Top of the Morning: The AI Economy 4Q26 update - Part 1
The desk interprets the UBS podcast update on AI's impact on the US economy as a significant indicator of upcoming market trends, emphasizing a shift driven by increased productivity and investment in AI technologies. Per the full note , Paul Hsiao outlines a 'flywheel' effect where advancements in AI lead to heightened demand and subsequent supply increases, potentially reshaping macroeconomic conditions. As this environment develops, institutional traders should be mindful of the implications for currency volatility, particularly in pairs sensitive to growth forecasts. Crucially, staying abreast of shifts in productivity metrics will be vital as markets trend toward year-end evaluations.
What the desk is arguing
The desk believes that the insights from UBS underscore a transformative phase in the economy influenced by AI, which could redefine market dynamics. Investment sentiment is expected to align with increasing productivity, translating to a more bullish outlook on sectors poised to benefit from AI integration.
Paul Hsiao mentions that this interrelated enhancement of supply and demand for AI capabilities creates a cyclical momentum that financial markets will need to navigate carefully. A notable factor here is the upcoming demand for GPUs, as AI applications proliferate, which stands to alter investment patterns significantly.
Where it sits in our coverage
Our current consensus target is 1.075 for the currency pair, with expectations for potential fluctuations within a range of 1.04 to 1.12. Aligned with our view, jpmorgan is targeting 1.10 for March 2026, while bofa holds a more conservative endpoint at 1.04 for the same tenor.
This imminent shift aligns the desk's outlook with jpmorgan, though there lies potential for divergence as we assess broader market responses against bofa's cautious position.
How other firms see it
Aligning firms, such as jpmorgan, predict that increased productivity tied to AI will bolster economic growth. Conversely, firms like bofa maintain a more reserved outlook, anticipating potential risks from economic deceleration.
In this context, watch the dynamics around growth-sensitive currency pairs, especially USD/JPY, as they are likely to reflect the broader economic alterations prompted by AI advancements. The intersection of these trends with central bank policies will further elucidate the market's direction.
What the calendar says
As of now, there are no imminent high-impact calendar events that directly intersect with these developments; traders should remain vigilant for any unexpected signals from macroeconomic data releases that could shift outlooks.
How firms align with this view
Aligned with the desk view
Contrary positioning
Key takeaways
- 01The UBS commentary highlights a cyclical 'flywheel' effect of AI on productivity and investment.
- 02Increased demand for AI infrastructure, such as GPUs, signals potential shifts in market dynamics.
- 03Traders should focus on growth metrics as indicators of currency volatility and investment trends.
- 04Current consensus positions show varied expectations, with **jpmorgan** being more optimistic compared to **bofa**.
Market implications
Traders should closely observe productivity metrics and market reactions as AI technologies gain traction. A deviation from expectations in growth signals may materialize around levels near 1.075, with any significant changes possibly prompting volatility.
Risks to this view
Should economic indicators fall short of expectations, signaling slower growth or productivity declines, it could reverse the positive sentiment surrounding AI investments. A pivot from central banks towards more dovish policies in response to such data may also present risks to the current bullish outlook.
Hi everyone, Dan Cassidy here. Welcome back to Top of the Morning on the UBS Market Moves podcast channel. Joining me for today's episode, glad to welcome back Paul Hsiao, Senior Asset Allocation Strategist from the UBS Chief Investment Office within UBS FSI.
Paul is one of the lead authors of the AI Economy, A Roadmap series. The UBS Chief Investment Office did release a quarterly update to the publication series that addresses 12 of the most pertinent questions heard from clients and investors relating to AI. On today's episode, which will be part one of a two-part series, Paul will preview some of the more market-focused questions with a following podcast set to be released next week that will feature U.S. economist Andrew Dabinsky providing an assessment of AI's impact on the economy.
So, Paul, to begin, can you walk our listeners through this quarterly update? What's new within the presentation? What we did was we updated on a quarterly basis what we wrote earlier this year, the AI Economy and Roadmap.
That was a larger presentation, really trying to framework our thinking about how to think about different parts of the AI stack, and we're really trying to break it down to two parts. The first part is what is the AI flywheel doing, how productivity, demand, and investment are all related. The more advanced things you can do with AI, the more demand that they're able to also prompt more investment into it, which also leads into, with more supply of GPUs and the amount of tokens that are available, you can do more things with the programs you have.
So it is somewhat circular in that front. And then with all that, how does it affect the macroeconomy, which our senior economists will delve more into. So we tried to make this presentation as user-friendly as possible, splitting it into four parts, really looking to the AI flywheel, some of the market and macro considerations, and then going to the risks of the AI trade going into 2027.
So we try to answer the 12 most pertinent questions that we're hearing from our clients and from investors going into 2027. In an environment of higher rates, how have CapEx intentions, Paul, changed for major hyperscalers? What have you picked up on?
I think when people look back in 2026, one of the major market news developments this year is really how rates have moved up higher, both on the short end as well as the long end. And typically, that tends to restrict economic activity with higher interest rates. The cost of borrowing is higher, and people expect a higher amount of return, and that tends to restrict activity.
The difference between this cycle and last cycle, with everything talking about AI, is that it's really supply-driven. So much less of it is dependent on interest rates than before, and that's why a lot of the hyperscalers have this ability to announce hundreds of billions, leading up to a trillion dollars per annum spent on CapEx going into 2030. So to answer your question, hyperscaler CapEx intentions were revised significantly higher this year.
