AI data center boom ‘stress tests’ insurers as private capital soars

AI data centers are becoming a “stress test” for insurers as rapid technological advances and the use of increasingly complex financial structures present unique challenges and opportunities for the industry.

Global spending on data centers could reach: 7 trillion dollars By 2030, most of this spending will no longer come solely from hyperscalers, according to McKinsey. Instead, Big Tech is increasingly leveraging private equity, private credit, and using debt to finance the capital-intensive construction of facilities.

Private infrastructure data center deals consistently totaled over $10 billion last year, according to data from Preqin. The biggest deal reached $40 billion, with Nvidia, Microsoft, BlackRock and Elon Musk’s xAI forming part of an investor consortium to buy Aligned Data Centers.

The fact that so much money is tied up in building, building and operating data centers has been “a real stress test” for major insurance companies over the last four to five years, said Tom Harper, data center leader at insurance brokerage. Gallagher’she told CNBC.

“When you put over $10 to $20 billion in a single location, it creates capacity issues in the market. There was always an appetite in the market for those risks because they were very high-end builds. They have state-of-the-art, AA-plus construction locations, but the capacity — the ability to provide insurance capacity in those locations — has been a challenge.”

According to Harper, it would be nearly impossible to reasonably insure a $20 billion campus in 2023. But in 2026, this has become a weekly conversation.

Estimated spend on AI data centers redirected It is considered the largest peacetime investment project in history. Rajat Rana, partner at Quinn Emanuel Urquhart & Sullivan, told CNBC he would take it a step further, emphasizing that it was “the largest peacetime investment project in human history, financed largely off-balance sheet.”

Rana, who worked on structured finance cases in the wake of the housing crisis triggered by the 2008 Financial Crash, said following developments in AI data center financing felt like “deja vu.”

“We’re talking trillions of dollars and we’re almost back to the same cycle where there’s almost no transparency in financing structures; the scale is astronomical,” he said.

The AI ​​boom is not only increasing demand for facilities, but also spurring rapid advances in energy generation and chips, the critical technology that data centers house. The developments and the large sums of money flowing into the industry represent both risk and reward for insurers and lenders.

The merger and acquisition boom is also keeping transaction lawyers busy; Kirkland & Ellis, some companies Creating data center-specific teamsIt appoints experts in the fields of real estate, energy, telecom, finance, insurance, trade, private equity and cyber security.

Professional service firm Swamp one started private digital infrastructure advisory group It is designed to assist clients as contracts become increasingly complex.

Last year Marsh also launched Nimbus, a 1 billion euro ($1.2 billion) insurance facility covering the construction of data centers in the UK and Europe. Seven months later this extended The facility will offer limits of up to $2.7 billion.

“Private credit can meaningfully complement banks and support non-hyperscale contract sales,” said Alex Wolfson, senior vice president of credit specialties at Marsh Risk.

Wolfson explained that as data center loans increase, insurers that protect lenders in case the borrower defaults are starting to reach limits. Marsh is working on solutions to support lenders.

But Quinn Emanuel’s Rana cautioned that when it comes to data centres, it’s not easy for insurers to fully understand the risk because financing moves off the balance sheet.

In January, four US senators in the name The government needs to investigate how Big Tech is increasingly turning to “sophisticated and opaque debt markets to borrow staggering amounts of cash.” Large debt loads could lead to “destabilizing losses” for financial institutions and trigger a broader financial crisis that would harm the economy, senators warned in an open letter.

Increased transparency in financing could lead to collateral litigation risks for downstream investors such as pension funds, insurers and asset managers investing in private loan funds who later learn they were not fully aware of the concentration risk, Rana he said in a note It was published in March.

He told CNBC that some private equity funds have reached out to him with concerns about commercial leases and the valuation of properties.

Tenants are trying to negotiate extensions to their properties and landlords are squabbling over value as they seek higher prices for AI data centres.

“I’m not a doomsday man saying, ‘Hey, it’s going to collapse. Whether it collapses or not, disputes are inevitable and we’ve already seen those disputes,'” Rana said.

A significant debate surrounds potential cracks in funding centers on GPUs and the risk that their lifecycles may not be compatible with the longer lifespan of the facilities that house them.

CoreWeaveThe company, which sells AI technology in the cloud, became the first to secure GPU-backed loans, essentially using the value of high-performance chips as collateral. company last week announced raised $8.5 billion in its first investment-grade GPU-powered deal. Its shares rose 12% on the day.

While data centers typically have a lifecycle of decades, the average lifecycle of a GPU is around seven years.

“There are different data centers that increase debt by explaining different life cycles to investors,” Rana said. He called the problem the “GPU debt wheel.” invented By AI commentator Dave Friedman.

“This is almost like a treadmill that AI data centers are running on,” Rana told CNBC. Even if the financing structure is ring-fenced and backed by an investable counterparty, the real risk may lie in whether today’s equity issuance will turn into a credit problem later over time.

“As these new chips come in, data centers will feel pressure to take on more debt, and then they’ll have to build new infrastructure, which will essentially create a billion-dollar question: How fast can you build these facilities? How fast can you raise the loan?”

Harper says the cost of financing these projects will likely continue to drive the recent growth in asset-backed securitization deals as more commercial mortgage-backed securities are sold to investors.

For some insurers, like Gallagher, changing dynamics in the industry are opportunities rather than challenges. Harper said GPUs’ lifecycles are increasing. In an environment where things were losing value rapidly, Gallagher had to get creative and write special insurance policies with a predetermined agreement on how the assets would be valued.

“The size and scope of these would be a nightmare.” [facilities] to determine [the value of] “Every single unit,” he said.

Harper also emphasized that GPUs are interchangeable. The firm has seen operators anticipate relatively short life cycles and build more modular facilities in response.

“There is a fundamental tension in data center project financing: lenders often want asset lives that exceed loan terms by a comfortable margin, and the shorter useful life of GPUs challenges this assumption,” said Marsh Risk’s Wolfson.

Therefore, lenders are structuring loans more carefully to protect themselves.