AI investment is adding another layer of pressure to US borrowing costs, as technology companies raise more debt to finance infrastructure, and expectations of future productivity gains influence long-term interest rates.
ING, in its October 1 report, estimates that artificial intelligence accounts for about 20% of the forces influencing long-dated rates, with productivity expectations representing the largest share of its estimated impact.
The bank’s analysis comes as US corporate bond issuance has crossed $1 trillion this year, led by technology, media and telecommunications companies.
The surge in investment is also creating a challenge for the Federal Reserve.
In an October 9 note, Apollo Global Management chief economist Torsten Slok argued that AI investment is relatively insensitive to interest rates.
Tighter monetary policy could therefore weigh on housing and autos while AI-related capital spending continues, complicating the Federal Reserve’s efforts to cool economic activity.
AI investment adds to pressure on long-term yields
In the October 1 note, ING’s global markets research team identified three channels through which AI influences long-term interest rates: direct spending, corporate borrowing and productivity expectations.
The bank estimated that productivity accounts for about 70% of AI’s overall influence on long-dated rates, while corporate issuance represents 25% and direct spending accounts for the remaining 5%.
The figures are ING’s estimates of AI’s contribution to interest-rate pressures, not a precise measurement of how much the technology has lifted Treasury yields.
The productivity channel reflects expectations that AI could improve the productive capacity of capital and labour.
ING said higher productivity could increase expected returns on investment, encourage additional capital spending and put upward pressure on real yields.
The bank also drew a comparison with the technology boom of the late 1990s and early 2000s.
It noted that the 10-year real yield reached about 4% during that period, alongside a nominal 10-year Treasury yield above 6%.
ING put the current 10-year real yield at 2.9% and the nominal yield at 5.2%.
However, the bank stressed that AI is not the only factor influencing long-term borrowing costs.
It considers inflation and fiscal deficits more important overall drivers of the rise in long-dated rates.
Corporate borrowing adds to the financing pressure
The growth in corporate debt issuance provides a more direct measure of the financing required to support investment.
US dollar corporate bond issuance reached $1.003 trillion year-to-date through September, exceeding the full-year 2025 total of $925 billion, according to an ING report published on October 6.
| Year | FY issuance ($bn) | YTD issuance ($bn) |
|---|---|---|
| 2020 | 1,166 | 1,038 |
| 2021 | 729 | 589 |
| 2022 | 564 | 452 |
| 2023 | 650 | 552 |
| 2024 | 811 | 721 |
| 2025 | 925 | 697 |
| 2026 | 1,003* | 1,003 |
September issuance totalled $112 billion, while net corporate supply reached $573 billion year-to-date.
Technology, media and telecommunications companies accounted for $381 billion, or about 38%, of dollar-denominated corporate issuance through September. Issuance in the sector rose 158% from the same period a year earlier.
The figure does not represent AI-related borrowing alone.
ING’s TMT category covers a broad range of businesses, and its data do not identify how much of the debt was raised specifically to finance AI investment.
The sector issued $44 billion in September alone.
ING’s figures cover investment-grade corporate issuers from America and Europe.
Nevertheless, the increase comes as technology companies commit substantial sums to data centres, computing infrastructure and other AI-related investments.
ING said technology companies have historically funded much of their investment through operating cash flows.
But rising capital expenditure is weighing on corporate savings, while the bank expects issuance volumes to remain elevated as annual investment approaches $1 trillion.
In its October 1 analysis, ING also highlighted a shift towards longer-dated corporate borrowing.
ING estimated that long-dated investment-grade corporate issuance was running $55 billion above the previous year, while Treasury issuance in the 10- to 30-year maturity range was unchanged at $340 billion.
Treasury buybacks were about $20 billion higher year over year, reducing the amount of outstanding government debt held by investors.
Combining the changes, ING calculated a $75 billion relative shift in long-dated supply pressure towards corporate bonds and away from Treasuries.
That is equivalent to about 22% of the $340 billion in annual issuance across the comparable Treasury maturity range.
ING said the shift was not destabilising but was a factor to consider when assessing long-term yields.
It also noted that investors had continued to absorb corporate issuance, with new-issue premiums for longer-dated and AI-related bonds remaining around 10 basis points.
AI spending could complicate the Fed’s task
The financing boom comes as the Federal Reserve faces persistent inflation and elevated borrowing costs.
US Treasury yields climbed to multi-decade highs this week.
Treasury yields reached their highest levels since 2002 earlier in the week, with the 10-year yield touching 5.36% and the 30-year yield rising above 5.7%, before retreating from their peaks.
Minutes from the Fed’s September meeting, released this week, showed that most officials anticipated another rate increase later this year as inflation remained above the central bank’s 2% target.
Apollo’s Slok argued on October 9 that AI investment is relatively insensitive to interest rates, creating a problem for monetary policy.
In his view, the investment boom is drawing capital, electricity and labour away from more rate-sensitive sectors, including housing and autos.
This leaves the Fed facing inflationary pressure from AI-related activity while higher rates weigh on other parts of the economy.
Slok said monetary policy cannot resolve this imbalance and pointed to expanding the supply of electricity, chips and infrastructure as the more durable solution.
The argument adds a further complication to the outlook for interest rates.
If AI investment remains strong despite restrictive monetary policy, higher rates may have less influence on that spending than on other sectors.
Still, the research does not establish that AI borrowing is the principal cause of the Treasury sell-off.
ING explicitly identifies inflation and fiscal deficits as more influential forces behind long-term yields.
The key question is how much additional pressure the AI investment cycle will place on financing markets as corporate spending continues to expand.
The post Why the AI trade is becoming a problem for the bond market appeared first on Invezz
