JPMorgan: New Cloud Providers' Financing Pressure Won't Shake AI Infrastructure Cycle; Tech Sector Still Has $400 Billion Surplus This Year

Concerns have recently intensified in the market about the sustainability of the financing model of neoclouds, but JPMorgan said in its latest report that these risks are not enough to shake the overall AI capital expenditure cycle. The key reason is that globally listed technology companies are expected to generate about $2.2 trillion in operating cash flow this year, with capital expenditure and R&D spending totaling about $1.8 trillion; after offsetting the two, the technology sector as a whole still has a financing surplus of about $400 billion—current investment levels can be maintained without relying on external financing.

The report also noted that even if the financing model of neoclouds runs into trouble under the worst-case scenario, their construction gap can be fully filled by technology companies with more robust balance sheets, such as hyperscalers. At the same time, JPMorgan's overall equity positioning indicator shows that market positioning has fallen significantly from its recent peak in mid-September, indicating that short-term downside room is narrowing.

Financing concerns about neoclouds—specialized AI infrastructure providers that rent out computing power specifically for AI workloads—have continued to build over the past week. The core of the problem is the timing mismatch between their capital expenditure and revenue, and rising financing costs in the high-yield bond market are squeezing the room for this business model to operate.

According to JPMorgan's estimates, of the roughly $2.2 trillion in expected operating cash flow of globally listed technology companies this year, neoclouds contribute only about $30 billion; of the roughly $1.8 trillion in capital expenditure and R&D spending, the corresponding scale for neoclouds is about $80 billion. JPMorgan's credit research team noted that as financing costs in the high-yield bond market rise, publicly listed neoclouds may need to adjust their financing structure, reducing debt financing at the company level and relying more on equity financing and project-specific debt.

JPMorgan's equity research team also noted that higher funding costs will indeed compress return space, but strong demand for computing power also gives providers the possibility of passing some costs on to customers. Even if this logic ultimately proves untenable, the construction scale of neoclouds can also be taken up by technology companies with stronger balance sheets, such as hyperscalers.

$400 billion internal surplus forms a safety cushion for the AI cycle

Report data show that this year the difference between operating cash flow and capital expenditure in the technology sector is a surplus of about $400 billion. In JPMorgan's stress scenario calculation, even if the technology sector raises capital expenditure by 25% each year for the next five years (through 2030) while cash flow grows 20% annually, it will still maintain a small surplus by 2030. This means that over a considerable time window, the technology sector as a whole can support AI infrastructure investment without relying on external capital market funding.

So why are technology companies still actively issuing debt?

JPMorgan explained that while increasing capital expenditure, the technology sector is still maintaining large-scale share buyback plans, and this is the main source of external financing demand. Once market financing pressure spreads from neoclouds to technology companies more broadly, these companies can in principle choose to suspend buybacks and activate an internal circular financing mechanism—that is, companies such as chipmakers with cash flow surpluses provide internal support to neoclouds with funding gaps.

JPMorgan's previous research showed that the deep support for the AI capital expenditure and cash flow cycle comes from direct AI procurement demand from large enterprises; survey estimates suggest that global related spending this year could reach $1.5 trillion.

Semiconductor VaR shock frequency rises, positioning adjustment signals appear

JPMorgan interprets the recent pullback in AI infrastructure stocks against the structural backdrop of increasingly frequent VaR shocks in the semiconductor sector.

Since the second quarter of this year, even with solid AI cycle fundamentals, semiconductor VaR shocks will become more frequent. The reason is that semiconductors (including memory chips) have seen a sharp rise in their share of global stock market capitalization, and their volatility premium relative to the broader market has also risen.

Measured from the perspective of risk capital occupation, JPMorgan estimates that the technology sector currently consumes about 40% of the risk capital in the MSCI ACWI index, far exceeding its 28% market capitalization weight; among this, the semiconductor industry occupies about 21% of risk capital, also higher than its 13% market capitalization weight—meaning semiconductor-related companies actually consume about one-fifth of total risk capital in the global stock market.

This feature has already been directly reflected in hedge fund behavior.

JPMorgan data show that since late September, the performance of the semiconductor ETF (SMH US) has been highly synchronized with the daily performance of the HFRX Equity Hedge Index, much like the situation from early April to late July this year. During market rebounds, this linkage reflects hedge funds' high concentration in semiconductors; during pullbacks, it means forced reductions after risk limits are triggered amplify the decline.

Currently, JPMorgan's comprehensive equity positioning indicator shows that positioning has fallen sharply from its recent peak in mid-September, and the semiconductor long positions of trend-following investors, including CTAs, appear to have been largely digested. The bank believes that the progress of this positioning adjustment is gradually narrowing the market's short-term downside room. Risk disclosure and disclaimer

Markets involve risk, and investment requires caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their particular circumstances. Investment based on this is at the user's own risk.