Market Update 21st September 2026

Memory loss

The US raised interest rates by 25bps on Wednesday. It seems a long time ago since we all thought US rates were bound to go down because the President wanted it and would hire someone specifically to do that. He apparently believes rates should be 1%.

Also on Wednesday, Emily and I were at a conference where a former head of the British army said the Iran war was “up there” as one of the most foolish ever – a crowded field.

We have touched before on the troublesome issue of Western governments wanting and needing to spend more – to boost growth and productivity, to reshore industry, infrastructure for more energy security, for defence now that the US has stepped back – while at the same time being constrained by lack of funds due to social and demographic trends despite tax rates as a percentage of GDP being at relative highs. Now the cost of this capital is increasing again creating further pressure on an already tight situation.

At the same time, businesses are also raising and spending historically large amounts of money. Technology stocks again take most of the headlines here. According to JP Morgan US hyperscaler capex was just over $400bn in 2025, is forecast to be over $800bn in 2026 and well over $1 trillion in 2027. The total amount to 2030 is forecast to be $5.6 trillion. After 2030 they believe it will gradually normalise to around $800bn per year.

Estimates for the amount of debt being issued by the hyperscalers are around $250bn in 2026. This is somewhere around half of the expected debt issuance of the US government this year.  JP Morgan think around half of their $5.6 trillion longer term capex estimate will be financed by debt.

There is some nervousness about why the previously asset light and cash generative tech companies have suddenly turned to vast debt issuance to fund themselves, but they are looking at what they hope is a long-term investment and product cycle and raising capital / liabilities in a way that matches their assets makes sense.

We are also seeing more and more people contextualise AI spending with previous transformative investments. The New York Fed estimated earlier this year that AI spending between 2025-2032 would average 2.8% of US GDP per year. This compares with 2.4% for railroads between 1865-1890, 1.1% for electrification between 1905-1925 and 1.6% for highways between 1956-1973. You might remember the current NATO core defence spending target is 3.5% of GDP for all of its members. The UK is currently at about 2.4%.

The absolute AI spending numbers always sound enormous and challenging, but they are not out of line with previous and current precedents as a percentage. Inflation of course distorts the absolute amounts, and you will notice that there are big differences in the number of years these investments took. AI investment will go on well beyond 2032. JP Morgan’s $800bn future steady state estimate is about 2.5% of US GDP.

Does the market mind lending all this money to the tech companies? Not so far. The US Technology and Electronics spread – the excess return demanded by lenders is currently 88bps. This is a little higher than the wider US Investment Grade spread which is at 78bps. The US 10Y treasury yield is 4.95% as I write, so Technology and Electronics yields are 5.83% vs 5.73% for the wider market.

The Technology and Electronics spread has risen from a low of 70bps earlier this year, but these are still historically low levels. It went to nearly 350bps during Covid, and 520bps during the financial crisis. No one was asking for $250bn back then. If the market minded lending all this money the spread would widen further as it demanded a higher return. It is a signal we can keep an eye on.

The high yield market – where much of the data centre money is being raised – tells a very similar story – spreads have widened a little this year but remain historically low.

Are they going to pay this money back? I think so. Using Factset I make it that the five largest hyperscalers – Microsoft, Amazon, Google, Meta and Oracle are set to make around $670bn of earnings before interest and tax over the next 12 months. Set against this they are forecast to spend about $1 trillion in capex (so slightly less in the next 12 months than the calendar year 2027 mentioned earlier). Against this they are borrowing $250bn. These numbers send Amazon, Google, Meta and Oracle into negative free cash flow in the coming years although they do start to improve again.

But borrowing $250bn means their debt is less than 0.4x their collective earnings before interest and tax. If we used their earnings before interest tax and depreciation this is forecast to be just over $1 trillion and the debt would be 0.25x. The implied depreciation over the next 12 months is around $355m, which is significant and will continue to grow in future years. If JP Morgan’s estimate is right, they will need to borrow more than $250bn per year between 2027-2030, but their earnings are also forecast to continue growing.

None of this is to say whether they are good or bad equity investments, but I am just trying to provide context to the very large numbers which are now regularly reported and whether or not they are affordable. I think they can afford it – the question then becomes whether they can make a return on the money spent.

When we look at equity markets it can feel counterintuitive to see them reaching successive all-time highs. There are large scale wars going on with no end in sight, geopolitics, tariffs, $100 oil, high rates, more volatile inflation. At best some of these increase uncertainty which is not supposed to be good for equities, and some are straightforwardly unhelpful.

But it is these increases in spending from both governments and corporates that has been a key driver of equity markets in recent years and also part of the reason performance has become concentrated in certain sectors. These trends look set to continue for the time being.

Going from near zero rates up to over 5% in 2022 was painful and took some time to adjust but some of this increase in the cost of capital is healthy – higher bond yields can be positive and memories are short. During a decade plus of zero interest rates the issue was “free” money leading to “anything” being financed with mixed results. Now everything is too expensive, and the US President wants to go back to near zero rates.

This applies to governments (of any party) as well – money is tight and this should focus minds on what is the absolute best way to spend it. There is even an argument that the large tech companies are crowding out government bond issues and forcing yields higher. I don’t really envy the UK Chancellor in the coming weeks, but these are long term problems in need of a meaningful solution.

Robert Fullerton – Senior Research Analyst

IMPORTANT INFORMATION

This is a Financial Promotion. Hawksmoor Investment Management Limited is authorised and regulated by the Financial Conduct Authority (www.fca.org.uk) with its registered office at Sterling Court, 17 Dix’s Field, Exeter, Devon EX1 1QA.

This document’s content should not be interpreted as investment or tax advice for which you should consult your independent financial adviser and or accountant. The information and opinions it contains have been compiled or arrived at from sources believed to be reliable at the time and are given in good faith, but no representation is made as to their accuracy, completeness or correctness. The information and opinions expressed in this document, whether in general or both on the performance of individual securities and in a wider economic context, represent the views of Hawksmoor at the time of preparation and may be subject to change. Past performance is not a guide to future performance. The value of an investment and any income from it can fall as well as rise as a result of market and currency fluctuations. You may not get back the amount you originally invested. FPC26769.