Home Insights Global macro

Global macro

Published 27 May 2026

Global markets

The deal we were all waiting for (and have been waiting for a while) continues to take longer. We blogged yesterday about how newsflow fell away, with the restart of military strikes. It seems that several Iranian boats were laying mines again, in the Strait, and hence the US employed “defensive strikes” to remove these IRGC vessels.

Also interesting, as I understand it, Iran has switched back on the internet, which had been closed to the masses for over a month. That would be a generally constructive sign, indicating things are intended to get back to normal.

So the two news items there stand somewhat in tension with each other, but there you have it.

Our thinking here is very muddled, but perhaps so too is everyone else’s thinking. If markets stay high, or ratchet higher, and time goes by and the Strait stays closed, we are inclined to sell, given the rising option value of cash.

If markets drop, and drop aggressively (these days we are accustomed to markets dropping 8% in a month, alas, when they do fall) we’d be inclined to revert to our standard play book, which is to monetise outperforming defensives. Running a well diversified multi-asset set of portfolios means you almost always have something that is doing well. But that something generally is determined by where the falls are the largest, i.e., do emerging markets get hit harder than developed, do credit spreads widen, does the AUD fall further/quicker, etc. It is sub-asset class dependant, on what looks like the best relative value in such a correction. That is probably pretty obvious, but, as we said earlier, there you have it.

Lurking in the background, I am simply amazed at how China has been able to curtail their oil imports, which on some reads have dropped by the most in well over a decade. There simply must be more inventory on hand that we can’t easily see, because that kind of energy import drop would normally be associated with the domestic economy lurching to a halt, which is unlikely.

AI

People often ask us what kind of markers should one look for, to determine if the AI trade is running out of puff. I suppose the problem as I see it is that bubbles almost always involve “good news”, in this case, earnings going up for stuff most directly linked to the bubble, and so searching for the below doesn’t necessarily tell you anything new.

Anyway, here’s the rental cost of GPUs. Remember GPUs are chips, similar to CPUs, they run the big matrix multiplication calculations that are the engines of/for LLMs. And to the degree they are going up strongly suggests ongoing demand.

But we knew that already.

And in any case seconds later you then read the below.

Clear as mud.

As I understand it (gosh I have to write that line so often lately) token usage has a) exploded b) might not be very effective. You burn a lot of tokens just trying to iron out bugs in what the AI gave you in the first place.

The Uber CEO (as in the ride hailing app owner) said something like “we spend all this money on AI but it isn’t obvious we are shipping all that much more”. Cost per token might have come down, but total tokens used have gone up, so that costs can sometimes “blow out” to the point where they match what a human workforce would’ve cost in the first place.

Again, I don’t know how to be definitive here. Xero and Wisetech (big figures in our local market) say it’s all amazing and they are doing so much more for so much less, and Richard White got into trouble for saying something like “why bother with a human when an AI agent can do it for so much less” (I am wildly paraphrasing from memory, so far better to check the exact quote). But then Uber and many others saying the cost is exploding and they are pulling back because the AI budget is being exceeded without tremendously obvious benefit works against this.

Perhaps reiterating our stance while trying to sum up. AI is very good and useful. It is probably not as useful as some of the most optimistic hope. Given the trillions of dollars being spent, and the market caps that are being bid up, the odds favour a mispricing (akin to a bubble). The very structure by which LLMs reason makes them vulnerable to hallucinations (let’s just call it errors, based on how pairwise correlations select for the answer in absence of what a human would call contextual reasoning) is likely not fixable. So you can’t get “6 9s reliability” and that will limit how far and wide these things can go.

And it shouldn’t be all that contentious to say “these are cool tools with clear commercial application” on the one hand and “don’t value them in the trillions” on the other. But to the degree you’ve positioned relative to market, it is contentious.

Some other quick stories. Starbucks scrapped an AI inventory led system after trialling it for 9 months. The AI just couldn’t quite (via cameras) get things right. Microsoft has done similar things, but I think it was so it could focus more on Microsoft branded solutions (like OpenAI/github etc).

