JPMorgan Asset Management: Gross Calls AI Buildout a Boom as Credit Market Gets Nervous

Friday, October 9, 2026. Bloomberg publishes an interview with Jared Gross, head of institutional portfolio strategy at JPMorgan Asset Management. His thesis: spending on AI infrastructure remains a boom, and he does not consider it a bubble. He says this at a moment when credit markets are nervous again about the prospect of new debt megadeals. The market is already saturated with loans that pay for AI projects. The short teaser for the interview gives no specific figures: no issuance volumes, no spreads, no borrower names. That makes the logic of the argument more important than any individual number.

Where the concern has shifted

Early bubble talk in AI centered on equity valuations: how much a chipmaker is worth, how highly a model developer was valued. Now attention is moving to the debt market. Data centers, power, sites and hardware require capital on a scale that company cash flow does not always cover. Part of that capital comes through bonds, credit lines and structured finance.

That is why Bloomberg's wording that credit markets are "nervous again" matters on its merits. A shareholder can live through a drop in share price. A lender expects to be repaid on schedule and reacts to risk differently. If the market starts demanding a bigger premium for each new giant deal, all subsequent construction becomes more expensive.

The "boom, not a bubble" argument

Gross handles asset allocation for institutional investors, so he looks at AI as a segment of a portfolio. The teaser makes one thing clear: he distinguishes between a boom and a bubble. A boom implies real demand and real investment in production capacity, even if the market is overheated. A bubble means prices detached from any reasonable revenue expectations.

The details of his argument are not disclosed in the available material. Positions like this usually rest on a few points: the capital comes from companies with large cash flows, demand for compute still exceeds supply, and the borrowers are mostly creditworthy. Those assumptions will have to be checked against the interview itself and market data.

What to keep in mind when reading any statement of this kind:

• Source of the optimism. Managers at large banks work with client capital and rarely declare a bubble in public. That does not invalidate their view, but it sets the context.
• Debt structure. It matters who the borrower is: a company with steady profits or a project backed by future contracts.
• Concentration. The more deals depend on the same compute customers, the higher the systemic risk.
• Cost of money. Wider spreads on AI issuance would be the first visible sign of cooling.
• Payback period. AI hardware ages faster than typical debt lasts.

A practical view

For investors, the takeaway is simple: it is more useful to break the "bubble or not" question into questions about specific instruments. Chipmaker equities, big tech bonds and project finance for data centers behave differently under stress. A single blanket bet on "AI" hides those differences.

For startups and infrastructure companies, something else matters more. If the credit market becomes more selective, expensive projects without long-term compute contracts will be the most vulnerable. Contracts with a clear customer, a transparent power supply plan and moderate leverage become an advantage when raising money.

For readers following the industry, the best indicator is not what fund managers say but how the market behaves: the size of new offerings, their terms and how quickly they are absorbed. If the next megadeal closes without a noticeable premium, the "boom" view gets confirmation. If not, Gross's thesis will have to be tested again. The full interview is worth checking on Bloomberg: the teaser states his position but not the reasoning behind it.