Eric Lascelles - Managing Director, Chief Economist and Head of Investment Strategy Research

Hello and welcome. My name is Eric Lascelles. I'm the Chief Economist and Head of Investment Strategy Research for RBC Global Asset Management.

I’m very pleased to share with you our latest monthly webcast for August 2026.

The title for this August 2026 webcast is: Running the gauntlet, AI concerns, oil shock, tariffs.

And so the title certainly speaks to some of the challenges that are out there right now, including in the AI space which I'll get to in a moment.

Certainly that oil shock alas continues. The Strait of Hormuz is more closed than open right now and there have been some new tariffs announced and potentially implemented.

So we'll grapple our way through that but I do want to mention this title perhaps doesn’t convey the nuance of the situation because we're still seeing fairly good economic growth.

We still have above consensus growth forecasts, so these are new issues to deal with, but they are not undermining the broader macro story completely.

Okay.

Report card: Let's jump in – and as we always do, we'll start with a report card and indeed within that, why don’t  we go with the positive themes first. And so, again, to reiterate that some things are going rather well.

I can say that macro variables are remaining broadly solid in terms of the data releases that we've been seeing and so consistent very much with economic growth as opposed to decline.

The global economy, in turn, is proving fairly resilient.

We are continuing to get what we would describe as pretty decent growth – and within that, bank lending is in general accelerating more than it is slowing down. That would reflect banks feeling confident in their clientele, but it also mechanically means more money is sloshing around and that is something that helps the economy as well.

We are continuing to think that productivity growth is more likely to be strong than weak. And we think that artificial intelligence is already starting to add to that and that that can continue – and that's an important underpinning for any kind of economic forecast. So perhaps not a surprise, then, we are maintaining above consensus economic growth forecasts right now.

So we are still relative optimists even though we are about to talk about some challenges that absolutely do exist in the world today.

Let's talk about those challenges. That's the negative themes in front of you.

So the war with Iran has, you might say, revived and so it's ebbed and flowed – I would say repeatedly at this point – over the span of the last several months. But in the last few days, there have been further attacks and the Strait of Hormuz is functionally closed right now. So that problem has not been fully resolved and we will talk our way through that in a moment.

Related to that, of course, oil prices are still fairly high. Inflation is itself still too high, even though it should be noted the June inflation prints around the world came down quite nicely – but they may not come down quite as nicely as the July numbers and beyond are released.

For central banks, of course, they're considering relative economic strength on the one hand, they're looking as per their other mandate at inflation, which is still too high. And so central banks are thinking more about rate hikes than not and the Fed in particular is continuing to give somewhat hawkish signals that would hint that there may be some rate hikes coming in the coming months. And then, of course, another policy decision, the White House has announced another round of tariffs – and many of those are just replacing prior tariffs that were blocked by the Supreme Court earlier in the year and replacing temporary tariffs that have been expiring in recent weeks. But nevertheless, tariffs to some extent on the rise again and indeed a completely new and unexpected set of tariffs threatened against Canada that we will talk to as well.

And then artificial intelligence, another concern. Obviously, an opportunity in many regards but there are concerns on a few fronts.

Some are just whether the so-called frontier models -- the expensive cutting-edge models – can be profitable when there are cheaper open models that exist that are nearly as good.

The big hyperscalers are still very profitable, but their free cash flow is deteriorating quite significantly as they continue to invest, so some concerns related to that. And the falling token prices mean they may not get to collect quite as much revenue in the future.

Alongside that, data centres are becoming less and less popular with politicians and voters and so on – and relating to higher electricity costs and they don't actually employ that many people.

And more generally, you can say some concern around AI valuations and perhaps of greatest focus right now, some of the Korean memory chip makers.

But more generally, just whether this can all hold together at current pricing.

So there are some concerns. I'll speak to those as we go through this presentation.

And so no surprise, then, though, that markets are choppy right now.

There's a lot going on and a lot of different things to consider.

And then lastly, just on the interesting file – and this was just an opportunity to wedge in an additional bit of research we've done recently – but we'll talk about the varied artificial intelligence implications by sector in terms of –the implication for demand for that sector’s wares, for profits, and for that matter, the workers in those sectors, as well.

Okay. That's the overview. Let's just push forward now and we'll add some pictures to the show and talk at more length about many of those subjects.