Fund managers

SGH ICE (a quality/growth franchise specialist in Aussie shares) put out an interesting graph in their April commentary. Nearly everything they own is at the bottom end of its historical PE range.

That means those companies have suffered enormous PE de-rates (the multiple the market was willing to apply to their earnings). Now you can probably say that many of those PEs were clearly historically quite high, at some point, given the average there is ~30x (10 years is a long lookback), but the point is present valuations are clearly more reasonable. If the underlying growth rates of those companies haven’t overly changed, then perhaps the future is appealing.

To be clear, these are general comments. We don’t know what’s in the fund, and we see the top 10 holdings same as everyone else. But some of those, like SDF, AUB (insurance brokers) have been caught up in the SaaS-acre, where fears of AI disintermediation have crushed enthusiasm for the stocks, are certainly of interest. If AI doesn’t in fact gobble up the business models of these companies (perhaps these companies can in fact use AI to further enhance their offering, or to remove cost, and thus offer better pricing etc) then they could be very good opportunities.

Walter Scott (another Quality Growth style manager) who targets (in part) stocks with desirable factor traits (e.g., high/stable/growing returns, cashflows, profit measures, low gearing, good growth rates (at the very least historically), that sort of thing) tabled the below.

Quality has been crushed. Growth has been crushed. Growth, as proxied by IT, has done well, as long as you were in the Mag 7 and chips/semis/memory (“picks and shovels, of the AI gold rush”) but awful if you were in Software, and stuff that is software adjacent. You can, I think, put platforms and payments in that camp of stuff that’s been affected by the AI led sell off in potentially impacted business models. Maybe AI means a competitor to Visa or Mastercard or Moodys or Intuit comes along, and eats their economic lunch. Either way, a very interesting graph.

Healthcare, software, quality, all very weak.

Aus inflation

Aus inflation still looking unhealthy. Still, 4.2% is “better” than 4.6% (monthly at 0.4% better than last month’s monster 1.1%).

Market finding some joy in the 10-year pulling back from volcanic-eruption style levels. Rates and dollar down, markets up is usually a combination of a dovish surprise. We did already have some degree of comfort from the RBA that they might be on hold for a little bit, even as more hikes are generally expected. The jobs print of last month was weak enough to change things a bit, at the margin.

We think in the long run restrictive monetary policy will work, and inflation will settle down. But it is clear it remains outside the target band, and hence the periodic market swings towards hikes, and a hawkish bias from the RBA, are very understandable.

Important Information: This document has been prepared by Aequitas Investment Partners Pty Ltd ABN 92 644 165 266 (“Aequitas”, “our”, “we”), a Corporate Authorised Representative (no. 1284389) of AIP FS Pty Ltd ABN 49 680 982 478 (Australian Financial Services Licensee no. 700016) and is for distribution within Australia to wholesale clients only.

This document is based on information available at the time of publishing, information which we believe is correct and any opinions, conclusions or forecasts are reasonably held or made as at the time of its compilation, but no warranty is made as to its accuracy, reliability or completeness. To the extent permitted by law, neither Aequitas nor any of its affiliates accept liability to any person for loss or damage arising from the use of the information herein.

Past performance is not a reliable indicator of future performance.

General Advice Warning: This document has been prepared without taking into account your objectives, financial situation or needs, and therefore you should consider its appropriateness, having regard to your objectives, financial situation and needs. Before making any decision about whether to acquire a financial product, you should obtain and read the relevant Product Disclosure Statement (PDS) or Investor Directed Portfolio Service Guide (IDPS Guide) and consider talking to a financial adviser.

Taxation warning: Aequitas is not authorised to provide tax (financial) advice and is not a registered tax agent. Investors should seek their own tax advice from a registered tax agent or a registered tax (financial) adviser if they intend to rely on this information to satisfy the liabilities or obligations or claim entitlements that arise, or could arise, under a taxation law. We recommend that you consult a registered tax agent or a tax specialist to confirm your taxation position.

Share this post

Do you want to stay ahead of the market?

Sign up to receive daily updates on investment topics and monthly updates on portfolios directly from the desks of our Joint Chief Financial Officers.
Stay ahead of the market and receive adviser insights straight to your inbox.