Notes · English

Does AI spending push up interest rates? What it means for Singapore

Big Tech plans about US$720-745 billion of capital spending in 2026 and has sold record bonds to fund it. Here are the three routes by which AI could raise interest rates, the evidence for and against, and how the boom reaches Singapore through chips and exports more than through bond yields.

CapsWindowPublished: 9 October 2026

AIInterest ratesBond yieldsInflation

Key takeaways

AI could raise interest rates through three routes: heavy bond issuance by tech companies (supply), stronger investment demand that lifts the neutral rate, and higher costs for memory chips and electricity that feed inflation.
The US 10-year yield rose 108 basis points in 2026 to 6 October, and 97 of those came from the real yield (TIPS); expected inflation rose only 11. That pattern fits stronger demand for capital better than an inflation scare, but it does not isolate AI.
The evidence is contested. PIMCO found no significant yield reaction around six large AI bond deals and argued AI could even lower neutral rates; Federal Reserve Governor Cook said AI appears to be adding to inflation in the short term.
In Singapore the boom shows up in exports and prices of what Singapore makes: electronics non-oil domestic exports rose 131.8% in August 2026 and the manufacturing deflator rose 14.1% in the second quarter. The 10-year SGS yield rose only about 35 basis points this year.

US long-term interest rates are at their highest in more than two decades, and a new explanation has joined oil and deficits: the money being poured into artificial intelligence. This note sets out how AI could lift rates, what the evidence shows so far, and what reaches Singapore.

Through which routes could AI raise rates?

Visual guide

Through which routes could AI raise rates?

Supply
Mechanism
Tech companies issue large amounts of bonds, which compete with government debt for investors’ money
What to look at
Bond issuance, spreads, Treasury reaction around big deals
Investment demand
Mechanism
A long investment boom raises the return on capital and the neutral rate
What to look at
Real yields, neutral-rate estimates
Cost
Mechanism
Chips, memory and electricity for data centres become dearer and feed into prices
What to look at
DRAM prices, electricity CPI, power auctions

Selected entries from the table below. Its dates, units, definitions and limitations still apply.

Sources ↓

Route	Mechanism	What to look at
Supply	Tech companies issue large amounts of bonds, which compete with government debt for investors’ money	Bond issuance, spreads, Treasury reaction around big deals
Investment demand	A long investment boom raises the return on capital and the neutral rate	Real yields, neutral-rate estimates
Cost	Chips, memory and electricity for data centres become dearer and feed into prices	DRAM prices, electricity CPI, power auctions

The first two would show up mainly in real yields. The third would show up in inflation expectations.

How much are tech companies spending and borrowing?

Four companies, Amazon, Alphabet, Microsoft and Meta, guided to roughly US$720-745 billion of capital spending in 2026 at their July results, according to company guidance compiled by Platformonomics. That is about 2.2-2.3% of US nominal GDP. Goldman Sachs expects five companies, including Oracle, to spend US$1.2 trillion in 2027.

Much of it is borrowed. Morgan Stanley expects about US$570 billion of AI-related bond issuance worldwide in 2026, and about US$236 billion had been sold by the end of May, four times the year-earlier pace (Reuters via MarketScreener).

Deal	Size and feature
Alphabet, February	US$20 billion in dollars, plus a £1 billion 100-year bond (6.125% coupon), the first 100-year tech bond since 1997
Amazon, March	about US$53.8 billion equivalent, the largest corporate bond sale on record, with about US$158 billion of orders (IFR)
Meta, April	US$25 billion, maturities from 2031 to 2066

Investors have charged more for it. The median spread on hyperscaler bonds with more than 20 years to maturity rose from 108.5 to 118 basis points, and the median new-issue premium from 2.25 basis points in 2025 to 12 in 2026, according to a Reuters tally.

Did the bond supply move government yields?

The direct evidence is weak. PIMCO examined six large AI bond deals and found no significant reaction in the 10-year Treasury yield, the term premium or swap spreads around them. Capital Economics wrote on 1 October that the effect was limited. On this reading, the extra cost fell on the companies’ own borrowing, not on the government’s.

What does the split between real yields and inflation show?

+97 bp vs +11 bp

Change in the US 10-year real yield (TIPS) and in 10-year expected inflation, 2 January to early October 2026 (FRED DFII10, T10YIE)

US 10-year measure (FRED)	2 January 2026	Early October 2026	Change
Nominal yield (DGS10)	4.19%	5.27% (6 October)	+108 bp
Real yield (DFII10)	1.94%	2.91% (6 October)	+97 bp
Expected inflation (T10YIE)	2.25%	2.36% (7 October)	+11 bp

Almost all of the rise came from the real yield, which reached 2.95% on 5 October, the highest since November 2008. That fits the investment-demand route better than an inflation scare. Estimates of the neutral rate have edged up too: the New York Fed’s Holston-Laubach-Williams r* went from 0.89% in 2025Q4 to 1.01% in 2026Q2, and the FOMC’s median long-run rate projection from 3.1% in June to 3.2% in September. None of this isolates AI from oil, deficits or the term premium.

