Bloomberg Audio Studios, Podcasts, radio News.
This is the Bloomberg Surveillance Podcast. I'm Jonathan Ferrow, along with Lisa Bromwitz and Amrie Hordern. Join us each day for insight from the best in markets, economics, and geopolitics from our global headquarters in New York City. We are live on Bloomberg Television weekday mornings from six to nine
am Eastern. Subscribe to the podcast on Apple, Spotify or anywhere else you listen, and as always on the Bloomberg Terminal and the Bloomberg Business Apple's turned back to Deep Seek's market impact of the last twenty four hours, losses extending beyond AI and ship giants as global investors reassess the sustainability of seeming the endless spending in the sector AI infrastructure companies including Constellation Energy and Vistra taking a
hit during yesterday's market term or for more on there. We're joined by Sharp Pureza of Guggenheimshah. Welcome to the program, sir, I just want to understand from your perspect if Hagen reacted to the news of not just yesterday, but of the past week or so, and whether it upends your thesis for twenty twenty five.
No thanks for having me on, but to be honest, we believe yesterday's reaction in the news over the weekend with certainly a knee jerk reaction in the utility and power space. One of the things we've been highlighting to investors and clients is that this is some of the strongest power demand environment we have seen in over twenty years.
Of covering the space.
The US is essentially decoupling from China. In our view, and we don't quite agree with our competitors. Ten to thirteen percent of power demand today is coming.
From AI on average.
That's nationally, So everywhere you look, the demands for energy and power continue to grow.
It's a supercycle. It's structural.
So whether you're looking at the domestic manufacturing space, whether you're looking at the onshoring and reshoring of the manufacturing sector, the electrification of the system. Trump's big focus is on energy policy, LNG, peture chemicals, chloralkali. Anywhere you look this, even this energy revival is going to push power demand maturely higher. Industrial growth is extremely healthy and that's the key driver of infrastructure needs. So we believe yesterday's sell off was well overdone Sorr.
Within your points there, I think there is an interesting nugget, which something is something that people have been talking about, which if you had the build out and the pace of the buildout of AI with the current capacity for the next five years, there is no way that the
energy infrastructure, even with additions, could support that. Can you give us a sense of just even if you did reduce the energy demands significantly going forward of AI adoption, how you still would have to increase dramatically some of the energy production in the United States and globally.
Yeah, I mean, I think that's a very fair point you're bringing up. US infrastructure has not kept up with the demand. When you sort of think about electricity consumption for the past fifteen years, it's been a decelerating command with energy efficiency and demand side management, and then that of nowhere, there's this energy revival, there's this industrial revival, and the infrastructure in the US is not kept up
with the demand needs. And this is why many of the utilities, the power companies, the ones that we cover, are building a massive amount of transmission system, They're building generation. We've seen the retirement of coal assets get delayed because of this very.
Type power supply demand environment.
So long and short of it is, infrastructure has not kept up with the demands, and we do not believe the news around deep seek is going to impact any of those fundamentals.
That said s aar I guess one of the potential headwinds to the entire energy sector has just said. Commodities are actually getting cheaper. When it comes to oil prices, Sa've been going down. People are getting more efficient at producing a lot of energy. How do you dovetail that into your optimism at a time where we see the actual commodity price is coming down for a whole host of different reasons.
Well, the commodity coming down is necessarily not a bad thing for the construms, correct, So, like you know, one of the arguments is to meet the needs of US infrastructure, small modular reactors likely not a viable option, large scale nuclear likely not a viable option. You're not building any more coal plants, offshore winds having a lot of hinderances.
So really, what's going to satisfy the needs in the.
Near term as a gridge to another technology is natural gas. So the more the commodity curves come down, the more it becomes economical to build the infrastructure that's needed to meet the power demands. So in our viewpoint, the drop in the commodity, the drop in natural gas, which is deconsolidated from utility spending to begin with, is a very big driver for spending needs and it's very pro consumer friendly.
Is there a line in the stand though, where it's good for the consumer, but it still makes sense for companies to go out create new utilities to make sure that they can drill and get these products to market without losing money.
I think you know the barriers to entry to get into the utility industry.
It's one of the most capital intensive.
Industries there is, so there is a barrier to entry to get in there. I think the point we're just trying to bring up is talking to many of these hyperscalers that you kind of mentioned talking to the utilities, there's a need for speed to market, right because, as Microsoft told us specifically on a client call, we're in a race to beat China, We're in a race to beat the three to four hyperscalers. We are not going to pull back spending, right.
