Tech companies went on a borrowing spree to fund their AI ambitions. Now some of their bonds are starting to look like bargains, at least to one of the sharpest eyes in corporate credit.
Robert Cohen, director of corporate credit at DoubleLine, is making a case that high-quality technology bonds, specifically those issued by Alphabet and Amazon, offer compelling value after AI-related borrowing pressures dragged down valuations across the sector. It’s a nuanced position: Cohen simultaneously believes an AI credit bubble is forming with near-certainty, yet argues that the best names in the space have been punished alongside the rest.
The numbers behind the borrowing binge
The scale of issuance is staggering. Hyperscalers have collectively issued more than $155 billion in unsecured bonds in 2026, representing an increase of over 45% compared to the full year of 2025. Bloomberg Intelligence projects roughly $5 trillion in AI-related capital expenditures over the next five years.
Cohen has been sounding alarms about this dynamic since at least late 2025, when he first flagged what he described as a near-100% probability of an AI-related credit bubble. His concern isn’t that AI itself is overhyped, necessarily. It’s that the flood of debt issuance is creating structural risks that most investors aren’t pricing correctly.
Cohen has drawn parallels to historical cycles including railroads and the early internet as templates for what might unfold in AI credit markets.
Why Alphabet and Amazon stand out
Not all tech borrowers are created equal. Alphabet and Amazon sit at the top of the corporate credit pyramid, with massive cash flows, diversified revenue streams, and investment-grade ratings that give their bonds structural protections most AI-adjacent issuers can’t match. Cohen’s thesis is essentially that the market is painting with too broad a brush, lumping Alphabet’s bonds in with speculative issuers in ways that miss an important distinction in balance sheet quality.
The broader credit landscape
Cohen sees the overall investment-grade corporate credit market as fundamentally healthy. An upgrade cycle is underway, meaning more companies are seeing their credit ratings improved than downgraded. The tension is sector-specific: over $155 billion in new bond supply from hyperscalers alone creates technical pressure on spreads, and the sheer volume of projected capital expenditure means the borrowing isn’t likely to slow down anytime soon.
What to watch from here
Cohen’s framework suggests a few things worth monitoring. First, the gap between top-tier tech borrowers and lower-quality AI issuers. Second, the pace of new issuance — a 45% year-over-year increase is already aggressive, and if hyperscalers accelerate borrowing further to keep up with AI infrastructure demands, the technical pressure on the bond market could intensify. Third, the $5 trillion in projected AI capex over five years assumes sustained corporate commitment to AI spending. Any sign that companies are pulling back could reshape the entire credit landscape for the sector.