Skip to content
The Next Marketers
November 20, 2026
Bitcoin-3.56%
Ethereum-3.86%
Solana-5.01%
The Next Marketers
Artificial intelligence

Lessons from the Telecom Bust Apply to AI’s Debt Boom

As artificial intelligence financing transitions from equity to debt, parallels emerge with the late-1990s telecom bubble. Massive bond issuances by tech giants raise questions about whether cash flows will meet strict debt obligations.

Lessons from the Telecom Bust Apply to AI’s Debt Boom

As financing for the artificial intelligence expansion transitions from equity to debt, veterans of the late-1990s telecom bubble may recognize the implications of that shift.

Stephen Hawking’s “A Brief History of Time” features an old anecdote about a scientist lecturing on the cosmos before a woman in the audience insists the Earth rests on the back of a giant turtle. When asked what supports the turtle, she responds that it is “turtles all the way down.” Philosophers call this infinite regress—a sequence of assertions where each depends on another, and solid footing is never actually reached.

This dynamic comes to mind when examining the AI expansion, as its foundational logic has adopted a similar structure. Valuations support borrowing, and borrowing constructs capacity. That capacity then serves as proof that the initial valuations were correct, meaning each assertion leans entirely on the one supporting it.

Over the past year, a larger portion of the technology stack has transitioned to debt financing rather than equity. Based on my experience, this crossing point makes the telecom bubble a relevant historical parallel, because debt fundamentally alters how these financial narratives conclude.

Until recently, the AI expansion operated primarily as an equity-driven narrative, accompanied by record levels of venture investment.

Data from Crunchbase reveals that global venture funding reached $679 billion through the first three quarters of 2026, marking the highest recorded amount for that timeframe. Artificial intelligence companies secured $102 billion during the third quarter alone, representing approximately 64% of global venture funding. Furthermore, a record 27 companies closed rounds of $1 billion or more in that same quarter, collectively accounting for roughly one-third of that overall total.

Although equity financing carries substantial risks, it generally does not impose fixed repayment mandates. When investments underperform, equity investors typically absorb losses via declining valuations, leaving companies with greater operational flexibility to pursue long-term strategies.

That debt is now arriving on a massive scale. Meta issued $30 billion in bonds last October in one of the largest corporate debt offerings on record, followed by another $25 billion in April, with investors reportedly demanding higher compensation during the second offering. Amazon, Alphabet, and Oracle have each borrowed tens of billions independently, while Meta’s expansive Louisiana data-center campus was funded by a record $27 billion bond secured by a joint-venture partner rather than Meta itself. Morgan Stanley projects that major AI and cloud providers will invest roughly $3 trillion in infrastructure through 2028, with only about half of that capital expected to originate from internal cash flow. The remaining balance must come from external sources, which increasingly appear to be the bond market and private credit.

I observed an earlier iteration of this cycle from a front-row perspective, initially working as a venture capitalist in the late 1990s and subsequently as a lender to growth-stage technology businesses. The telecom industry borrowed staggering amounts against demand forecasts that were widely repeated yet rarely verified—such as the famous assertion that internet traffic was doubling every 100 days.

In reality, traffic was doubling about once a year. While that represented spectacular growth, it amounted to only a fraction of the assumptions underpinning the borrowing. By 2002, the telecom sector carried roughly $1 trillion in debt—much of which the Federal Communications Commission chairman told Congress would never be repaid—and WorldCom had filed for what was then the largest bankruptcy in American history.

A significant amount of the fiber-optic cable laid during that era sat dark for years before eventually being utilized by buyers who acquired it for pennies on the dollar through bankruptcy proceedings, spanning a decade of restructurings and private mergers that completely overhauled capital structures. Some enterprises survived that grueling process, while many others failed.

