Is AI Pushing the Technology Sector Back Toward Junk Bonds?

AI tools—including search engines, report generators, and increasingly autonomous agents—are advancing rapidly, addressing a wide range of consumer and industrial use cases. While early results have been promising, the scale of organizational change required, combined with structural and financial constraints, makes a clear and near-term return on investment (ROI) unlikely for most businesses.

This uncertainty has become a meaningful headwind for technology companies that have made substantial bets on AI, often at the expense of their existing software and platform business models. These companies have largely positioned AI as a long-term investment, arguing that current costs will be justified by future gains. However, capital markets are rarely patient. Investor attention tends to shift quickly, and from a capital allocation perspective, commodities such as gold, silver, and other hard assets are increasingly competing for attention.

Historically, most software companies have operated as asset-light businesses. Their models emphasized intellectual property, services, and virtual assets, with minimal investment in physical infrastructure. This structure supported high margins, as marginal production costs were close to zero, aside from research and development expenses.

AI changes that equation. Competing at scale now requires access to large data centers, advanced computing infrastructure, and high-end accelerators such as NVIDIA GPUs. Training large language models and deploying agentic AI systems have forced many software companies to rethink their operating and capital structures. As a result, hardware investments have increased sharply, drawing greater scrutiny from analysts. Valuations, balance sheets, and accumulated debt are now under closer examination.

Additional pressure comes from falling chip prices, increasingly rapid three-year hardware refresh cycles, and unresolved questions around asset capitalization and depreciation. Together, these factors are raising concerns about long-term sustainability, particularly in the absence of stable, recurring AI-driven revenue streams.

A core issue remains unresolved: there is still limited visibility into commercially scaled AI revenue or credible forecasts that clearly link AI deployments to durable financial returns. Many organizations remain dependent on ecosystem-wide changes that could take several years, including upgrades to hardware, software platforms, data architectures, and operating models. These dependencies are rarely emphasized in marketing narratives or sales discussions around AI adoption.

For much of the past decade, relatively low interest rates and subdued bond markets supported elevated risk-taking across capital markets. This environment contributed to the investment euphoria surrounding AI over the past three years. However, media reports indicate that hundreds of billions of dollars in new debt may be raised over the next two years by data center operators and infrastructure providers to support AI expansion, as equity financing alone may no longer be sufficient amid cooling investor interest.

If bond markets become less accommodating, or if commodities continue to divert capital away from technology equities, the financial dynamics could shift quickly. At the same time, if AI deployments continue to underperform relative to ROI expectations, executives may become increasingly reluctant to publicly defend or expand these investment plans.

There is historical precedent. When the dot-com bubble burst, many telecommunications companies and equipment suppliers experienced sharp stock declines, mounting debt burdens, and eventual downgrades to junk status—creating opportunities for distressed and vulture investors. It has been some time since capital markets have seen a cycle dominated by such dynamics. Whether AI investment leads the technology sector back toward similar conditions remains uncertain, but the parallels are becoming harder to ignore.