The AI Supercycle Meets the Full Equation

“The decisive metric is no longer GPUs, tokens or CapEx — it is productive intelligence per unit of capital.”

This is one of those days when three apparently separate market stories converge into a single economic equation. Data cut: 07:45 CEST.

Later today, Nvidia reports results that could determine whether the extraordinary expansion of AI infrastructure continues to validate the capital being committed underneath it. Before that, at 14:30 CEST, the United States publishes July PCE inflation. And overnight, the third component of the equation moved sharply in the opposite direction, as oil fell on renewed Iran–Oman negotiations over the Strait of Hormuz.

Markets are therefore simultaneously repricing the return on artificial intelligence, the cost of capital, and the cost of the physical system on which artificial intelligence depends. That is the complete equation.

1. 01 Nvidia Has Become the Earnings Report for the AI Economy

   Nvidia reports today with revenue expected near $92.2 billion and data-center sales set to more than double, while investors scrutinize the transition from Blackwell to Vera Rubin. Tuesday's positioning was constructive but cautious: Nvidia gained 2.2%, the Philadelphia Semiconductor Index 1.4%, the Nasdaq 0.66% and the S&P 500 0.32%, though Nasdaq futures were about 0.5% lower early this morning.

   The market already assumes Nvidia will beat expectations. Execution is therefore no longer the upside catalyst — it is the minimum requirement. The relevant question is what happens after Rubin, and today's result should be read through one new metric: economically useful intelligence produced per dollar of AI infrastructure. signal: Nvidia's earnings are now a macro reading of the AI capital cycle, not a sector event.

2. 02 Hormuz Relief Has Become Duration Relief

   Iran and Oman have resumed negotiations over the Strait of Hormuz, including a temporary joint navigational corridor and mine clearing. Brent fell more than 3% Tuesday and another 2.6% to around $86.28, while U.S. 10-year yields fell roughly 6.5 basis points Tuesday to near 4.63%. But only five commodity ships crossed Tuesday, against a ten-day average of 15.

   The recursive chain is now visible: Hormuz risk ↓ → oil ↓ → inflation premium ↓ → Treasury yields ↓ → WACC ↓ → technology-duration valuations ↑. Lower oil acts indirectly as AI financial easing. But five vessels are not normalization — the market is buying the possibility of normalization before the infrastructure has actually normalized. signal: Oil and Nvidia increasingly belong inside the same financial model.

3. 03 PCE Will Test Whether the Denominator Can Really Fall

   July core PCE is expected at 3.3% year-on-year and +0.2% month-on-month, with Fed policy at 3.50%–3.75%. Boston Fed President Susan Collins warned that further tightening may soon be appropriate absent sustained progress. Meanwhile consumer confidence fell to 89.4 in August and July new-home sales dropped 10.5% to a 607,000 annual rate.

   The consumer is weakening before inflation is defeated. The Great Split is becoming visible inside the U.S. economy: AI investment velocity rising sharply while household purchasing velocity declines. The Fed increasingly faces two economies — one capital-rich and accelerating, the other consumption-dependent and rate-sensitive. signal: If core PCE holds near 3.3%, the Fed cannot cut for the consumer without raising the hurdle rate on capital.

4. 04 Europe Is Beginning to Grow Again — Just as the ECB Prepares to Tighten

   Germany's Q2 GDP was revised up to +0.3% quarter-on-quarter, following 0.4% in Q1, with exports up 2% and the Ifo index at 88.8, its strongest in a year. Yet Reuters reports ECB policymakers preparing for a September hike, probably taking the deposit rate from 2.25% to 2.50%.

   For years the problem was insufficient growth; now it may become how to finance a recovery without extinguishing it. Europe must simultaneously fund defense, energy security, grids, rail, accelerated computing, AI factories and demographic obligations — while entering that cycle with materially higher sovereign yields. signal: Can productivity growth rise faster than the cost of capital being mobilized to produce it — the same question Nvidia faces, at continental scale.

5. 05 AI Has Reached the Sovereign Balance Sheet

   Morgan Stanley estimates $350–400 billion of AI-related public investment-grade debt could be issued in 2026, with $1.4 trillion of AI CapEx planned for 2027 and some $92 billion financed since mid-year. At the same time, U.S. corporate tax receipts are down roughly 23% — partly because 2025 legislation allows immediate expensing of large capital investments — while deficits run near 6% of GDP and interest on ~$40 trillion of debt exceeds $1 trillion annually.

   The chain becomes: AI CapEx ↑ → corporate debt issuance ↑, corporate tax receipts ↓ → government funding pressure ↑ → competition for duration ↑ → long yields ↑ → AI WACC ↑. Artificial intelligence can raise future productivity while making the short-term sovereign financing problem harder. signal: The AI Supercycle now has a fiscal dimension markets have barely begun to price.

Key takeaways

the recursive system

• Nvidia tests the numerator, PCE tests the denominator, Hormuz tests the physical cost base — three variables aligning in one trading day.
• Lower oil acts as indirect AI financial easing, but five Hormuz vessels reflect diplomacy priced ahead of physical supply.
• The U.S. now runs two economies: capital-rich AI investment accelerating while household consumption weakens before inflation is beaten.
• AI is reaching the sovereign balance sheet — raising corporate debt issuance while cutting tax receipts and intensifying competition for duration.
• The Great Split will separate firms and nations by whether intelligence generates a return greater than the full cost of producing it.