AI is causing the collapse of US Treasuries, and even if it ultimately succeeds, it will face heavy taxes

**Key points**

**1. Tech giants issue debt at any cost, crowding out long-term capital**

As of August, the five major cloud providers have issued nearly $230 billion in debt this year, more than twice last year's total, with a large share in 20- to 40-year maturities; Meta, Alphabet and others are locking in long-term money regardless of rates and are insensitive to price, because they firmly believe AI returns will far exceed interest costs.

**2. The reverse crowding-out effect pushes up US Treasury yields**

Yields on long-dated tech bonds (such as Alphabet's 30-year at 6.4%) attract scarce long-term capital, forcing Treasuries to raise yields to compete for buyers; Nomura estimates tech giants have borrowed about $200 billion of long-term funds, a quarter of the Treasury's medium- and long-term issuance.

**3. The AI boom actually erodes the government's tax base**

More than 75% of US federal revenue comes from individual income and payroll taxes; AI compresses labor income and employment, reducing total wages and the corresponding tax revenue; the "One Big Beautiful Bill Act" lets tech giants take large depreciation-based tax breaks, and corporate income tax revenue plunged by a quarter year-on-year in the first 11 months of this fiscal year.

**4. The fiscal gap forces AI profits to face heavy taxes**

Shrinking tax revenue combined with a ballooning deficit (over $2 trillion in the first 11 months of this fiscal year) means government spending is expanding instead, due to retraining and other needs; the only growing tax source is AI profits, and the more successful they are, the more likely they are to be heavily taxed to fill the fiscal gap. 2026-10-03 20:52 AI is crashing US Treasuries, and even if it ultimately succeeds it will be heavily taxed Wall Street Insights at a Glance

This article is from the WeChat public account: Wall Street Insights, author: Yue Ming

On September 24, US long bonds collapsed across the board. The 30-year Treasury yield rose to 5.48%, a 20-year high; the 10-year broke through 5.2%, a high since the financial crisis. Long-end yields have risen almost in a straight line over the past two months, and Treasury Secretary Bessent twice stepped up buyback interventions, each time buying only a day of respite before the market pushed them back the next day.

Deficits too large, oil prices too high, the Fed raising rates again—these are all real underlying forces, but none is the new variable in this round. What truly makes this round different from previous ones is that a competitor that does not ask about price at all has emerged on the demand side.

To rush the construction of data centers and computing power, tech giants are issuing ultra-long bonds at a record pace, and almost without regard to price.

As of August, the five major hyperscale cloud providers have issued nearly $230 billion in debt this year, more than twice last year's total, with a large share concentrated in 20-, 30- and even 40-year maturities.

These deals would be headlines in any year.

Meta is only 22 years old, yet last October it issued a bond locked in for 40 years—by the time it matures, Zuckerberg will be 80. That single deal was $30 billion, setting a record for a single corporate bond issuance. In April this year, Meta added another $25 billion.

Alphabet even issued a 100-year sterling bond. Amazon raised $54 billion in a single deal in March. Oracle is now rated only two notches above junk, has negative free cash flow, yet still issues bonds one after another.

Why don't they care about rates? Bessent himself remarked in an August interview that these companies don't look at rates at all when borrowing, because they firmly believe AI returns will be so high that current interest costs will be negligible.

And there is only so much money willing to lock in "long money": pension funds, insurers, long-duration bond funds—the long end is supported by these buyers, and the money they have is limited. Money that flows into tech companies' long bonds cannot flow into Treasuries.

Alphabet's 30-year corporate bond yield of 6.4% is right there, and if Treasuries want to win the money back, they can only push their own yields up.

Nomura estimates that tech giants alone have borrowed about $200 billion of long-term funds, a quarter of the Treasury's full-year medium- and long-term issuance. Investment-grade bond funds have also been cutting Treasuries and adding corporate bonds this year, and foreign private money buying US corporate bonds has already surpassed buying US Treasuries—hard to imagine in the past.

A massive, price-insensitive private borrower permanently stationed at the long end rewrites the rules of the game in itself.

Bessent's buybacks cannot reach here either—each is only a few billion dollars, a drop in the bucket in a Treasury market of more than $30 trillion, and it can only adjust the Treasury's own maturity structure, not the pace of tech giants' issuance.

There is also a self-reinforcing loop hidden here: tech giants' issuance pushes up long-end rates, and higher rates in turn raise the funding cost of AI projects themselves—the premium AI giants pay on new bonds has soared from just over 2 basis points last year to 12 basis points.

But the problem is that the AI boom is very likely not to expand the government's tax base. Productivity and GDP may soar, yet the government's debt-servicing capacity may actually deteriorate.

This deterioration shows up mainly in two ways: long-term and near-term.

In the long term, US individual income and payroll taxes together contribute more than three-quarters of federal revenue, meaning the government's fiscal lifeline is tied to payrolls, while the economic effect of AI is precisely to compress the value and income of labor—companies do more with fewer people, profits rise, but total wages fall, and the government collects less and less from individual income and payroll taxes. If AI reduces labor income's share of GDP by 10 percentage points, most developed countries' finances would be in serious trouble.

In the near term, the "One Big Beautiful Bill Act" passed in 2025 sharply cut US corporate taxes; the law restored full expensing of capital expenditures in the year incurred. The bill's intent was to stimulate investment, but for tech giants already spending whatever it takes on data centers, this is not an incentive but a windfall. Microsoft disclosed in July that its tax payable for the period was only $2.5 billion, versus $14.1 billion a year earlier—an 80% shrinkage, while profits were soaring. That is the magic of depreciation. Data from the US Congressional Budget Office in September confirms this: in the first 11 months of this fiscal year, corporate income tax revenue plunged by a quarter year-on-year, a drop of nearly $100 billion.

Revenue shrinking, spending expanding, tax revenue falling off a cliff. The deficit can only continue to be covered by issuing debt. In the first 11 months of this fiscal year the federal deficit has already piled up to $2 trillion, total national debt just crossed the $40 trillion mark in August, and interest alone consumes about $1 trillion a year.

At the same time, government spending will not shrink but expand—because more people will need retraining and welfare support to get by.

As the government's fiscal gap tears wider, the money can only be found in the one place still expanding—the profits created by AI. Channel: Finance and Economics This content is published with the author's authorization; the views represent the author only and not the position of Huxiu. If you have objections or complaints about this article, please contact tougao@huxiu.com. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]