JupiteX Get the app
Markets & Finance21 Nov 2026 · about 7 min

What The Telecom Bust Taught Me About AI’s Debt Boom

The brief

Equity financing gives companies money in exchange for ownership; debt is borrowed money that must be repaid. A shift means more AI infrastructure is funded by borrowing rather than investors buying equity. The distinction matters because debt adds scheduled obligations even if a project earns less than expected. Venture funding reached $679 billion globally in the first three quarters of 2026, and AI companies attracted $102 billion in the third quarter. But major firms are also issuing bonds. Meta sold $30 billion in bonds last October, then $25 billion more in April. Private credit is another expected source of funding. Debt can let companies build without selling as much ownership, but it creates fixed repayment obligations. Equity investors generally absorb losses through falling valuations, giving companies more flexibility. The shift does not prove AI investment will fail; it changes who bears risk if cash flows fall short or arrive late. Morgan Stanley expects about $3 trillion in infrastructure spending through 2028, with only about half from companies’ own cash flow.

01

What does it mean that financing for the AI buildout is shifting from equity to debt?

Equity financing gives companies money in exchange for ownership; debt is borrowed money that must be repaid. A shift means more AI infrastructure is funded by borrowing rather than investors buying equity. The distinction matters because debt adds scheduled obligations even if a project earns less than expected.

Venture funding reached $679 billion globally in the first three quarters of 2026, and AI companies attracted $102 billion in the third quarter. But major firms are also issuing bonds. Meta sold $30 billion in bonds last October, then $25 billion more in April. Private credit is another expected source of funding.

Debt can let companies build without selling as much ownership, but it creates fixed repayment obligations. Equity investors generally absorb losses through falling valuations, giving companies more flexibility. The shift does not prove AI investment will fail; it changes who bears risk if cash flows fall short or arrive late. Morgan Stanley expects about $3 trillion in infrastructure spending through 2028, with only about half from companies’ own cash flow.

02

How much do major AI and cloud companies expect to spend on infrastructure through 2028, and how much is expected to come from their own cash flow?

Morgan Stanley expects major AI and cloud companies to spend about $3 trillion on infrastructure through 2028. The article says only about half of that spending is expected to come from the companies’ own cash flow. The scale matters because the remaining funding must come from other sources.

Those sources increasingly include bond markets and private credit. Large companies have already borrowed tens of billions of dollars. Meta, for example, sold $30 billion in bonds last October and another $25 billion in April. A joint-venture partner also raised a record $27 billion bond to finance Meta’s Louisiana data-center campus.

The article does not give a precise dollar figure for the share not covered by cash flow; it describes it as roughly half. It says that portion has to come from somewhere, increasingly borrowing. This makes financing conditions and repayment demands important to the buildout’s future. The central question is whether the cash arrives on the schedule required by the debt.

03

Why can borrowing become especially risky if revenue arrives later than expected?

Borrowing creates fixed repayment obligations, while revenue can arrive later than a company expects. If cash comes in too slowly, a company may face debt payments before its projects generate enough money. That timing mismatch is why debt can make a mistaken forecast especially costly.

The article contrasts debt with equity. Equity investors generally absorb losses through declining valuations, and companies may retain more flexibility to pursue long-term plans. By contrast, the writer says equity can usually wait out a wrong timeline in a way debt cannot. The telecom industry provides a warning: companies borrowed heavily against a demand forecast that proved far too optimistic.

The AI buildout may have strong businesses and real demand, the writer acknowledges, but that does not settle whether revenue will arrive in time. Morgan Stanley expects about $3 trillion in AI and cloud infrastructure spending through 2028, with only about half coming from companies’ own cash flow. More reliance on borrowing makes repayment timing a central risk for lenders and companies.

04

What internet-traffic forecast fueled telecom borrowing in the 1990s, and how did actual growth compare?

In the late 1990s, telecom companies borrowed heavily against the claim that internet traffic was doubling every 100 days. The article describes this as a widely repeated demand forecast that was rarely checked. That projection helped justify staggering sums of borrowing for network infrastructure.

Actual traffic growth was still rapid, but much slower: it doubled about once a year. The gap mattered because borrowing and building decisions were based on a much faster increase in demand than the industry experienced. By 2002, the telecom sector carried roughly $1 trillion in debt, and the FCC chairman told Congress much of it would never be repaid.

WorldCom entered what was then the largest bankruptcy in American history. The article draws a parallel to today’s AI demand claims, while noting that the internet’s growth did arrive and transformed society. Its caution is that strong long-term demand does not answer whether cash arrives quickly enough to meet debt payments.

05

What happened to telecom companies and their unused fiber after the sector’s debt-fueled bust?

The telecom boom left companies with enormous debt and infrastructure built for demand that arrived more slowly than forecast. By 2002, the sector carried roughly $1 trillion in debt. The FCC chairman told Congress that much of it would never be repaid, and WorldCom entered what was then the largest bankruptcy in American history.

Much of the fiber laid during the boom remained dark, meaning it was not being used, for years. It eventually went to buyers who paid pennies on the dollar for it in bankruptcy. The purchases came through a decade of restructurings and private mergers that rebuilt entire company ownership and financing structures.

Some companies survived that process, and many did not. The example shows that useful infrastructure can outlast the businesses and financing arrangements that first paid for it. The article raises this history as a caution for AI investment: if demand or revenue falls behind expectations, companies with debt may have less room to wait than equity-backed businesses.

06

How can quarterly audited results help investors check claims that demand for data centers is already secured?

The current demand claim is that data centers are effectively leased before they are built, and that the main mistake would be building too little. The article says much of this narrative rests on projections supplied by the companies doing the building. When investors hear the same claim repeatedly, they still need evidence to check it.

The writer points to companies reporting financial results every quarter. The article identifies Paradox as public and says investors can compare the demand story with audited numbers once companies of that scale report results. Regular reporting can make claims about demand and revenue easier to scrutinize over time.

Audited results provide a check, not a guarantee that future demand will match forecasts. The article does not claim that one report settles the question. It says investors have begun demanding answers, and that company results can help test the narrative. For lenders, the crucial issue remains whether cash arrives on the schedule debt requires.

07

What is a company’s capital structure, and why does its mix of debt and equity affect who bears losses?

The article does not define capital structure. In finance, it means the mix of borrowed money and ownership investment a company uses to fund itself. Debt is generally a claim that must be repaid; equity represents an ownership stake. The balance matters because the two sources place risk and obligations differently.

The article describes the AI buildout as shifting from equity toward debt. With equity, losses typically appear as declining valuations, and companies may retain flexibility to pursue long-term plans. With debt, fixed repayment obligations remain even when a project underperforms. Meta’s bond sales and borrowing by other major technology companies illustrate the growing role of debt financing.

A more debt-heavy structure can make the timing of cash flow especially important. The article’s central lender question is whether cash arrives on the schedule debt requires. If demand or revenue comes later than expected, equity can generally wait out a wrong timeline more easily. The financing mix therefore affects both the company’s flexibility and who bears losses.

This brief was written by AI from the original reporting and checked by other models. Names, figures and quotes come from the source; read it for full context.

Read more in the JupiteX app

Pulse is free. New stories every 4 hours, each one broken into the questions that explain it.

Or read more news on the web