Written by: Rita

As of 2026, the issuance of high-rated bonds in USD related to AI capital expenditure has reached $266 billion. Among these, ultra-large companies issued $182 billion, data centers issued $42 billion, and semiconductor companies issued $42 billion. This figure exceeds the total of $139 billion for the entire year of 2025 and is more than nine times the $29 billion for the whole of 2024. JPMorgan noted in its strategic report published on September 9, 2026, that the financing demand for AI capital expenditure is changing the structure of the U.S. credit market and the way macroeconomic risks are transmitted.

JPMorgan believes that AI investment is both a market risk and an economic risk. The sensitivity of household wealth and consumption to asset prices is increasing. Report author Joyce Chang wrote that history shows investment booms seldom end gently.

This article is developed along three lines. The scale and sources of financing for AI capital expenditure, how fiscal deficits and net interest costs push up term premiums, and how the rejection of a winner-takes-all narrative by middle forces affects global capital allocation.

AI Capital Expenditure Financing Exceeds Last Year

JPMorgan's U.S. high-rated strategy team predicts that over the next five years, the financing demand for AI capital expenditure in the high-rated bond market is expected to reach $2.1 trillion. Among the issuance composition from 2026 to date, ultra-large companies issued $182 billion, data centers issued $42 billion, and semiconductor companies issued $42 billion. For the entire year of 2025, ultra-large companies issued $93 billion, and only $17 billion in 2024.

The sources of financing for capital expenditure are shifting from internal cash flow to external debt. JPMorgan estimates that the total funding need for AI capital expenditure is $5.5 trillion, of which organic cash flow contributes $1 trillion, incremental equity capital contributes $400 billion, the structured products market contributes $300 billion, the high-rated bond market contributes $2.1 trillion, leveraged finance market contributes $400 billion, and there is another $1.4 trillion that needs alternative capital. The high-rated bond market is the largest single external financing source.

This change in financing structure indicates that the capital cost of AI investment is directly linked to credit market conditions. The report points out that long-term debt issuance by ultra-large companies has significantly increased, crowding out other investments. The surge in capital expenditure pushes up real interest rates, and momentum trading and leverage may amplify withdrawals and correlations, undermining the effectiveness of old adages like “do not fight the Fed.”

Net Interest Costs to Double by 2035

The U.S. fiscal deficit for FY 2026 is expected to reach $2.1 trillion, which is about 6.5% of GDP. The deficit from 2026 to date exceeds last year's amount by about $170 billion. JPMorgan points out that long-term yields will not be anchored until actions are taken on the fiscal side. There are currently no measures to cut spending or increase revenues before the end of the year. Tariffs will continue to be the government’s only new source of fiscal revenue.

JPMorgan’s long-term rate model assumes a 2.5% real yield on U.S. Treasury bonds. If expected inflation remains at 3%, the long-term target nominal yield for the 10-year U.S. Treasury bond is approximately 5.5%. When interest rates reach 5.5% of GDP, the U.S. net interest costs are expected to double from the current approximately $1 trillion to $2.1 trillion by 2035, nearing 5% of GDP.

Fiscal risks are transforming into financial risks. The report points out that interest rate levels are influenced by deficits, issuance, and market structure, as significantly as they are affected by the Fed's path. Changes in the structure of U.S. Treasury bond holders and a decline in foreign exchange reserves are now fundamental drivers of rising term premiums. Financial repression measures, such as policies aimed at keeping the yield curve flatter, have re-entered discussions.

AI Investment Amplifies Dual Risks

JPMorgan defines AI investment as a macro shock. Global capital expenditure on data centers and IT infrastructure is expected to reach $6.7 trillion by 2030. The U.S. is expected to build 200GW of data centers between 2026 and 2032, possibly corresponding to an investment of $8.2 trillion, roughly 2.8% of cumulative GDP. AI capital expenditure is projected to account for 5% of GDP.

Economic risks and market risks are becoming one and the same. The sensitivity of U.S. household wealth and consumption to asset prices is increasing, with the total market capitalization of the U.S. market now reaching 248% of GDP. The report states that given the level of global financialization, this sensitivity allows market withdrawals to directly translate into economic downside pressure.

The report mentions that history shows investment booms rarely end gently. This judgment points to tail risks in the AI capital expenditure cycle. As the financing structure shifts from internal cash flow to high-rated bonds and leveraged financing markets, the sustainability of capital expenditure plans will face tests if credit conditions tighten.

Middle Forces Reject Winner-Takes-All Narrative

JPMorgan points out in the report that the U.S. tends to define the AI race and U.S.-China strategic competition through a single-winner framework, which is not recognized by middle forces. Many middle-force countries are optimizing autonomy, diversification, and bargaining leverage, seeking a balance between trade, security, technology, and capital flow.

The report also acknowledges that the responses from middle forces have been disappointing. As the definition of middle forces expands, networks become more diverse, and common interests diminish. Group formations become more chaotic, risks of fragmentation rise, and the predictability of cross-border policy coordination decreases during times of pressure. The report believes Europe has significant global influence but does not always fully exercise it.

The issue of global imbalances is the core divergence that prevented the G20 finance ministers and central bank governors from reaching a joint communique. The U.S. believes that persistent current account surpluses, excessive reliance on exports, and non-market practices are distorting the global economy. China rejects this description, emphasizing that it does not deliberately pursue trade surpluses and is working to boost domestic demand, supporting a balanced multilateral approach to resolve global imbalances.

Governance Framework Determines Payment Winners

Stablecoins, tokenized deposits, and central bank digital currencies may reshape settlement and cross-border finance, but JPMorgan points out that these technologies will not eliminate hard constraints such as sanctions, illicit finance, capital controls, monetary sovereignty, and systemic risk transmission. The core argument of the report is institutional. Technology expands the feasible set, but governance determines the outcomes.

Financial stability depends on leverage, term transformation, and interconnectedness. The report states that the issue lies in who bears the losses and where hidden vulnerabilities accumulate; the speed of settlement itself is not critical. This judgment shifts the analysis focus of payment system innovations from technical capability to institutional design.

The report defines the current period as a transitional phase. The world is in a transition between established orders and has not yet stabilized into a new international order. Each new system may trigger further transitions and will not stop at a stable endpoint. The U.S.-led unipolar system once facilitated the rise of emerging markets and China, which now threatens the previous equilibrium. Within this framework, the winner-takes-all narrative does not align with the preferences of middle forces, nor with the realities of the transition period.

Disclaimer

This article is a整理 and interpretation of a third-party brokerage research report (JPMorgan, September 9, 2026) by Chao Xiang Research, combined with整理 of public market information. The ratings, target prices, profit forecasts, and related judgments cited in the text are the views of the analysts from that brokerage and only represent the position of their institution, not the views of Chao Xiang Research, and do not constitute any investment advice.

The market has risks, and decisions must be independent. This article should not be used as a basis for buying or selling any securities.