So the point is, is the amount of spending that is.
Happening on US infrastructure, whether it's AI, whether it's on the energy space, whether it's industrial manufacturing, et cetera, has to be met with a generation source, right, and so there's a lot of money being put into it. I think natural gas is clearly the fuel, the bridge fuel to meet those needs, which is why there's so much investment going into the natural gas space. This is why one of the largest renewable develop lupers in the country, next to our Energy is now did a ge ge
deal to build gas. This is why Constellation Energy, the biggest nuclear operator, acquired Colpine, the biggest gas operator.
In the country.
I think the shifts from decarbonization to resiliency and reliability are starting to take shape because I agree with you that as you're looking at the demand outlook, the infrastructure has not been able to keep up with it.
Sure how much excitement is driven by politics, because of course Trump just last week lifted the permitting freeze that we had on LNG applications that were put under the Biden administration.
Yeah, politics is certainly part of it.
I think under the Trump administration, we have a viewpoint that it's really good for energy, right, so when you're thinking about carbon rules for natural gas, when you're thinking about pipelines, when you're thinking about the LNG markets. This clearly going to be a push to drill, drill, driu and the focus is becoming energy sort of independent and so part of this is but to be honest, like even before the Trump administration took the HELM, our viewpoint is the demand.
Picture wasn't going away. There's such a big focus on.
Shifting manufacturing back to the US, There's such a big focus around electrified electrification of the system.
All that is very much pulls on demand. And that is before even the Trump administration took the HELM.
Now, post Trump, it sort of adds a lot of additional drivers to it, which is like what you've highlighted, there's going to be a major ease around the energy infrastructure, business, permitting, et cetera. Again, all this that you're highlighting is going to push power demand growth higher.
It's going to prompt.
These utilities to build the infrastructure to meet the needs.
I shall I appreciate your input and your opinion. Thank you, sir Champe, Resident of Gegenheim. So here's the lightest investors assessing USAI dominance after Chinese startup Deep Seeks newest AI model, triggered a sell off in global equities. Max Ketner of HSBC saying yesterday was not a game changer, writing when we look at the S and P five hundred breadth, it has been quite positive. The equal weighted index went
up yesterday, not down. We might also see more broadening of the market, not only of the index, but also of the AI theme to enable us and users. Max joined us now for more Max, Welcome to the show, sir. I want to pick up on that last point. We've talked a lot over the last year about chips, about data centers, about energy providers. Do we need to talk a little bit more about who will benefit the adopters? Is that where you see the trade guming now?
Yeah?
I think so.
I think when we look at the last year, or really a year and a half, there was a lot of questions, mark question marks on the AI theme, and we talked to clients, for example, around the monetization of AI. Who can really make money of that?
Well?
Number one we have already and we are already seeing, even on an index level, we're already starting to see I think some product activity gains, some efficiency gains. When we look at cross sectors in the S and P, there's a bunch of sectors already you know, very very clearly deploying AI and clearly having some beneficial sort of
beneficial factors from that. But most importantly, I think we've always had sort of this feeling, Okay, it's really in video or maybe two three behemoth that really make money off this, but how we're going to make money with
this on a broader scale. So I don't quite get the reaction from yesterday when people say, oh my goodness, it puts the whole years exceptionalism case really under threat because we've been moaning about it for really more than a year that you know, this is not a broad based story, and actually across all the names, it's highly
questionable whether anyone can make money of that. Whereas what you're saying now is with that if that really comes to fruition that it is much much cheaper, which again is actually still remains a heavy, heavy question mark behind that.
But if that's a theme, then this is a game changer actually all sort of the broader sectors to put AI to a test into practice and really really get some some gains from that, so it should then shift more to the AI and ablus that clearly have been lagging the trade over the last twelve to fifteen months.
MAXI sound very optimistic, which is as I would expect considering that you've been taking a good dose of Goldilocks on steroids. I am wondering, though you talk about it might be a game changer, but it's not the game changer, and certainly not in a negative way with respect to tech. Tariff's not a game changer immigration. The real game changer
is disinflation. How do you separate out this idea of disinflation with growth from these other ideas that are whip swing markets, whether it's tech advancement or whether it's tariffs.
