What The Telecom Bust Taught Me About AI’s Debt Boom
As artificial intelligence infrastructure shifts from equity financing to massive corporate debt, tech market veterans see striking parallels to the late 1990s telecom bust and its risky reliance on leverage.
As the funding model for artificial intelligence infrastructure pivots from equity to debt, veterans of the telecom crash may recognize familiar patterns.
Stephen Hawking’s “A Brief History of Time” features a classic anecdote about a lecturer discussing the universe, only for an audience member to claim the world rests on a giant turtle. When asked what supports that turtle, she answers that it represents “turtles all the way down.” Philosophers call this infinite regress—an unbroken chain of claims lacking any solid foundation.
This dynamic comes to mind when examining the AI expansion, as its foundational logic mirrors this exact structure. Valuations justify borrowing, and that debt funds new capacity. The resulting capacity then serves as validation for the original valuations, leaving each layer resting entirely on the one beneath it.
Lately, a larger portion of the tech stack relies on debt rather than equity. In practice, crossing that threshold signals that the telecom bubble becomes a relevant historical parallel, especially since leverage can fundamentally alter how these market cycles conclude.
Until recently, the AI expansion relied heavily on equity financing, driving venture capital to historic highs.
According to Crunchbase data, worldwide venture funding hit $679 billion across the first three quarters of 2026, marking the highest recorded total for that timeframe. Artificial intelligence ventures secured $102 billion during the third quarter alone, representing nearly 64% of all global venture financing. A record-breaking 27 companies closed rounds of $1 billion or more within that same quarter, combining to make up roughly one-third of the aggregate total.
While equity financing introduces distinct risks, it typically lacks fixed repayment terms. If ventures underperform, equity backers generally absorb losses via dropping valuations, while firms maintain greater operational flexibility to execute long-term strategies.
However, debt has now arrived on a massive scale. Meta issued $30 billion in bonds last October—one of the largest corporate debt offerings in history—before issuing an additional $25 billion in April, reportedly facing higher investor yield demands the second time around. Amazon, Alphabet, and Oracle have also secured tens of billions in debt individually. Furthermore, Meta’s massive Louisiana data-center campus relied on a record $27 billion bond floated by a joint-venture partner instead of Meta directly. Morgan Stanley projects that major AI and cloud providers will pour roughly $3 trillion into infrastructure through 2028, with only about half funded by internal cash flow. The remainder must come from external sources, which increasingly points toward the bond market and private credit.
Having witnessed the earlier iteration of this phenomenon firsthand—initially as a venture investor during the late 1990s and subsequently as a lender to growth-stage tech firms—the parallels are striking. The telecom sector took on staggering liabilities backed by widely accepted yet rarely verified demand forecasts, such as the famous assertion that internet traffic doubled every 100 days.
In reality, traffic doubled roughly once a year—impressive expansion, but merely a fraction of what borrowing models assumed. By 2002, the telecom industry carried approximately $1 trillion in debt, much of which the FCC chairman informed Congress would never be repaid, while WorldCom plunged into what was then the largest bankruptcy in U.S. history.
A vast amount of the newly laid fiber sat dark for years before eventually being utilized by buyers who acquired assets for pennies on the dollar through bankruptcy proceedings, enduring a decade of restructurings and private mergers that completely overhauled balance sheets. Some enterprises successfully navigated that turmoil, while many others collapsed.
The demand narrative
The prevailing demand narrative echoes past cycles and raises the most concern. Today’s consensus suggests data centers are essentially pre-leased prior to construction, meaning the sole risk is under-building. That thesis may prove accurate, but it is echoed with the exact uniformity seen in the 1999 traffic forecasts, heavily relying on projections generated by the very companies constructing the facilities. Lately, however, investors have started demanding accountability.
With Anthropic now public, enterprises of that magnitude must release quarterly financial reports, enabling investors to test demand claims against audited metrics.
Skeptics may reasonably argue that this comparison falls short, noting that the largest corporate borrowers today rank among the most profitable companies in history, boasting cash flows telecom operators never approached. Additionally, select AI startups are scaling revenues faster than any previous business model.
The internet fundamentally transformed society, and the anticipated data traffic ultimately materialized. Even if all of that holds true, it bypasses the fundamental question every lender must address: whether cash inflows will arrive on the precise schedule mandated by debt obligations. Unlike debt, equity can generally endure an extended timeline. When borrowed capital expands infrastructure faster than incoming revenue can service the debt, corrections typically manifest through formal restructurings rather than quiet market markdowns.
Enduring lessons
None of this implies that the AI buildout must culminate in a telecom-style crash. The most crucial takeaway from the telecom bust centered on capital structure, which determined survivor status once consumer demand finally aligned with available capacity, long after the underlying technology had proved its worth. Experience shows that while growth companies often fear debt unnecessarily, that principle remains strictly conditional.
Debt benefits borrowers whose liabilities align with visible revenue streams and lenders who underwrite fundamental business metrics rather than hype-driven narratives, but it proves unforgiving to everyone else. The borrowing supporting the AI boom will ultimately be measured by that standard. Every core assumption may prove correct, but historical precedent shows that once investors begin questioning the foundation, they rarely stop until they find solid ground.
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.
?Frequently Asked Questions
01How is the AI buildout being financed compared to earlier stages?
While the AI buildout was initially financed largely through equity and record-setting venture capital, it is increasingly shifting toward debt financing, including corporate bond sales and private credit.
02Why is debt riskier than equity for funding infrastructure?
Equity generally does not impose fixed repayment obligations, allowing companies to absorb underperformance through declining valuations. Debt, however, requires regular cash payments on a strict schedule regardless of whether revenue has materialized.
03What lessons does the telecom bubble offer for the current AI boom?
The telecom crash highlights the dangers of taking on massive debt based on aggressive, unverified demand forecasts. It demonstrates that capital structure and cash timing dictate which companies survive when capacity outpaces actual revenue growth.
Illustration: Dom Guzman
Thi Nien
Thi Nien is an AI, finance and global research analyst, specializing in global markets, macroeconomics, AI infrastructure, startups and emerging technologies. Her work focuses on analyzing the trends shaping the future economy, including artificial intelligence, institutional capital flows, digital assets and global financial innovation.
More from this author