Attacks on vessels in Strait of Hormuz have resumed: So we'll start with this war with Iran, this energy shock of sorts.

So we are tracking the number of vessels attacked in the Strait of Hormuz – and it’s never quite died down to nothing – but there was a period of relative tranquility from let's say May through to the end of June and you can see there have been a notable number of attacks more recently.

It is not safe to transit that Strait – and indeed, perhaps more evocatively, we can go to the next chart and observe that while there had been a tentative if incomplete revival of shipping throughout the Strait – and this is specifically oil tankers broadly -- you can see there is very little transiting again.

Strait of Hormuz re-opening paused: So Iran is attacking those ships and there is quite a battle both physically but also a debate as to whether Iran will be allowed to essentially apply a toll to ships crossing the Strait.

And so I think the takeaway here is one in which it's not clear there is going to be a near-term neat and tidy resolution to this war. We would like to think some progress can be made in the coming months, the price of oil probably doesn't get to settle all the way back to where it was before the war.

Oil prices have significantly reversed: And so as we look here at the price of oil, you can see of course, for a moment, a few months ago it rose to as much as $120 a barrel and there was a moment actually where it briefly fell to just over $70 a barrel.

And it's since been wavering back and forth.

But it does look as though, again, it probably doesn't get to settle all the way back to where we were before the war started. So in turn, inflation doesn't get to completely settle over the next six months or so and there is at least a subtle drag on the global economy as well.

Divided U.S. government ahead – implications (at the margin): Let's shift focus here and talk about something that we actually haven't mentioned, mid-term elections in the U.S. These are sneaking up on us.

As I'm recording this at the very end of July, we are getting awfully close, just over three months away from the U.S. mid-term elections. And so at this juncture, it looks likely that the government will emerge from that election divided and so that's a round-about way of saying the House of Representatives probably flips from the Republicans to the Democrats, so it's no longer a Republican sweep between the White House and the Senate and the House of Representatives

So what does that mean for the economy and for markets?

I want to emphasize that these are all at the margin implications.

We don't think this is likely to be the dominant force for markets. It's just an extra consideration to think about. But we would argue that perhaps the U.S. economy runs a little bit less quickly. Perhaps inflation runs a little bit lower, so that's a good thing.

That's why that's in green.

Stock marketss might advance a bit less quickly, bond yields could fall as opposed to rise. That's also a useful decline.

And perhaps the U.S. dollar could be a little bit weaker as well.

Again, those aren't actually explicitly our forecast.

We think other things will matter as well. But this is what the election itself might do.

And just to talk our way through, with far too many words, with some of our thoughts here.

Well, in terms of negatives – and it's hard to define what a negative is – and it might be bad for GDP but good for inflation –  so it gets blurry awfully quickly.

But talking through some of the key thoughts we have on this subject, one would be classically a divided government just has less fiscal stimulus.

There's less scope for agreeing on tax cuts or big spending plans and so on.

So that hurts growth, though as we'll talk to later, not the worst thing in the world for the big deficits and big debt.

The U.S. is running probably harder to pass a budget, and so that's not ideal. And in particular, the next debt ceiling is set to be reached perhaps next spring – and so that could make for a precarious moment that no one will much like.

The Democrats in general, you might say – at least the current incarnation – are somewhat less business friendly, and as much as they don't get to set legislation or raise taxes or anything like that, you can say well there's not going to be as much tax cutting to the extent the Republicans are no longer purely in charge.

But also, you know, a party in charge of the House of Representatives can conduct investigations and they do become members of important committees and so on. So it does change the dynamic somewhat.

Related to that, the deregulatory push in the U.S. probably weakens. And so if the oil and gas sector had benefitted before, it probably doesn't benefit as much.

The banking sector was benefitting from deregulation. Not clear that's over, but nevertheless, your could imagine a bit less of a push in that direction going forward.

We would posit that perhaps there's a bit less support for the AI industry.

Some would debate that, by the way.

The general view is both parties are pro-AI – but between electricity costs and data centre unpopularity and concerns about the implications for workers, it wouldn't surprise us if the Democrats perhaps provided a bit less support than the current government does.

Pharmaceuticals, well, you know, there have been a number of drugs that have been price restricted, where the price is capped and it wouldn't be a surprise if the Democrats pushed for more.

