AI & ComputeLOT 261120 · 4 min read
Meta's $55 Billion Bond Binge Signals AI Buildout Runs on Debt
Runway Growth Capital's David Spreng warns AI's shift to debt financing mirrors telecom's $1 trillion bust, as Meta, Amazon and Oracle borrow tens of billions against unverified demand.
Key takeaways
- Meta sold $30 billion in bonds in October 2025 and another $25 billion in April 2026
- Morgan Stanley expects AI and cloud companies to spend about $3 trillion on infrastructure through 2028, only half funded by cash flow
- Global venture funding hit a record $679 billion in the first three quarters of 2026, with AI taking $102 billion in Q3 alone
- Telecom sector carried roughly $1 trillion in debt by 2002 after traffic forecasts proved wrong by a factor of about 3.5x
- Anthropic is now public, allowing investors to check the AI demand story against audited quarterly numbers

Meta sold $30 billion of bonds last October and returned for $25 billion more in April, and that $55 billion in six months marks the point where the AI buildout stopped being an equity story. So argues David Spreng, founder and co-CEO of Runway Growth Capital, who watched the telecom implosion from a front-row seat as a venture capitalist in the late 1990s.
His warning is blunt. The financing of AI infrastructure is shifting from equity to debt, and in his experience that crossing is precisely when the telecom bubble becomes the relevant history. Debt, unlike equity, imposes fixed repayment obligations that equity investors never face.
How large is the debt?
The numbers are already substantial. Meta's October sale ranked among the largest corporate debt offerings on record, and investors reportedly demanded more compensation the second time around, in April, after the company raised its spending outlook.
Amazon, Alphabet and Oracle have each borrowed tens of billions of their own, according to the report. Meta's Louisiana data-center campus was financed with a record $27 billion bond raised not by Meta itself but by a joint-venture partner.
Morgan Stanley expects the large AI and cloud companies to spend roughly $3 trillion on infrastructure through 2028. Only about half of that, the bank estimates, will come from their own cash flow. The remainder must come from somewhere, and Spreng believes that somewhere is increasingly the bond market and private credit.
What happened in telecom?
The parallel Spreng draws is specific. Telecom carriers borrowed staggering sums against a demand forecast that was widely repeated and rarely checked — the claim that internet traffic was doubling every 100 days. Traffic was actually doubling about once a year. That is spectacular growth, but a fraction of what the borrowing assumed.
By 2002 the telecom sector carried roughly $1 trillion in debt, much of which the FCC chairman told Congress would never be repaid. WorldCom entered what was then the largest bankruptcy in American history. Much of the fiber the industry laid sat dark for years, and buyers eventually put it to work after paying pennies on the dollar in bankruptcy.
Is the demand narrative checked?
The echo that concerns Spreng most is the demand narrative. Today's version holds that data centers are effectively leased before they are built and that the only real mistake is building too little. He notes the claim may prove correct, but it is repeated with the same uniformity the traffic forecast enjoyed in 1999, and much of it rests on projections supplied by the companies doing the building.
One thing has changed in recent months: investors have begun demanding an answer. Anthropic is now public, and once companies of that scale report quarterly results, investors can check the demand story against audited numbers.
Spreng concedes the comparison may fail. The largest borrowers may be among the most profitable enterprises ever built, with cash flows the telecom carriers never approached. A few AI companies also appear to be reaching revenue scale faster than any businesses before them. The internet, he notes, really did change everything, and the traffic really did arrive.
None of that, in his view, answers the question a lender has to ask: whether the cash arrives on the schedule the debt requires. Equity can wait out a wrong timeline in a way debt cannot. When borrowed money builds capacity faster than revenue arrives to service it, the correction tends to come through restructurings rather than quiet markdowns, he writes.
What is the durable lesson?
Spreng is careful to state this is not a prediction that the AI buildout must end the way telecom did. The most durable lesson of the telecom bust concerned capital structure, which decided who was still standing when demand finally caught up with capacity, long after the technology itself had been vindicated.
He has spent years arguing that growth companies fear debt more than they should, and he still believes it, but the argument has always been conditional. Debt serves borrowers whose obligations are sized to revenue they can see and lenders who underwrite fundamentals instead of narratives, and it treats everyone else harshly.
His closing image borrows from Stephen Hawking's "A Brief History of Time": every assumption in the stack may well hold, but once investors begin asking what the turtles are standing on, they rarely stop until they reach something solid.
The next test of that stack arrives with Anthropic's audited quarterly filings, where the demand story meets numbers that cannot be narrated away.
news.crunchbase.com (Source)
More from Sophie Lindqvist
Show full bio
News editor covering money, funding and market shifts around Semiconductors, chip manufacturing, hardware, AI compute at Chip Dispatch.
1,194 articles