AI data centers are becoming a “stress test” for insurers as rapid technological advancements and the use of increasingly complex financial structures present a unique set of challenges and opportunities for this sector.
Global spending on data centers could reach $7 trillion By 2030, according to McKinsey, and the majority of that spending may no longer come solely from hyperscalers. Instead, Big Tech is increasingly tapping private equity, private debt and using debt to finance capital-intensive construction of facilities.
Private infrastructure data center deals were consistently above $10 billion last year, according to Preqin data. The largest deal was for $40 billion, in which Nvidia, Microsoft, BlackRock and Elon Musk’s XAI were part of a consortium of investors to buy aligned data centers.
The fact that there is so much money involved in building, building and operating data centers has been a “real stress test” for major insurance companies over the past four to five years, said Tom Harper, data center leader at the insurance broker. GallagherTold CNBC.
“When you invest more than $10 to $20 billion in a single location, it creates capacity issues in the market. There has always been an appetite in the market for these risks because they are higher quality builds. They have cutting-edge technology, they are AA plus plus build locations, but the capacity – the ability to provide insurance capacity at these locations – has been difficult.”
According to Harper, it was almost impossible to properly insure the $20 billion complex in 2023. However, in 2026, this has become a weekly conversation.
We’re talking about trillions of dollars, and almost going back to the same cycle where there is almost no transparency about financing structures – the scale is huge.
Rajat Rana
Partner at Quinn Emanuel Urquhart & Sullivan,
Estimated Spending on AI Data Centers has been referred to As the largest peacetime investment project in history. Rajat Rana, partner at Quinn Emanuel Urquhart & Sullivan, told CNBC that he would take it a step further and emphasize that it is “the largest peacetime investment project in human history, financed largely from the balance sheet.”
Rana, who worked on structured finance litigation in the wake of the housing crisis caused by the 2008 financial crisis, said tracking the growth in AI data center financing feels like “deja vu.”
“We’re talking about trillions of dollars, and almost going back to the same cycle where there is almost no transparency about financing structures – the scale is huge,” he said.
The AI boom is not only accelerating demand for facilities, but it is also driving rapid advances in power generation and chips – critical technology that resides in data centers. Advances in the sector and the flow of large amounts of money create both risks and rewards for insurers and lenders.
special policies
Data centers require a special approach from insurers, covering both real estate and technical assets. Gallagher’s Harper said some of the world’s largest insurers are creating data center-specific pathways to managing projects.
The facilities present unique challenges because of the high concentrations of value, required power generation and “bleeding edge technology,” which typically provide them with advantageous pricing and make them “very desirable,” Harper told CNBC.
Insurers want to spread the risk, which reduces costs. But problems arise when your $20 billion of assets are concentrated in an area prone to high winds or hurricanes, he said.
Disruptions in the supply chain can increase complexity when it leads to the concentration of high-value equipment that has not yet been installed. Customers are importing large dollar shipments from overseas and then storing them — often in facilities they don’t own or operate — which creates additional risks, he said.
The M&A boom is also keeping transactional lawyers busy, with Kirkland & Ellis noting that many companies Formation of data center specific teamsEnlisting experts in real estate, power, telecom, finance, insurance, trading, private equity and cyber security.
professional services firm swamp launched a Dedicated Digital Infrastructure Advisory Group It is designed to help customers as contracts become increasingly complex.
Last year, Marsh also launched Nimbus, a 1 billion euro ($1.2 billion) insurance facility to cover the construction of data centers in the UK and Europe. Seven months later, it Expansion Facility to offer limits up to $2.7 billion.
“Private debt can meaningfully complement banks and support non-hyperscale contracted offtake,” said Alex Wolfson, senior vice president of credit specialties at Marsh Risk.
Wolfson noted that as data center debt grows, insurers that protect lenders if borrowers don’t pay are beginning to have their limits hit. Marsh is working on a solution to support lenders.
However, Quinn Emanuel’s Rana cautioned that when it comes to data centers, it is not easy for insurers to fully understand the risk because financing moves off the balance sheet.
He said that in January, four US senators called The government has been asked to investigate how Big Tech is increasingly “turning to complex and opaque debt markets to borrow excessive cash.” In an open letter, the senators warned that the huge debt load could cause “unsustainable losses” for financial institutions, leading to a broader financial crisis that hurts the economy.
Increased opacity in financing could create second-order litigation risks for downstream investors such as pension funds, insurers and asset managers investing in private credit funds, who later discover they were not fully informed about concentration risk, Rana said. said in a note Published in March.
He told CNBC that some PE funds have reached out to him with concerns about commercial leases and valuation of properties.
Tenants are trying to negotiate expansion of their properties and landlords are disputing the value as they seek higher prices for AI data centres.
Rana said, “I’m not a doomsday person saying, hey, it’s going to crash. What I’m saying is whether it crashes or not, controversies are inevitable, and we’ve already seen those controversies.”
‘GPU Loan Treadmill’
A major debate about potential cracks in financing centers on GPUs and the risk of their life cycles not aligning with the long lifespan of those features.
corewaveThe company selling AI technology in the cloud is the first to secure GPU-backed loans, essentially using the value of high-performance chips as collateral. Last week, the company announced It secured $8.5 billion in the first investment-grade rated GPU-backed deal. Its stock jumped 12% that day.
While the lifecycle of data centers is typically decades long, the average lifecycle of a GPU is about seven years.
“There are different data centers that are raising debt by exposing investors to different life cycles,” Rana said. He referred to the problem as the “GPU debt treadmill,” a phrase fabricated By AI commentator Dave Friedman.
“It’s almost like a treadmill that these AI data centers are running on,” Rana told CNBC. Even if the financing structure is ring-fenced and backed by investment-grade counterparty, the real risk may lie in whether today’s equity issue develops into a credit problem over time.
“As these new chips come out, data centers will feel pressure to raise more debt, and then they’ll have to build new infrastructure, and then it basically creates a billion dollar question: How fast can you build these facilities? How fast can you raise debt?”
Harper says the cost of financing these projects is likely to continue the recent increase in asset-backed securitization deals, as large amounts of commercial mortgage-backed securities will be sold to investors.
For some insurers like Gallagher, the changing dynamics in the sector are opportunities rather than challenges. Harper said the lifecycle of GPUs is increasing. Where things depreciated rapidly, Gallagher had to get creative and write special insurance policies with predetermined agreements about the value of the assets.
“With the size and scope of these (facilities) it would be a nightmare to determine the value of each individual unit,” he said.
Harper also emphasized that the GPUs are interchangeable. The firm has observed that operators anticipate relatively shorter life cycles and build facilities that are more modular in response.
“There is a main tension in data center project finance: Lenders typically want asset lives that exceed the loan term by a comfortable margin, and the short useful life of GPUs challenges that assumption,” said Marsh Risk’s Wolfson.
So lenders are structuring loans more carefully to protect themselves.