The demand narrative

The parallel that raises the most concern for me is the prevailing demand narrative. Today’s argument insists that data centers are effectively leased before construction even begins, meaning the only true risk is building too little capacity. While that perspective may ultimately prove correct, it is repeated with the exact same uniformity that the internet traffic forecast enjoyed in 1999, and much of it relies on projections provided by the very companies constructing the facilities. What has shifted in recent months is that investors have begun asking for verification.

With Anthropic now operating publicly, enterprises of that magnitude report financial results on a quarterly basis, allowing investors to cross-reference the demand story against audited figures.

Skeptics may reasonably argue that my comparison is flawed, noting that the companies engaging in the largest borrowings rank among the most profitable enterprises in history, boasting cash flows that telecom carriers never approached. Additionally, certain AI enterprises appear to be achieving revenue scale faster than any businesses before them.

It is equally true that the internet ultimately transformed everything, just as the predicted internet traffic eventually materialized. However, none of these points answer the fundamental question a lender must ask: whether cash arrives on the precise schedule required by the debt obligations. Equity can generally weather an inaccurate timeline in a way that debt cannot. When borrowed capital builds capacity faster than incoming revenue can service those liabilities, corrections tend to manifest through formal restructurings rather than quiet asset markdowns.

Enduring lessons

None of this serves as a prediction that the AI buildout must conclude in the same manner as the telecom crash. In my view, the most enduring lesson from the telecom bust involved capital structure—specifically, determining which organizations remained solvent long after demand finally caught up with capacity and the underlying technology had been vindicated. I have spent years asserting that growth companies fear debt more than is warranted, and I maintain that viewpoint, though that argument has always been conditional.

Debt effectively serves borrowers whose liabilities align with visible, verifiable revenue, as well as lenders who underwrite fundamental metrics instead of marketing narratives. It treats all other scenarios harshly. I believe the debt accumulated beneath the AI boom will ultimately be evaluated against that exact standard. While every assumption in the financial stack may prove accurate, the lesson from the past is that once investors begin questioning what the underlying turtles are standing on, they rarely stop until they unearth something solid.


David Spreng is the founder, co-CEO and co-CIO of Runway Growth Capital. He has more than 30 years of experience as a Silicon Valley venture capitalist and growth debt lender, has been involved in the formation and development of nearly 50 technology companies, and has appeared four times on the Forbes Midas List of top venture capitalists. He co-founded Decathlon Capital Partners in 2010, served on the board of the National Venture Capital Association and has participated as a panelist at the World Economic Forum in Davos.

Note: This article expresses the author’s views as of the publication date and is provided for general informational and educational purposes. It is not investment advice or a recommendation to purchase, sell, or hold any security or to enter into any financing transaction. References to companies and transactions are illustrative and do not reflect a conclusion regarding their investment merits or creditworthiness. Historical market comparisons are inherently limited, and future conditions may differ materially. The author’s views may change without notice.

Illustration: Dom Guzman

Frequently Asked Questions

How is the financing of the AI buildout changing?

The financing of the AI expansion is shifting away from equity toward debt, with major technology companies and joint ventures issuing billions of dollars in corporate bonds and utilizing private credit.

How much infrastructure spending is projected for AI and cloud companies?

Morgan Stanley expects major AI and cloud corporations to spend approximately $3 trillion on infrastructure through 2028, with only about half anticipated to come from internal cash flow.

What are the risks of debt compared to equity financing?

Equity financing generally does not impose fixed repayment obligations, allowing investors to absorb losses through declining valuations during underperformance. In contrast, debt requires strict, scheduled repayments that can force restructurings if revenue fails to match the timeline.

What historical event does the author compare the current AI debt boom to?

The author compares the current shift to the late-1990s telecom bubble, during which companies accumulated staggering amounts of debt against overhyped demand forecasts before facing widespread bankruptcies and restructurings.

Who is the author of this article?

The article was written by David Spreng, founder, co-CEO, and co-CIO of Runway Growth Capital, who has over 30 years of experience as a Silicon Valley venture capitalist and growth debt lender.


Related stories

Comments 0 responses

Join the discussion

Comments are moderated and appear after review.