Look, I think on the tech side it really remains or on the AI side specifically, I think it really remains to be seen whether that is that big game changer. If anything, I think that's probably going to be something that we will see panning out in the next few years, not instantaneously in the next three six months. The same thing is on tariffs. We don't know how that will pan out, if it comes, how it may come. The same thing on immigration on any kind of really big,
big immigration flows. Well, number one, we've already seen a peak in immigration flows in the second quarter last year,
and ever since it's clearly gone down. So that and the inflationary impact on that, whether that's inflationary, and how much that's going to be, I think that is much more a story for age two, possibly only for age one twenty twenty six, but certainly not for the next three to six months, because, Lisa, what you were saying, the story now is the disinflationary impact from things like super co inflation, from the underlying inflation measure from the FED,
and some of those insurance components that were sort of artificially inflating those inflation baskets at the beginning of last year that now perhaps not repeating at least to the same magnitude like we've had it in the first quarter twenty twenty four and first quarter twenty twenty three. I think that's the much bigger story, particularly from a timing perspective. So to your question, how do you separate them? I think people are sort of confusing the timelines a bit.
What we're discussing about Tariff's immigration, you know, possibly life.
The President is talking about potentially tariffs as soon as this Saturday, and if not a review by April first, do not take him seriously or literally.
No.
I think a lot of it might be tactics, right. We've seen that possibly also last Sunday with Columbia, for example, So I don't think it. I think that all let's put it away. I think the number one mistake that we can make, both on a portfolio basis, but or even on a just a trade by trade basis, is
to trade off every single headline, day by day. Right. Otherwise, you know, Sunday evening we would have said, oh, really bad on em EM currency is really bad latam, and then twelve hours later you would have said, actually, scrap that,
Let's go the entirely opposite direction. I think you're really risking then flip flopping around in terms of your views and really trade off every single headline rav then really and trade the bigger picture, the underlying inflation and growth picture, which again to me really really is sort of outweighing those kind of headline risks that we're facing in the next three four months.
A Max got to hear from your Max Canda of HSBC the one stop to watch and video recovering after its worst day since March twenty twenty. The stop wiping out almost six hundred billion dollars in market cap, the biggest single day drop ever for a US company. The move coming after Deep Sik's cheap AI model fueled concern of a US tach dominance.
So Jim read Over at Deutsche Bank compared the drop in market value as being more than the total market cap of Exxonmobile and Master Cards. You can basically put most companies into that bucket that the one day drop in valuation is a commensurate or superior surpassing in some capacity. What does this say? Number one, it says how high elevation,
how high valuations had gotten. And b it talks about consensus trades, which is the reason why you're hearing a lot about bubbles in two thousand like behavior and all sorts of studies. I expect the history onics to maybe die down, but there has been a mentality shift and it seems like it may not go away so quickly.
The President wank in yesterday as wow. Donald Trump, addressing concerns about Deep Sak killing the Chinese startups progress in AI, a positive Trump posts are delivering a warning send the companies. Apparent breakthrough should be a wake up call for US companies.
He basically wants US companies to be laser focused on this. He says we should be out in front of China. He even said Chinese officials have told him himself that the US has the best scientist. Also weighing in on this was the new AI cryptos are within the administration. That's David Sachs. He went on Twitter and he was talking about how Trump was right to rescind the Biden
executive order, which he says hamstrung American AI companies. Without asking whether China would do the same, He said, obviously not. And I'm confident the US and in the US and we cannot be complacent. This goes back to Trump's push last week ahead of all this with stargate. But how much more money and how much more effort is really going to go towards it.
We'll catch up with Bernstein later. They put out a great note yesterday. They said we believe that one deep Seak did not build Open Ai for five million dollars. Two the models look fantastic, but we don't think they are miracles. And three the resulting Twitter verse panic over the weekend seems overblown.
Basically, it doesn't change the whole model, and that's the reason why you see the little bit of a balance today. Nonetheless, there has been a sort of denial of the shift away from just investing in a lot of chips to the broadening out. This may open up that discussion a little bit more.
You'll notice how towhere some dollar strength just want to squeeze in this story. Here's the why. President Trump also saying he wants to enact across the board tariffs that are quote much bigger than two point five percent. His comment's coming after The Financial Times reported Treasury Secretary scomp Person is backing universal tariffs on US imports starting at two point five percent and then maybe rising over time.