I think market consensus was about $500 to $600 billion for 2026, and year-to-date, it's really looking more on track to a $700 to $800 billion pace in 2026, and then going into 2028 and 2029, we expect that to accelerate to $1 trillion per year. But one thing that we will keep in mind is that the delta of CapEx growth, both on a dollars basis and on a percentage basis, these hyperscalers, I think one of the reasons why we used to call them the MAG-7 was one thing that was magnificent about them was the amount of cash that they have on hand. We thought that they would use their free cash flow as well as cash on hand to really help finance the data center build-out.
A lot of it has taken place so far, and looking to projections, we do see free cash flow in total for some of the major hyperscalers dipping into negative territories this year, staying in negative territory in 2027 before going back into positive territory in 2028. So what that means is they can't just rely on the free cash flow to finance more and more investment needs, so they do have to look at either the equity market or the bond market. I think a lot has been said about hyperscaler issuance in the bond market right now.
I think in general, the market has digested it pretty well. The big question is, will this really become something that's monetized in 2028 and stabilize those free cash flows? How long will rates stay high?
Because if rates continue to stay high, then the amount of interest into the hyperscaler issuance tends to be higher, so that will be a little more challenging for some of the hyperscalers to really raise the funding for that. So that's something we're watching out for as a risk going into 2027 and 2028. Okay, now what trade worked or didn't work back in Q3?
Right. So I'd say in general, investors still remain concentrated in AI, but we did see a pretty significant rotation from semis into hyperscalers and software. I think part of the reason is, going into this year, semis was already a crowded trade, and part of it, like power, sold off because of regulatory concerns such as Texas pause in data center demand, and that happened, I believe, in August.
So that part of the AI trade seemed to be dented. That all said, hyperscaler demand, MAG7, improved going into Q3. I think partially seen as a book called safe haven in the AI trade, just given its revenues as well as high cash balance that I mentioned before.
At the same time, the software, which has been rather unloved by equity markets at the start of the year, started to pick up. So instead of the AI trade really fading, we're seeing more of a rotation from a crowded part of the trade to a more defensive part, if you can call it that, in the AI sector. That's all for equities.
And really, in the credit market, bond spreads remain rage-bound and demand from AI-linked paper remain robust. So we're not seeing that sort of repricing of risk that I mentioned in earlier questions. Bond markets still think it's a pretty—bond market hasn't changed its opinions on the credibility of these hyperscalers in Q3 compared to Q2.
As we look ahead, Paul, we're recording here in mid-October. Looking ahead to 2027, which will be here before we know it, what should investors be looking for as we head into the next year? Yeah.
So I think this really comes back to what we think is the AI flywheel, right? The productivity part, what can these agents do? Can this actually be a visible return on investment capital for a lot of these companies that justifies that capex?
Then we go to demand, and we kind of split demand into two parts. Overall demand, and we're kind of measuring that with token usage and proxy, but also with the introduction of open models and more and more sophisticated Chinese AI models, how much of that demand is really for U.S.-based AI platforms versus open source and Chinese, which can be much cheaper for a lot of businesses, because that's another question that I think equity markets are deciding on. The third thing is how sophisticated these models can be.
So we've had the, I think, broader adoption of agentic models this year. Some of the hindrances for businesses to adopt them is just how expensive tokens are. But since token prices are coming down, going to next year, can these AI agents really start becoming much more productive or replacing part of the labor market ahead of schedule?
So how sophisticated these models can be. And then finally, I think the underpriced list in the market really is security, and that comes into two parts. I think over the summer, we've heard some concerns that open AI and Gemini, we've had some of these models being able to take advantage of some vulnerabilities in some large institutions.
So the question on the regulatory side is, will U.S. start having a more robust federal framework around AI regulation when it specifically comes to security? And then on the equity front, and especially in the software space, are there companies right now in cybersecurity that see the potential challenges that AI has and are able to design a defense that other companies can use to mitigate the risks of AI exposing their digital vulnerabilities? So I think those are some of the risks that investors are watching out for going to 2027.
I realize that a lot of what I've talked about today is much more focused on the market aspect about AI, but this presentation is really split on what is the market impact and as well as what is the macro impact, and I leave a lot of those macro questions to our senior economist, Andrew Dubinsky, who I think will also speak upon this in a later podcast. Thank you, Paul. Again, today we have been joined by Paul Hsiao, Senior Asset Allocation Strategist for the Americas.
From the UBS Chief Investment Office to you, our listeners and our clients, please take a look at the Q4 update to CIO's report series, The AI Economy, and be sure to join us next week here on Top of the Morning for Part Two, where we will look at AI's effect on the economy with U.S. economist Andrew Dubinsky from CIO. Until then, from UBS Studios here in New York, I'm Dan Cassidy. Thank you for joining us.
Thank you for tuning in. Be sure to visit ubs.com slash studios to view the entire UBS Studios suite of podcast channels along with our video offerings, such as UBS Trending. You can also follow us on Instagram for content highlights at UBS Trending.
UBS Studios is part of the UBS Chief Investment Office within UBS Global Wealth Management. Visit ubs.com slash CIO to view the latest research. This material has no regard to the specific investment objectives, financial situation, or particular needs of any specific recipient and is published for informational purposes only.
For information, please visit our website at ubs.com slash workingwithus. For a full legal disclaimer applicable to the independent investment views produced by UBS, please visit our website at ubs.com slash CIO-Disclaimer. Thank you for listening.
Sources & References
How we cover this story