So good for inflation, good for households, bad for pharmaceutical makers, you might say.

And then for defence spending, I think in general, both parties are reasonably supportive of spending there, but you might argue perhaps the Democrats a little bit less –  though it's very nuanced, and the Democrats are perhaps more supportive of the war in Ukraine, Republicans perhaps more supportive of the war in Iran.

So it does get nuanced but nevertheless less support there, too.

Conversely, it's good if there are fewer budgetary excesses.

That's a welcome thing from a fiscal standpoint and could pull yields down and so on.

There could be a little less protectionism. Tariffs have been primarily a Republican pursuit.

The Democrats are not exactly pro free traders but they might be less supportive, nevertheless, of that sort of initiative.

Maybe a bit less support or even just a bit less antagonism toward green industries which we've seen in recent years.

And then both are very much advocates for, you might say, pocketbook economics, for households, for consumer spending. So there could still be friendly measures implemented in a bipartisan way there.

Ultimately, a little bit slower economy, a little bit less inflation, maybe stocks a bit weaker as well.

Okay. We'll keep pushing forward here.

S&P 500 weighed down by recent AI concerns, though broader market still advancing: Here is me pivoting into artificial intelligence and I'm just going to hit a number of different quasi-related subjects here consecutively.

So I'll start with this, which is you would be used to looking at the blue line on this chart. This is just the S&P 500, the big U.S. stock market.

It has been a good year. But you'll notice that blue line has been running roughly sideways since May.

So the prior advances have stalled out to some extent and that reflects significantly the fact that the hyperscalers, the biggest U.S. tech companies, are no longer going upwards as reliably and have even come off somewhat.

So some concerns about AI you might say.

Interestingly, just to provide a bit of a contrast, the gold line is still rising and actually has outperformed the overall stock market since the start of the year.

So what is that gold line - that is an equal weight index.

Instead of giving giant, giant weights to the biggest companies, which is naturally what the S&P 500 does, you give an equal weight to every company. You'd still be seeing gains so that's a round-about way of saying that many companies in the stock market are still going up.

In fact, most are still going up. It's these hyper scalers that are underperforming right now.

So this is not a broad market decline, this is a narrow market decline and there are places to hide in the stock market that can help you avoid that.

Still, it does reflect the fact that markets are more wary and a bit more skeptical about those big tech companies and you can see that expressed in a different way in this next chart.

Credit market demanding more compensation on AI borrowing: These are credit default swap spreads which means the risk of a company failing but the notion is the same, which is the credit market is less friendly.

Up is bad here. These are basis points and how much essentially a company has to pay to borrow –and they're having to pay more. And I'll say, you know, some of that is perhaps again the market just thinking a bit harder and worrying a little bit about some of these very expensive business models – and wondering whether all of the growth projections can actually come true.

Some, though, is also these companies are now issuing more debt. And so they've been historically so profitable they didn't have to do that.

They are now investing so incredibly in their models and in data centres and so on that they're starting to borrow.

And so, as a result, the cost of borrowing for them is going up.

I should emphasize they're broadly still in very sound financial health and for the most part don't fully have to borrow, or borrow as much as they have.

They're just finding the interest rates to be attractive.

Those interest rates are becoming less attractive now. And it does again reflect markets becoming a bit less favourable toward that sector.

Now, don't get me wrong, we shouldn't lose sight of the big AI story here, which is when you try to measure the quality of artificial intelligence, it's improving astonishingly.

AI improving at exponential pace: So these are various bleeding-edging models and how good they are according to at least one metric of model intelligence. And you can see every model more or less getting better at a rapid rate.

If you weren't paying close attention to the Y-axis, you might miss this is logorhythmic scale.

So the fact this looks like a straight line upwards and to the right means there is exponential increase in the quality of the AI, so it is getting a lot better and we really shouldn’t underestimate how remarkable it can become and we shouldn't underestimate any of the amazing sort of statistics associated with this AI boom. And, you know, I believe Google had announced there's been a 7 times increase in the number of tokens demanded over the last year and so demand is rising. The quality is rising.

This may all hold together beautifully but, you know, the market is becoming a bit more discerning.

Frontier models are very expensive to develop: One thing it's concerned about and you can see this on this next chart is just how expensive it is to train the next frontier model.

