JPMorgan's most optimistic path for US federal debt now runs straight through AI. In a May 28 analysis, JPMorgan Asset Management chief global strategist David Kelly mapped five scenarios for the debt trajectory through 2036 — and the only one that keeps debt from blowing past 130% of GDP assumes an AI-driven productivity surge that exceeds what even the bulls currently price in.
The numbers
Federal debt held by the public is on track to hit $32.2 trillion, or 100.4% of GDP, by the end of this fiscal year — up from just 31% of GDP in 2001. The FY2026 deficit is running near $1.89 trillion, the gap between roughly $7.4 trillion in spending and $5.5 trillion in revenue. Annual interest payments have already crossed $1 trillion. Total gross national debt sits near $39 trillion, a figure JPMorgan CEO Jamie Dimon has flagged repeatedly.
Five paths, one swing factor
Kelly's baseline scenario has debt reaching about 127.7% of GDP by 2036. The best case caps it at 115%, but only if three things break right simultaneously: AI lifts productivity beyond expectations, immigration restrictions ease enough to grow the labor force, and a prolonged stretch of divided government prevents either party from adding fresh unfunded stimulus. He frames that outcome as a slow deterioration in federal finances that the bond market largely shrugs off. The downside is blunter — Kelly writes that "a fiscal crisis scenario is somewhat more likely" than a serious, sustained deficit-reduction effort. Two further scenarios, large spending cuts or tax increases, he treats as politically improbable under current conditions.
Why AI is load-bearing
Productivity is the lever markets are betting hardest on, and it is the one most directly tied to the AI capital cycle. The IMF has cautioned that AI cuts both ways for public finances: it can expand output and the tax base, or it can concentrate wealth and hollow out labor-tax revenue if gains accrue narrowly to capital. That ambiguity is the difference between Kelly's 115% and his fiscal-crisis tail.
What it means for enterprise and infra
The practitioner takeaway is that the fiscal best case and the AI capex thesis are now the same bet. Hyperscalers are committing close to $700 billion this year on the premise that AI throughput translates into measurable enterprise productivity. Kelly's framework makes the macro stakes explicit: if those gains show up in output and earnings, the optimistic debt path stays plausible and capital keeps flowing. If they stall, the more likely scenario is rising rates, tighter financing, and pressure on the very deals funding the buildout. For enterprise buyers, that raises the bar on demonstrating realized productivity — not pilots — from AI spend.
