak, and a range of sovereign wealth funds have similarly flagged AI data centers and power assets as priority deployment targets.

What distinguishes Helix from these broader infrastructure plays is the operational model. Rather than simply providing capital to existing data center operators or acquiring colocation facilities, Helix is positioning itself as an active developer and operator — closer to a project developer in the energy sector than a passive infrastructure investor. This requires domain expertise alongside capital, which is where Selipsky’s hyperscaler background becomes operationally relevant rather than merely decorative.

The initial backer roster includes a sovereign wealth fund (name undisclosed) and two strategic partners, suggesting that Helix’s early capital base has both the patient return expectations typical of sovereign wealth and the operational connection that comes from strategic investors with skin in the hyperscaler game.

The Larger Shift This Signals

The AI investment narrative of 2023 and 2024 was dominated by model companies, GPU vendors, and application-layer startups. The narrative of 2025 and 2026 has shifted toward infrastructure, and Helix represents its most ambitious expression to date.

The underlying logic is compellingly simple. If AI compute demand continues growing at the rates hyperscalers are projecting — and the $725 billion in combined capital expenditure commitments announced by the major cloud providers through Q1 2026 suggests they believe it will — then the physical infrastructure required to host that compute will be one of the largest infrastructure investment opportunities of the next decade. It is not glamorous. It is not the kind of thing that gets covered at AI conferences. But it is, in a very literal sense, what makes the rest of the AI stack possible.

KKR, with Adam Selipsky building the operational team and $10 billion in committed capital at launch, is making a clear bet that the next phase of value creation in the AI economy will flow to whoever owns the physical substrate underneath it.