And so, you know, that cost is going up a lot as well.

This is also an logorhythmic scale, so a straight line actually represents an exponential increase there.

It's getting 3.4 times more expensive per year to train these models. So ever more money being thrown into them and they are getting ever better and the rate of improvement is running faster than the rate of increase in the cost. But you do still need to be able to charge for those.

So there are still some questions in part because of this next chart

Open models offer 93% of the performance at 1/8 the cost: So it tries to show, the blue line tries to show just a different metric, the improvement in quality of the – you might say – the closed weight models which is really the frontier models, the hyperscalers, the big fancy ones that you've probably heard of and maybe are even using.

However, that gold line is kind of sneakily keeping pace if not quite as good. Those are open weight models. Loosely, you would say these are models that are not quite as good.

That's why the gold line is below the blue line but what you don’t see here is they are radically cheaper to develop.

A lot of time they're sort of standing on the shoulders of blue models and copying them a little bit.

But in the end, they're much, much cheaper.

So what have you here are some open weight models-- a lot of them are Chinese but not exclusively -- and they're only about 7% worse in performance than the expensive closed weight models, but they cost in the realm of 1/8 the cost.

So I guess the point we would make here is that it presents a real challenge to the frontier models – and not every bit of AI needs to be the greatest in the world. And it might well be that a slightly less intelligent model is more than enough for a lot of things.

So, you know, at one point, we've been making the point it's not clear this is going to be a winner take all market.

It looks as though it may be a competitive market with room for many players.

It may be a lot of the AI work can be done by very inexpensive models.

It may be that only a subset needs to be done by these greatest models

That is a challenge, of course, to the huge sums of money being spent on the very best models.

Memory chips are now more than half the cost of an AI semiconductor: And, you know, another challenge out there is the cost of memory chips. And so these are a nontrivial, a very large fraction of the cost of building out a data centre.

In fact, I believe  more than half the cost of an AI chip are the memory chips that are embedded within it. And you can see sort of out of the blue and incredibly the cost of those memory chips have gone up on the order of 700 to 800% over the last year.

And so this is an incredible increase. This speaks to why some of these memory chip makers, many Korean, have soared absolutely.

Historically, that's a market, though, with low margins. Historically, it's a very commoditized cyclical business, so the fear is that it all just falls apart and goes back to normal – and usually these sorts of industries invest in expansion maybe not exactly at the wrong moment, but the expansion itself just removes the shortage and margins go away and they sort of regret it later.

So the question is whether this can stick – and for the moment, it's a challenge to the AI sector. It's a big, big cost.

Equally, though, the market is getting a bit nervous of some of the valuations of these particular companies and at least in recent weeks have been selling them off to some extent.

So maybe sanity prevailing, in particular as it pertains to the Korean Kospi index.

AI hyprescalers are significantly ramping up CapEx: So, don't get me wrong, I know I'm whip sawing you here.

We're seeing remarkable CapEx spending by the hyperscalers.

The amount being spent is incredible.

Could well be a trillion dollars by just the five biggest companies next year.

The percentage rate of growth is incredible through 2026, even though it's expected to slow in 2027 and 2028.

Just the sums involved are such that you're still talking about an extra quarter of a trillion dollars perhaps spent next year versus this year.

That still adds a lot to economic growth.

So this is still economic growth supportive.

It does reflect the conviction these companies have that these models are for real and there are, you know, powerful things to do I should say with these models.

But ultimately you can also say it's a lot of money and they do need this to pay off.

So one way, in fact, to frame this is just to say that when we look at conventional and I'll speak to this chart in a moment, when we look at conventional hyper scaler earnings, we see big, impressive gains.

Hyperscaler earnings improve but actual cash flow getting worse: When we look at cash flow, though, the actual money they're taking in versus the money they're sending out, that's actually getting worse, not better.

So let's, you know, first frame that in this chart.

So this chart shows this in a different way.

It shows a price earnings ratio and a price to free cash flow ratio.

So the blue line is what you would normally see, it’s actually getting a little bit better.

That is to say even though we've seen such a remarkable run-up in the big tech company stock valuations over the last few years, their earnings have actually fully kept pace and the situation is looking a little better now than it was six months ago.

