Long-term US Treasury Sell-off Continues! 10-Year Treasury Yield Breaks 5.2% Again, “AI Boom vs. Rising Financing Costs” Narrative Showdown Intensifies

According to Zhitong Finance APP, the wave of AI computing infrastructure investment is facing increasingly sharp contradictions: the strong expectations for computing power driven by the rapid expansion of AI agent applications reinforce the necessity for enterprises to lock in computing resources on a large scale in advance. However, energy inflation and the Federal Reserve’s tightening monetary policy expectations are simultaneously raising the funding costs of these investments. In early Asian trading on September 28, after Donald Trump rejected Iran's latest proposal to reopen the Strait of Hormuz, Brent crude rose 1.4% to $105.80 per barrel, and U.S. Treasury bonds suffered another round of heavy selling: the yield on the two-year Treasury, which is sensitive to policy rates, rose 5 basis points to 4.90%, while the ten-year Treasury yield rose 4 basis points, once again exceeding the 5.20% mark for the second time since last Thursday.

In early Asian hours, Japanese and Australian government bonds also continued to come under significant selling pressure, following the trend in U.S. Treasuries. For companies involved in computing infrastructure—such as Oracle, Amazon, Google, and CoreWeave—expanding AI data centers with borrowed funds, this means new financing will require paying higher interest rates and financing costs. For equity investors, it means future profits must continually withstand evaluation under higher discount rates and sustained higher financing costs. Whether AI computing power demand can continue to grow, and how much of this growth will ultimately return cash to shareholders, are now two separate questions that need distinct answers.

The immediate catalyst for the global bond market adjustment is the market's reassessment of the timeline for energy supply recovery and how much more the Federal Reserve needs to hike rates. Moreover, for the U.S. Treasury yield curve beyond ten years, a more structural and critical force is the "rising fiscal deficits + AI bond issuance" competing for the pool of global long-duration bond funds.

To understand the ten-year U.S. Treasury as the "anchor of global asset pricing," one must distinguish between the expected path of future short-term rates and the term premium required for holding long-term bonds. On September 16, the Federal Reserve raised the policy rate by 25 basis points to 3.75%–4.00%. Expectations of further hikes under an inflationary regime will transmit through the yield curve to long-term bonds. Besides the AI bond issuance frenzy and U.S. Treasury expansion jointly competing for global funds, fiscal financing, inflation uncertainty, and the long-term bond supply-demand dynamic are all factors affecting the extra compensation investors require.

In Japan, financial markets are also losing the cushion previously provided by a low interest rate environment. Recent quotes show that Japan’s ten-year government bond yield reached 3.115% on September 25, the highest since 1996. The yields for 20-year and 30-year bonds on September 24 rose to 3.900% and 4.130%, respectively. On September 18, the Bank of Japan decided to raise its policy rate from around 1.00% to about 1.25%, and explicitly noted that high oil prices, yen depreciation, and AI-related demand are pushing up corporate costs, with some of the pressure beginning to pass on to consumer prices as wage pass-through and inflation expectations strengthen.

Oil Prices Surge After Trump Rejects Iran Proposal, Bond Selloff Reappears

On Monday, U.S. Treasuries underwent another round of sell-off. Previously, U.S. President Donald Trump rejected Iran’s latest proposal to reopen the Strait of Hormuz, driving up international oil prices. Brent crude price surged more than 2% in early trading, approaching $107 a barrel, intensifying market concerns about global inflation.

The two-year U.S. Treasury yield—sensitive to rate changes—rose by 5 basis points to 4.90%, while the ten-year yield rose by 4 basis points to 5.20%. Sovereign bond prices in Japan and Australia also fell. Last week, in the context of hawkish comments from Federal Reserve officials, U.S. Treasury yields across maturities soared to multi-year highs. Monday's Treasury selloff occurred on the back of those movements.

As shown in the chart above, the ten-year U.S. Treasury yield is hovering near its highest level since 2007.

Early in the Asian session, Brent crude rose 1.4% to $105.80 per barrel. Previously, Iran stated it would not relax its conditions for reopening the Strait of Hormuz, once again increasing the Fed's pressure to curb inflation through rate hikes. According to Axios, Trump said that despite rejecting Tehran’s latest proposal, talks are expected to resume later this week.

Damien McCullough, Head of Fixed Income Research at Westpac, said, “The Fed’s persistently hawkish signals, as well as oil prices remaining above $100, are key factors behind the weakness in the bond market.”

Meanwhile, U.S. Treasury Secretary Scott Besant urged Fed policymakers to remain “open-minded” on interest rates, arguing that productivity gains driven by artificial intelligence and regulatory easing would help control U.S. inflation.

The Yield Curve “Bite Back” of the AI Boom: The More Robust the Demand for AI Computing Power, the More Meticulous the Funding Calculations Need to Be

AI financing expansion has already extended from investment-grade tech giants to record high-yield bond transactions. On September 24, SoftBank finalized issuance terms for $10 billion and €1 billion (about $11.1 billion) in new bonds, widely recognized by the media as the largest high-yield corporate bond deal on record globally; according to SoftBank’s announcement, the scheduled issue date is September 29. The 7.5-year USD bonds carry a coupon of 9.75%, and funds raised include payment for the final $1 billion installment in additional investment to OpenAI. This deal shows that while AI financing channels are still open, the price for raising large sums has become quite expensive.

Large tech firms are likewise scaling up financing and extending debt maturities. Alphabet, Google’s parent company, completed a $25 billion USD bond issuance on August 10, with the longest maturity due in 2066. On September 9, Amazon raised £4.25 billion (about $5.76 billion) in its first British pound bond deal. More noteworthy is the supply expectation: Goldman Sachs’ forecasts show that debt issuance by hyperscale cloud providers will hit a record $420 billion by 2027, 60% higher than the estimated total for 2026.

Financing pressure is coming from both benchmark rates and credit spreads. As of September 22, market data showed that spreads for AI-related issuers were about 115 basis points, higher than the broad investment-grade market's 78 basis points. Interviewed institutions stress that the extra compensation is due not only to credit risk but also to the continuously increasing supply of bonds, the uncertainty of issuance pace, and portfolio concentration limits; these spreads are not directly equivalent to default probability. For new USD fixed-rate financings, costs are typically comprised of the base rate for the corresponding maturity (often closely linked to the ten-year U.S. Treasury yield known as the “anchor of global asset pricing”) plus credit and liquidity premiums. When both components rise at the same time, even if companies are still able to finance smoothly, project returns may be compressed.

This contradiction can to some extent be summarized as the “yield curve bite back” of AI prosperity for long-duration U.S. Treasuries: to realize future productivity gains, companies first expand demand for chips, electricity, construction, and funding; if near-term demand grows faster than supply expansion, costs and inflation pressures may increase, in turn tightening the financing environment. Besant emphasizes that AI and regulatory easing can enhance productivity and help control inflation—logic focused on improvements in long-term supply capacity. Researchers at the Federal Reserve Bank of St. Louis, however, note through model analysis that even before productivity gains materialize, optimism about future growth can first stimulate current demand and inflation. These assessments concern different time frames: the potential for long-term cost reductions does not automatically offset the financing cost pressures and resource constraints during the build-out phase. From an AI infrastructure project financial perspective, what truly requires attention is how much space can remain between the investment return rate and capital costs, and whether cash flows can keep up with the debt repayment schedule.