AI, tariffs and the trouble with Treasuries

Making sense of the latest trends in property and economics from around the globe
Written By:
Liam Bailey, Knight Frank
5 minutes to read

Whether or not US President Trump had triggered a full blown financial crisis felt like the key question when I wrote this note two days ago. ‘Not quite,’ was the answer, though the sell-off in US government bonds was beginning to feel like the first act of a Netflix cautionary tale.

A lot has happened since then; the President's 90-day reprieve for everybody but China triggered a 9.5% surge in the S&P 500 on Wednesday, the largest single day gain since 2008. The ratcheting up of China's tariff to 145% didn't seem to matter until the following morning, when sober analysts pointed out that the tariff mix was now actually worse for US consumers, given the scale of US imports from China (see chart).

So what has changed since Wednesday? Not much, on the face of it. Stocks have resumed their falls and Treasuries are still trading like emerging market bonds. The key question as to whether volatility in the Treasury market is due to hedge funds deleveraging or investors issuing deeper judgment on the safe haven status of US government debt remains unresolved.

UK borrowing costs

Granted, large falls in swap rates - a view that the Bank of England would pivot to propping up growth by cutting rates - have unwound to some extent. The five year SONIA swap rate closed at 3.96% yesterday, higher than it was a month ago and up from as low as 3.7% at the peak of the chaos.

That didn't stop Barclays from becoming the first major lender to cut mortgage rates since the onset of the tariff crisis yesterday. The bank now offers a 3.99% rate on two-year fixed, three-year fixed and five-year fixed mortgages with a deposit of at least 40%.

“Even with the President’s 90-day reprieve, risks to the growth outlook have clearly risen in the past four weeks,” Hina Bhudia of Knight Frank Finance told the FT.

The trajectory of borrowing costs from this point is anybody's guess. Bank of England policymaker Sarah Breeden said the tariffs would clearly weigh on growth, but admitted that the impact on inflation wasn't "clear cut". That uncertainty will be compounded by the fact that any UK deal with the US will take time - Trump is focused on countries that run large trade surpluses with the US, the FT reports.

Digital infrastructure

Lofty stock market valuations are often touted as a gauge of investor confidence in AI’s transformative potential. A more telling indicator might be the capital earmarked for the digital infrastructure needed to power it.

Knight Frank's Global Data Centres Report has plenty of numbers to boggle the mind. The data centre market will, on a rental basis, grow from $1.4 trillion today to $4 trillion by the end of the decade, a compound annual growth rate of 18%. For 2025 alone, the "Big Four" (Amazon, Microsoft, Alphabet, Meta) have committed €320 billion, most of which will go to AI infrastructure.

As AI's computational demands surge, data centres have become mission-critical assets for both hyperscalers and investors. The balance sheets of the world's biggest tech firms increasingly resemble utility companies, allocating capital not just to innovation, but to the physical infrastructure needed to support it.

Over the next three years, the global market will receive 35GW of new data centre capacity, of which 30% is expected to be dedicated to AI workloads. Realising this AI-based capacity will require approximately $120 billion in capital deployment and a real estate footprint of 120 to 150 million square feet.

Green tensions

Data centres are challenging from an ESG perspective. They are energy intensive, and our report picks ESG compliance challenges as among the key risks to the sector as governments enforce stricter energy usage and carbon emissions standards.

Indeed, data centres will use more than twice as much electricity by 2030 than they do today as artificial intelligence drives demand, according to new International Energy Agency forecasts. Processing a request made to ChatGPT takes about ten times as much electricity as a typical Google search.

That may make them a target for regulation, but policymakers would do well to consider that usage in the context of their contributions to productivity. From the Times: The IEA said, however, “concerns that AI could accelerate climate change appear overstated”. Although it sees data centre power demand as one of the fastest growing sources of emissions, increasing from 180 million tonnes today to 300 million tonnes by 2035, this would remain less than 1.5 per cent of total energy sector emissions. Such emissions could be more than offset by savings elsewhere, thanks to the widespread adoption of AI making things more efficient.

Algorithm anxiety

AI models like ChatGPT are irresistibly convenient, especially when the stakes are low. Skipping the scroll through search results to find a cake recipe is a real leap forward—after all, who cares about the source as long as it tastes good?

Higher stakes interactions with AI present a different proposition, and the Bank of England is worried: "A likely area of development over the coming years is advanced forms of AI increasingly helping to inform firms’ core financial decisions," policymakers highlighted in a new paper this week.

The BoE's key concern relates to herding, in which investors panic and dispose of assets in a way that makes crises worse - see the market for US Treasuries this week.

Greater use of AI to inform trading could "lead market participants inadvertently to take actions collectively in such a way that reduces stability," the authors add. "For instance, the potential future use of more advanced AI-based trading strategies could lead to firms taking increasingly correlated positions and acting in a similar way during a stress, thereby amplifying shocks. Such market instability can then affect the availability and cost of funding for the real economy."

The BoE wants better governance and oversight, including clear escalation paths and safeguards when AI tools are involved. Most importantly, it wants ultimate human accountability.

In other news...

Why holiday rentals are winning the war for Europe’s city centres (FT).