The problem is the earnings do not need to subtract the full extent of the capital expenditures on new data centres and so on.

That's just how financial accounting works.

There's nothing funny to that, but nevertheless, in actual practice, these companies are pumping money out at a remarkable rate which is why they're now borrowing and so on and so their free cash flow is getting worse and worse and worse and that's what's setting that goal line up.

And so if you look at the price to free cash flow ratio which is maybe a metric of how expensive these companies are in the stock market versus the actual money they're physically getting a hold of, you can see the values are looking very challenging indeed.

The way to frame this, then is, to say if all these investments by the companies in data centres and models pay off, this all makes sense.

It will all work out beautifully, don't worry about that gold line going up.

If though, there's regret later and they realize that the open models that are cheaper are almost as good and they end up capturing a lot of the market and so on, there could be some regret and that's where you would maybe have a problem in a broader sense.

So the market is maybe paying a bit more attention to that downside risk as it exists right now which is probably a healthy thing to do because it is a real downside risk.

Okay.

U.S. productivity growth has picked up relative to expectations: And then I guess back to the good news side, we are seeing faster productivity growth in the U.S.

We think already artificial intelligence is helping to drive productivity growth faster.

That's a big support for economic growth.

And so in a purely economic lens, this is all great.

The thing we're just all nervous about is there's a long history, whether it's the railroads way back when or whether it's the fiber optic cables in the late '90s and early 2000s of maybe over investment that ultimately benefits humanity and society and is a helpful thing and increases productivity, but maybe doesn't fully benefit the companies doing the spending as much as they thought it would.

That's what the risk is right now.

Potential AI employment implications by sector: And then let's get into this.

Full points for colour and exciting numbers here I suppose.

I can't possibly speak to all of the nuance and detail in here and, of course, you're always welcome to click pause and take a look at your own leisure, but I'll just say a few things which is we've been trying to feel out what sectors have opportunities here, what sectors there could be more labour disruption, what sectors could see profits soar.

This is not an exact science.

You will find other opinions out there.

But ultimately useful to do and so we sorted this across a few columns.

The first column is just the AI potential for the sector to the extent to which it could be relevant to the sector.

You can see software and information technology right at the top as you might imagine.

You got the sort of food type products towards the bottom where it's just less obvious immediate large-scale application.

We then have scope for sector demand growth and, so of course, linked very much to the potential for AI for that sector but also connected to the extent to which demand might rise if AI makes that product cheaper so that does vary to a significant extent.

We have the scope for sector profits to grow as well which is a very similar concept to demand growing but a little bit different as well.

And then scope for job displacement and so, you know, no one quite knows how this is going to play out and we would say we're seeing some evidence of localized job losses right now but not at the economy-wide level and it's not unusual for the natural process of creative destruction to see some job losses and other exciting opportunities created elsewhere which so far really has been the case.

But in theory, the sectors that might be not need as much work or at least workers relative to a certain unit of output, you can see towards the top and again, it skews towards services more than goods.

It skews towards you might say high knowledge industries which is very much the opposite of the kind of drain we've seen out of blue collar jobs and manufacturing in prior decades so it is very much a new challenge.

But ultimately, food for thought and something we're weighing carefully as we decide what sectors and what companies are worthy of our capital.

Okay.

New tariffs provide continuity, but fresh threats could push effective rate higher: On to tariffs here and so,  here's us trying to log the trajectory for tariffs and I wouldn't want to overstate the movement.

The big story of the last let's say six months is that the tariff rate for the U.S. went down notably in the early spring and that was the Supreme Court overturning a certain type of tariff.

The U.S. did immediately then introduce other temporary tariffs.

Section 122 tariffs.

But those didn't fully replace the prior tariffs.

What we're seeing now is - and this has just happened in the last week or two, is the temporary tariffs have expired.

They had a limited duration they could be applied.

Those have come off.

The White House is now applying other tariffs.

It thinks it's found a different existing law to do so with its section 301 and so we're seeing new tariffs brought in.  Broadly, they're in line with the temporary tariffs that are going away.

Broadly, they are still lower than the tariff rate that prevailed last winter.

So a bit of a switch but without net increasing tariffs all that much.

They are still looking at other possible tariffs to apply, though, so this could inch a little bit higher and then I guess I should say for Canada, something else has happened.

