Helix Digital Infrastructure: KKR’s $10B Bet on Owning the Full AI Infrastructure Stack

What Helix Is and Why the CEO Choice Tells You Everything

Most private equity data center plays are straightforward: buy an existing data center operator, add leverage, sell at a higher multiple in 5-7 years. Helix Digital Infrastructure is a different kind of bet. According to reporting by Bisnow, KKR is building Helix to “partner with large-scale cloud providers to design, own and run its own data centers, power generation and transmission.” That is not a real estate transaction — it is a full-stack infrastructure company targeting the complete physical layer of AI computing.

The CEO appointment underscores the ambition. Adam Selipsky ran AWS from May 2021 until June 2024 — overseeing a period when AWS revenue grew from roughly $54 billion annually to over $90 billion. He was not a salesperson who knew cloud customers; he was the operator who knew how hyperscaler infrastructure is built, scaled, and contracted. His presence as Helix CEO signals that KKR is not building a passive landlord — it is building an active infrastructure operator that understands the technical requirements of hyperscaler customers from the inside.

The Chief Investment Officer role goes to Waldemar Szlezak, KKR’s global head of digital infrastructure — ensuring that the investment discipline of one of the world’s largest alternative asset managers is embedded in the venture’s capital allocation from day one.

The Investment Thesis: Why Now, Why Physical Infrastructure

The Helix bet is grounded in a clear market signal. Private equity invested over $108 billion in data center deals in 2024, with projections showing the market reaching $635 billion by 2028. McKinsey estimates $7 trillion will be needed to meet total data center demand by 2030. These are not AI optimism numbers — they are derived from announced hyperscaler capex commitments and current capacity backlogs.

The three hyperscalers have confirmed the demand signal with their own balance sheets. In Q1 2026 alone, Amazon committed to $200 billion in infrastructure capex for the full year, Microsoft to $190 billion for fiscal 2026, and Alphabet to approximately $185 billion. Google Cloud’s contract backlog nearly doubled to over $460 billion during Q1 2026, driven by enterprise AI solutions demand. These commitments represent actual purchase orders that must be fulfilled — and the physical infrastructure to fulfill them does not yet exist at the required scale.

Helix’s positioning as a partner to hyperscalers rather than a competitor to them is the key architectural choice. The venture will design and own data centers specifically to serve hyperscaler demand — taking on the capital-intensive, long-lead-time construction and power procurement work that hyperscalers increasingly prefer to outsource to specialized operators. This mirrors the model already working for CyrusOne, in which KKR co-owns a significant stake alongside Global Infrastructure Partners.

What This Means for Data Center Buyers, Hyperscaler Partners, and Infrastructure Investors

1. Evaluate Helix as a potential long-term infrastructure partner before the initial capacity is committed

Hyperscalers commit data center capacity years in advance. If Helix builds the operator relationships and technical credibility that its leadership team’s background suggests it can, initial capacity allocations will be locked up by 2026-2027 before the first facility breaks ground. Enterprise buyers that need co-location within hyperscaler-adjacent facilities — not the hyperscalers themselves, but their enterprise customers seeking proximity — should monitor Helix’s development closely. Early conversations with an infrastructure operator before capacity is committed are worth more than late conversations after waiting lists form.

2. Use the Helix model as a benchmark for evaluating other private infrastructure investments

Helix’s full-stack approach — data center + power generation + transmission — is the emerging template for serious AI infrastructure plays. Operators that own only the building but not the power supply face critical dependency on utility grid availability, which has become the binding constraint in AI data center development in the US, Europe, and increasingly Asia. When evaluating any private infrastructure investment with an AI angle, ask whether the investment includes energy generation control or merely grid dependency. Helix’s integrated model is the higher-value version of the same bet.

3. Model the competitive implications for cloud pricing over a 3-5 year horizon

More privately capitalized AI data center capacity competing to serve hyperscalers changes the supply dynamics for cloud infrastructure. When hyperscalers have multiple high-quality operator partners competing for their business — rather than building exclusively themselves — they gain pricing leverage on construction costs. That cost pressure eventually flows through to cloud service pricing. Enterprise cloud buyers on long-term contracts should include renewal terms that allow renegotiation as infrastructure supply conditions change.

The Structural Question: Can Private Capital Deliver at Hyperscaler Speed?

KKR’s Helix faces the same challenge that all private infrastructure plays targeting hyperscalers confront: speed. AWS CEO Andy Jassy said in Q1 2026 that “AI revenue is growing triple digits year over year … We’re monetizing capacity as fast as we can install it.” Hyperscalers are constrained not by demand or money, but by construction timelines — permitting, power interconnection queues, equipment lead times, and skilled labor availability.

Private capital does not solve these bottlenecks through money alone. It solves them through operational expertise, pre-positioned land and power assets, and relationships with regulators and utilities that allow faster site development. This is why the Selipsky appointment matters operationally: his experience running AWS’s infrastructure delivery machine means Helix starts with the playbook, not just the capital.

The data center market also has a geographic concentration problem. The majority of current AI data center development is concentrated in a handful of US markets — Northern Virginia, Phoenix, Dallas, Chicago — where power and permitting are becoming constraints. According to Data Center Knowledge, the pipeline of new developments in May 2026 is spanning an increasingly broad set of secondary markets as primary markets hit power availability limits. Helix’s ability to develop alternative geographic markets — particularly in regions with available power capacity and faster permitting — will be a key differentiator if it can execute. KKR’s existing CyrusOne relationship, including a negotiated US Army data center project in Utah, suggests the firm already has relationships in non-traditional markets.

The energy angle deserves specific emphasis. AI training clusters require power at a density and reliability that the commercial grid struggles to provide without substantial infrastructure investment. Nuclear energy — including small modular reactors — has emerged as a serious power source for AI data centers. Microsoft has signed agreements for nuclear power capacity, and Amazon Web Services has explored similar arrangements. Helix’s mandate to own power generation alongside compute capacity positions it to pursue the same integrated energy strategy. Whoever controls reliable, high-density power supply near major internet exchange points in the next decade holds the most defensible position in the physical AI infrastructure stack.

Frequently Asked Questions

Sources & Further Reading

• KKR Launching $10B AI Firm for Data Centers — Bisnow
• Helix Digital Infrastructure: KKR Plans $10B AI Data Center Firm — Data Center Dynamics
• Q1 2026 Cloud Revenue: Google Cloud Wins — Motley Fool
• KKR Helix Digital Infrastructure Analysis — The AI Consulting Network
• New Data Center Developments May 2026 — Data Center Knowledge