The main event, though stocks are raising some of yesterday's losses after China, these AI startup deep Seak sparked a one trillion dollar routing the US tech stocks. Matt Bryson of Webbush writing, the AI race is still more about being the first to create models that realize greater value from AI versus lowering cost and software efficiency yields historically have led to more hardware spending and not less because the ROI of investing in hardware only increases. Madam pleased
to say, join us right now for more. Matt, Welcome to the program. I shared a quote from Nvidia a little bit earlier this morning. I'd like to read it again and get your thoughts on it. When they say that deep Seek's work illustrates how new models can be created leveraging widely available models and compute that is fully export control compliant. Do you believe what you have seen is fully export control compliant?
I mean, it's really hard for me sitting here in Boston to say what resources Deep Seek used. Obviously, we've all seen the speculation that they had process there is the were an expert compliant, that they were using processors that they had before the export restrictions went into place, and that the cost of their model was a whole lot more than five million dollars.
That's said mattter. Is there sort of a takeaway here that we really have changed the mentality and maybe companies won't be as investing as much in capital expenditures to build out AI because there is an example of how things could be done more efficiently going forward.
So I still stick by what I wrote, which is the question is or the thought process is AI changes how technology works. And so if you think AI is going to enable a great new technology like autonomous driving, it's not that it's too expensive right now, that it doesn't have the capabilities. And so the investment you're putting into your data centers is to enable that technology. And you want to be first thatechnology because we've seen time and time again when you have leaders in technology, they
tend to stay leaders. Microsoft, Google, Amazon, you can go down the list that said.
Mark Bennioff of Salesforce had a great quote, this is kind of classic in our industry. The pioneers are not the ones who end up being the victors. Matt, how much are we learning that some of the biggest gainers so far on the AI craze are not necessarily going to be the big gainers? Was this the Cisco moment for the likes of Nvidia?
So the question then, I think is did we overbuild?
So you look back to two.
Thousand, did we simply build too much infrastructure like we did back in two thousand and we'll never use it? I guess I would be surprised I think that the differences in two thousand we created the backbone. Nineteen nine nine, nineteen ninety eight, we've created the backbone the internet. The Internet was working at that point. Again, with AI, we still haven't really realized what the benefits of AI are.
So I think until we see, you know, here's a great new chatbot, or here's a great new personal assistant, and I'm buying a new iPhone because it's changing my life with this personal assistant, I think until then, it's really hard to say that we've invested enough.
Well, where do you think we are now in the cycle? Is there a chance that it's over invested.
I think there's a chance that that next technology takes longer to show up. So we get to the end of this year and there isn't a breakthrough in autonomous driving, There isn't a breakthrough and a personal assistant, and so companies look and say, it is that goal that we've been shooting for, is it really achievable the way that we're going about it? And people rethink that. And that's been the concern with Nvidio last year heading into this year,
that at some point you can't justify the investment. But I don't think this changes that goal or the amount that people are willing to invest get to that goal, because again, if you create a tonments driving the returns are amazing. It's kind of if you became dominant search market back in the early two thousand and two thousand and one time frame.
Well, you talk about the fact that in this intermediate term, if it is more efficient AI, like we're seeing from deep Seek that on the margins we could see increase adoption. What does that mean? What does that look like?
Yeah, so if you think about AI right, part of it is part of the question is the expense. So I think about it in search for for instance, I think people have said a search on using AI something like ten x a traditional Google search. So if all of a sudden I can bring the pricing down by ninety percent, then you know, maybe I moved my infrastructure from traditional search to two AI search and then moving forward beyond that. So I create this this this new app.
Lets let's say it's again a personal assistant on my phone. If if it's easier, if it costs less to UH support that app from a hardware perspective on my phone, then they'll be quicker adoption internal leads to you know, a faster refresh cycle. More seven conductor is getting sold. And so that's why I'd say in the intermediate term lowering costs. We've seen it again and again server virtualization for instance, it leads to faster adoption.
I just want to squeeze this in. This does not raise questions about valuations as some of these firms though, do we have to sort of readdress that just a little bit, even if you can be optimistic about the future.
I mean, I think the valuation question it is always is always a fair question, and certainly markets have particularly recently fluctuted wildly. But no, again, I think the value the value of these firm is what we might be able to change in the future dot not how much it costs to get there.
Matt, I appreciate your time and your impus, sir. Thank you, Matt Brice and there of webbers. This is the Bloomberg Surveillance podcast, bringing you the best in markets, economics, and geopolitics. You can watch the show live on Bloomberg TV weekday mornings from six am to nine am Eastern. Subscribe to the podcast on Apple, Spotify, or anywhere else you listen, and, as always on the Bloomberg Terminal and the Bloomberg Business app.