So the something else is that the U.S. has just threatened another type of tariff and threatened as of August 19th to apply a 50% tariff not to everything Canada sells to the U.S., to about 5% of Canadian exports to the U.S.

So very problematic for certain niche industries and they have been it seems targeted explicitly by the U.S. to either favour the equivalent U.S. business or perhaps exact maximum punishment, but ultimately not welcome, something that would subtract we think a couple of 10ths off of Canadian growth if it happened. We’re hoping it’s more in the threat space and Canada coming to the negotiating table may manage to help that be avoided.

We still think if we were to fast-forward a year, most like is the USMCA trade deal does manage to get extended and that the tariff environment isn’t  materially worse than it is right now for Canadians.

But as it stands now, tariffs inching higher, not lower, after the opposite having been the trend for a period of time.

Okay.

More hawkish Fed and Warsh: Over to central banks.

Let's talk about the U.S. Federal Reserve as the focal point here.

Two things – one is the gold line just conveys the Fed funds rate which had some rate cuts over the last few years, but has been on hold for a period of time.

The thinking increasingly is that the Fed probably does do a bit of rate hiking over the next six months or so and, you know, one way of gauging this is there's this wonderful natural language model, the blue line, which just looks at how, you know, interprets the very speeches and commentary from Federal Reserve officials and tries to get a sense of whether they are more inclined to raise or cut rates and hawkish comments are dominating here.

The blue line is rising so that would in theory suggest that we could be approaching rate hikes and I would say that's consistent with the most recent Fed decision which is just yesterday as I'm recording this so wouldn't be a surprise to get a rate hike at some point before the end of the year.

We don't think a lot of hiking is necessary but between inflation that is still well above target and economic growth that's holding together pretty well, that is probably the right course of action.

And then I guess turning over to the good news side, I focused on a lot of challenges so far, but as I mentioned off the top, ultimately the economy is still moving forward and we think it can.

Bank lending growth accelerating – positive macro signal: Take a look at bank lending growth.

This is European bank lending growth on the left.

It's U.S. bank lending growth on the right.

We are seeing a notable acceleration and as I mentioned earlier, that's relevant in a few ways.

It means that, first of all, customers are willing to borrow.

They think there are good opportunities out in the economy perhaps to invest in.

It means that banks are willing to lend.

It means that banks themselves are in healthy shape and they believe their own customers are in healthy shape as well.

And then just mechanically, when you lend in a fractional reserve banking system, it means that the money supply is effectively increasing.

There's more money sloshing around which, of course, is an economic support as well.

So that's a positive macro signal.

Our U.S. Beige Book Sentiment Indicator is strongest in years: I can say we look at the U.S. beige book.

This is a kind of hilariously a qualitative indicator that we then insist on quantifying which you might say defeats the purpose but we do it anyways

Gives a good sense of the general economy. You can see it's actually been accelerating recently.

Nothing special in the context of the last decade, but nevertheless, about the happiest, most optimistic level we've seen in American businesses going back a number of years.

So, again, in theory, consistent with more economic growth.

Canadian economy stabilizing?: I'll finish with this, a nod in a Canadian direction.

Canada has had its challenges and of course, we talked about the new tariffs that were threatened and that could well be implemented in a few weeks time at this juncture, but I do want at least celebrate the fact that Canada's unemployment rate does seem to have peaked over the last several months and if anything has been coming a little bit down. You might say why are there so many lines all trying to say the same thing.

This is something we're fairly proud of.

We have sort of constructed and looked very carefully at the variety of economic indicators.

This is provincial level unemployment for at least some of the major markets.

And we can say again, you know, pretty consistently across markets, we are seeing a stabilization and if anything a slight decline in unemployment in Canada. And, of course, it's so important to remember for Canada right now, the population is shrinking and so when you see perhaps Canada not create very many jobs in a month or even suffer some mild job losses, that's actually what you should expect when the population is shrinking.

The real test is what is it the unemployment rate doing and it's been doing a little bit better.

So we're hopeful that can hang on.

Okay.

I'll stop there and say thanks so much as always for following along.

Please do if you're looking for more, take a look at our website, RBCGAM.com or on Linkedin or you can use your phone to access that QR code and thanks for your time.

I wish you well your investing and please tune in again next month.