Is the cost route showing up?

Yes, in specific prices. Contract prices for conventional DRAM rose 93-98% quarter on quarter in the first quarter of 2026 (TrendForce). US electricity CPI was up 3.8% in August. In the PJM grid region, the 2027/28 capacity price cleared at the US$333.44 per MW-day cap for the third record in a row (Utility Dive).

Fed officials have noticed. Governor Cook said on 28 September 2026: “In the short term, AI appears to be adding inflationary pressures to the economy.” Governor Barr said in February that “the AI boom is unlikely to be a reason for lowering policy rates.”

What is the case that AI lowers rates?

PIMCO argued on 1 July 2026 that AI could lower the neutral rate over the short to medium term. Households may save more as a precaution against job loss, the labour share of income may fall, and fiscal expectations may change. Across 43 major AI model release days, the US 5-year 5-year forward real rate fell by about 100 basis points in total, while it rose about 100 basis points over the full period. That is an event study, not proof, but it shows the direction of the effect is not settled.

What reaches Singapore?

Singapore is a producer in this boom, and the signal is clearer in trade and prices than in bond yields.

Indicator	Value	Period
Electronics non-oil domestic exports	+131.8% year on year	August 2026
Integrated circuits / personal computers within that	+90.9% / +237.9%	August 2026
GDP deflator	+6.1%	2026Q2
Manufacturing deflator	+14.1%	2026Q2
Share of Micron’s NAND made in Singapore	98%	2026
Data centre capacity	about 1.4 GW in more than 70 facilities, plus at least 300 MW planned	May 2024
10-year SGS yield, 2 January to 7 October 2026	2.13% to 2.48% (about +35 bp)	ADB daily data (Bloomberg)

Export figures are from August trade data; deflators from the MTI Economic Survey appendix. Micron is building a US$24 billion NAND plant here, due to start production in the second half of 2028. Data centre figures are from IMDA’s Green Data Centre Roadmap, which allocates new capacity to operators that prioritise sustainability.

For households, the effect so far is in the prices of devices, covered in chipflation, more than in borrowing costs. Singapore’s ten-year yield rose about a third as much as the US yield this year.

To compare what a sum of money was worth in different years, try the past money calculator.

Open the past money calculator →

No study cited here isolates the AI effect on interest rates from oil, fiscal deficits and the term premium. Capital spending figures are company guidance and can change. Singapore's export and deflator figures show where AI demand lands; they do not show that it moved Singapore's bond yields. The data centre figures date from May 2024.

Questions & answers

How does AI investment raise interest rates?

Through three possible routes. Tech companies borrow heavily in bond markets, adding to the supply investors must absorb. A sustained investment boom can raise the neutral interest rate. And demand for memory chips and electricity raises costs that can feed inflation. Each route is debated, and none has been measured precisely.

Is there evidence that AI bond sales lifted Treasury yields?

Not clear evidence. PIMCO's 21 September 2026 study found no significant reaction in the 10-year yield, term premium or swap spreads around six large AI-related bond deals. Hyperscaler bond spreads and new-issue premiums did widen, so the cost showed up more in corporate credit than in government yields.

How does the AI boom affect Singapore?

Mainly through what Singapore produces. Electronics non-oil domestic exports rose 131.8% year on year in August 2026, Micron makes 98% of its NAND flash in Singapore, and the second-quarter GDP deflator rose 6.1%. Singapore's 10-year SGS yield rose only about 35 basis points in 2026, much less than US yields.

Could AI lower interest rates instead?

Some argue so. PIMCO wrote on 1 July 2026 that AI could lower the short- to medium-term neutral rate through precautionary saving, a lower labour share and fiscal expectations. It found that across 43 major model release days, the US 5-year 5-year forward real rate fell by about 100 basis points in total.

Source

Calculators

Past money valuePast money value · SGD

Explore more

Global bond yields 2026: US, UK, Germany and Japan at multi-year highs. Where is Singapore?Ten-year yields in the US, UK, Germany and Japan hit their highest levels since 2002, 2007, 2011 and 1996 in 2026. Singapore's 10-year SGS yield rose only about 35 basis points this year and remains below the middle of its own history.Singapore 10-year SGS yield: how high is 2.35% in its own history?The 10-year Singapore Government Securities yield ended August 2026 at 2.35%. Across 339 month-ends since June 1998 it ranks 208th, below the middle of its own history, while US, UK and Japanese yields sit at multi-decade highs.What is chipflation? Singapore laptops +26%, CPI PCs −2%Chipflation is the AI-driven memory shortage raising gadget prices. Singapore laptop prices rose 26%, yet the CPI for computers fell 2.0% in August 2026.How a supply shock reaches your currencyFreight, refining, inflation, bond yields, exchange rate: the 2026 energy shock traced through five measurable steps, and what each step does not tell you.