CURRENTS OF RISK
Goes BUST?
Hyperscale Capital Spending
No technological boom in history has drawn bigger bets than today’s race to dominate AI.
Analysts forecast that trillions will be spent on AI infrastructure over the next five years — on chips, data centers, and the power to run them.
Jim Footh
Managing Director, Global Data Center Investments, Real Estate, PGIM
“
In all likelihood, there’s going to be consolidation amongst the large AI companies. It’s hard for me to believe that these hundreds of billions of dollars [in infrastructure spending] by all of these companies is sustainable.”
The world’s largest tech firms have spent hundreds of billions on AI and will spend even more in the years ahead, according to Bloomberg estimates, mostly on data centers used to train and run large language models (LLMs) like ChatGPT.
But cracks are appearing in the large language model thesis, including high costs, diminishing returns and reliability concerns.
Carousel Controller (Do NOT delete)
Today’s AI buildout could become tomorrow’s overbuild.
*Oracle, Apple, Tesla Sources: Company filings, Bloomberg Intelligence, Nov. 2024
Forecasts are not guaranteed and may not be a reliable indicator of future results.
Amazon
Microsoft
Meta
Google
Tier 2 Cloud*
*Asset-backed securities, commercial mortgage-backed securities Sources: Morgan Stanley, Bloomberg, Oct. 2025
Private credit can help Big Tech borrow off balance sheet
Non-Depository Financial Institutions
One of the engines of the AI infrastructure boom is debt.
According to Bloomberg News, approximately $2.9 trillion is needed to fund data centers through 2028, including $1.6 trillion for hardware (chips and servers) and $1.3 trillion for infrastructure (real estate, construction and maintenance).
The gamble is that demand will prove enduring and profitable. The risk is trillions in stranded assets if it doesn’t.
Source: Bloomberg, Nov. 2025
The world’s biggest tech companies are pouring fortunes into hyperscale campuses. But in a market that will likely see consolidation, which company will rise to the occasion—and which companies could see their investments disappear?
The Great Data Center Race
Consolidation adds another risk.
If the winner-take-all AI ecosystem narrows to a handful of dominant platforms, not every company will need its own hyperscale footprint. That would leave entire campuses underutilized or even abandoned, with billions invested in stranded assets.
The dot-com era left behind fiber-optic cables and network infrastructure that continue to serve the global economy. AI may leave a far less durable legacy. While data-center shells might be repurposed, the silicon at their core could turn into scrap.
AI BOOM
WHAT IF THE
Source: BloombergNEF
Power Demand From AI Data Centers to Quadruple in 10 Years
Source: PJM Interconnection
U.S. utilities are projected to invest over $200 billion annually in grid upgrades through 2026. Financed largely with debt and locked into regulated rate structures for decades, structurally higher energy costs will persist whether hyperscale data centers operate at full tilt or sit half-empty.
Source: US Energy Information Administration
Electricity costs are near a record high
Utility Bills Keep Rising
Utility Bills Keep Rising
Source: McKinsey, Oct. 2025
(Projected through 2032)
“
