Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
Enter your email address below and subscribe to our newsletter

The largest capital spending cycle on record is beginning to strain the financial system that underpins it. Goldman Sachs now projects that the four leading hyperscale cloud companies — Meta, Microsoft, Amazon, and Alphabet — will collectively spend $5.3 trillion on AI and data center capital expenditures from 2025 through 2030, up from a prior estimate of $4.5 trillion before first-quarter earnings reports.prismnews
But the bank’s chief credit strategist, Amanda Lynam, cautioned in a June report that the sheer scale of spending is pushing against the limits of traditional financing. “We expect liquid credit market saturation and issuer concentration constraints to become somewhat more binding in coming years,” Lynam wrote, noting that hyperscalers already account for a large share of new corporate borrowing and that public debt investors may balk at absorbing ever-larger amounts from the same handful of issuers.goldmansachs
Goldman’s analysts observed that AI-related capex estimates are climbing faster than actual data center construction, warning that “future bottlenecks may shift from model demand to financing capacity, electricity supply, and project execution”. The bank estimates hyperscaler capex could reach roughly $1.1 trillion in 2027, well above the $920 billion Wall Street consensus, with a bullish scenario reaching $1.4 trillion.bitget
Morgan Stanley has offered a complementary view, estimating that global data center construction capex will reach approximately $2.9 trillion through 2028. Of that, only about $1.4 trillion would come from hyperscalers’ own cash flows, leaving a $1.5 trillion financing gap to be filled by private credit, corporate bonds, securitized products, and sovereign capital.msn
Goldman sees private infrastructure and real estate funds as essential to bridging the gap. Private infrastructure funds raised a record $221 billion in 2025 and held roughly $400 billion in dry powder as of September that year. Lynam’s report suggested total private infrastructure assets under management could exceed $3 trillion by 2030 if growth accelerates.goldmansachs
The complexity of data center financing amplifies the challenge. A single facility combines land, power access, network infrastructure, cooling systems, and servers — spreading financing needs across infrastructure funds, real estate vehicles, private credit, and corporate bond markets. Should a market correction occur, the transmission of losses would be far more complex than during previous technology booms.x
On the demand side, rising token costs are already prompting corporate belt-tightening. Companies including Uber and Walmart have imposed caps on internal AI usage, while OpenAI CEO Sam Altman acknowledged this month that costs have become a “major challenge” for customers. Goldman has flagged these dynamics as a risk to the revenue growth assumptions underlying AI lab valuations, even as AI-related investment amounts to roughly 1.5 percent of U.S. GDP — still below the 2 to 3 percent peaks seen during past infrastructure booms like railroads and electrification.businessinsider