By ByteBot
Private equity giant KKR committed over $10 billion last week to launch Helix Digital Infrastructure, a company that will build and operate end-to-end AI infrastructure for hyperscalers. The announcement on April 30, 2026 comes with a twist: former AWS CEO Adam Selipsky is leading the charge. When the executive who built Amazon Web Services into a $100 billion business leaves cloud computing to solve the physical infrastructure problem, that’s a signal worth heeding.
The move reflects a brutal reality: hyperscalers are burning through $725 billion in AI capex this year, yet 30-50% of planned data center capacity won’t come online until 2028. Power shortages, copper supply chains, and helium scarcity are creating bottlenecks that money alone can’t solve. Private equity is betting it can build faster than Big Tech.
The Infrastructure Crisis Hyperscalers Can’t Buy Their Way Out Of
Microsoft, Amazon, Google, and Meta collectively plan to spend $725 billion on AI infrastructure in 2026—a 77% jump from last year’s $410 billion. Microsoft alone increased its forecast to $190 billion. Amazon is at $200 billion. Google added another $5 billion to hit $190 billion. Meta raised its projection by $10 billion to $145 billion.
The money is there. However, the infrastructure isn’t. The U.S. grid interconnection queue exceeds 2,100 gigawatts—more than total grid capacity—creating 2-3 year wait times for power connections. Copper prices hit a record $6 per pound in January 2026, currently sitting at $5.61. Moreover, each megawatt of data center capacity requires 27 tons of copper. Do the math on hundreds of megawatts, and the supply chain problem becomes clear.
Then there’s helium. Following strikes on Qatari production—which accounts for one-third of global supply—prices doubled. Semiconductor fabs in Taiwan and South Korea are now rationing helium, essential for wafer cooling and leak detection. Industry analysis projects 30-50% of 2026 data center capacity will slip to 2028. Hyperscalers have capital but can’t buy their way past physical constraints.
Related: AI Chip Shortage Crashes Consumer Hardware: 70% to Data Centers
Why Adam Selipsky’s Pivot Matters
Selipsky spent 11 years building AWS from 2005 to 2016, leading marketing, sales, and support as the platform grew from startup to cloud leader. He left for Tableau, led it through Salesforce’s acquisition, then returned to AWS as CEO in 2021. Over three years, he accelerated AWS to $100 billion in annualized revenue and launched transformative AI services like Amazon Bedrock and Amazon Q.
His September 2025 move to KKR as Senior Technology and AI Strategy Advisor surprised few. Meanwhile, his April 2026 appointment as CEO of Helix Digital Infrastructure is the reveal. When someone who knows exactly what hyperscalers need—because he ran the world’s largest cloud—leaves to solve infrastructure, it validates two things: the problem is real, and it’s solvable.
Selipsky’s decision signals infrastructure specialization is the play. Consequently, hyperscalers focus on AI services and software. Specialists like Helix handle power generation, grid connections, data center construction, fiber networking, and cooling systems. Specialization worked for cloud computing. Private equity is betting it works for AI infrastructure.
What Helix Actually Does
Helix provides full-stack infrastructure as a single counterparty. In other words, it delivers power generation and transmission, land development and permitting, data center construction, fiber networking, and cooling systems. Hyperscalers lease the infrastructure rather than building it themselves, converting capital expenditure into operational expense.
The value proposition is speed and risk transfer. Traditional hyperscaler builds take 36-48 months from land acquisition to operational status. Furthermore, Helix targets 12-24 months by pre-developing sites with power secured, managing permitting relationships, and controlling supply chains. Hyperscalers pay for operational capacity, not construction risk.
Funding comes from KKR’s $10 billion+ commitment, an unnamed sovereign wealth fund, and two strategic partners. Leadership includes Selipsky as CEO and Waldemar Szlezak—KKR’s global head of digital infrastructure—as CIO. The company is purpose-built for hyperscale, single-tenant AI facilities under long-term contracts.
Private Equity Floods Into AI Infrastructure
KKR has already deployed $40 billion into digital infrastructure and operates six global data center platforms. Helix is the next evolution. Blackstone boasts an $85 billion data center platform, the largest globally. Brookfield has major infrastructure funds targeting AI. Additionally, Morgan Stanley and BlackRock describe 2026 as the beginning of a “golden age” for private infrastructure, driven by AI demand and energy transition needs.
The math is compelling for private equity. McKinsey forecasts data center demand could reach $7 trillion by 2030, growing at 22% annually. Goldman Sachs projects $765 billion in annual AI capex in 2026, rising to $1.6 trillion by 2031. Hyperscalers are publicly committing hundreds of billions, creating demand visibility that infrastructure investors crave.
The question is whether private equity ownership accelerates deployment or creates new extraction points. Hyperscalers built cloud computing by convincing companies to outsource on-premises infrastructure. Now they’re outsourcing their own infrastructure to private equity. The irony isn’t lost on anyone paying attention.
Related: Cloud Waste Hits $182B: Why 35% of Cloud Spend Is Burned
Execution Is TBD
Helix launched one week ago. The $10 billion commitment is real. Nevertheless, the execution isn’t proven. The company needs to announce first projects, secure hyperscaler contracts, navigate permitting and environmental reviews, and deliver on the 12-24 month timeline promise. Hyperscalers have struggled with 36-month timelines despite massive resources. Can a new company really move twice as fast?
Announcements are easy. Building data centers with power secured, copper delivered, and helium available is hard. If Helix delivers, it validates the infrastructure specialization model and opens the door for more private equity capital. If it stumbles, it becomes another example of financial engineering meeting physical reality.
Key Takeaways
- Hyperscalers are spending $725 billion on AI infrastructure in 2026, but 30-50% of capacity is delayed due to power, copper, and helium shortages
- Former AWS CEO Adam Selipsky’s pivot to Helix validates that infrastructure—not software—is the critical bottleneck for AI deployment
- Helix provides full-stack infrastructure (power, data centers, networking) as a single counterparty, targeting 12-24 month deployment vs traditional 36-48 months
- Private equity is flooding into AI infrastructure with tens of billions, betting on specialization and long-term hyperscaler demand
- Execution remains unproven—Helix must deliver on speed and cost promises to validate the model
