Bloomberg Surveillance TV: October 15th, 2026

Bloomberg Audio Studios.

Podcasts Radio News.

This is the Bloomberg Surveillance Podcast. I'm Jonathan Farrow, along with Lisa Abramowitz and Anne-Marie 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 6 to 9 a.m. 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. We begin this out with renewed tech jitters dragging stocks lower. Sebastian Page of T. Rowe Price writing, traditional valuation says the market is in a bubble.

I disagree. Seb joins us now for more. Seb, welcome to the program. Got to start with Corweave and a bit of distress in the credit market. What's the signal coming from that name?

Well, there is going to be volatility in the AI trade because it is more levered than it was. But fundamentally, I am optimistic for the next six to 12 months on that trade. And Jonathan, I look at end demand. Token spend is up 700% year over year, quarterly token spend. And token demand, because they're getting cheaper, is up 18x.

So we're only scratching the surface in terms of adoption and accelerating demand. The demand is accelerating. There's data that estimates that about 20% of enterprises have at least one agentic process in their enterprise, 20%. consultants are estimating that we'll get to 80%. So my signal from this is there's volatility, there's more leverage, there's more speculation, there's winners, there are losers, but fundamentally, the demand is still pulling us forward on that trend.

Seb, to your point, that demand of the application layer should flow downstream to demand for the infrastructure build-out. The infrastructure build-out, though, is done so much ahead of time to meet the demand of the application layer. When we look at the neocloud firms, the leveraged players like CoreWeave, if they face capital constraints, are you confident that demand will come from elsewhere to fill the gap for the chips trade?

John, you mean the supply of capital for financing the build-out? I think there's still capital available. And if you look at some of the more solid companies in the AI build-out on the cloud side, for example, their corporate spread is not that high, and it is actually comparable to, in the case of Microsoft and Alphabet, lower than the average IG investment.

Let me jump in just to clarify. That's a different bucket. There's two buckets I'm talking about here. On the one side, you've got a pool of demand, and that's legit, hyperscaler, investment-grade, well-funded balance sheets, that demand. A lot of people might say that's a largely rate incentive. They're going to keep on buying. They're going to keep on buying. You've got another pool of demand, not as large, and then you've got the leverage players, the likes of CoreWeave. I'm trying to understand... If the likes of Corbyn face capital constraints, can't borrow anymore, are facing those constraints and can't build out in the same way, is the demand from the well-funded players sufficient to fill the gap left by them

if indeed they get in trouble?

I think so. I mean, you have a big shift, not a big shift, but an incremental shift from the frontier models to open weights. And when you tally all the demand, it keeps growing. And yes, the build out has happened earlier and now we're monetizing it. I remember, John, we talked about this a while back, but one of our investors said when everybody was asking if AI was going to be monetized or not. We didn't see the monetization.

He compared it to basically the crack cocaine phase. This was maybe a year, year and a half ago. He basically said, everybody's getting it for free. You get hooked on it, and then we charge you later. And believe me, in enterprises, it's only beginning. So, yes, John, there will be fragile players. Some of them will get in trouble.

But I look at it from a top-down perspective, and I find the trade could continue for the next six to 12 months. I just think this is so transformational. And we do have still capital markets that are behaving. I think, you know, I just look at the end demand, and that's That's how I think about it.

So would you be buying the chip players off the back of a six-day run, a move of more than double digits, double digits, about 11% right now over the last six sessions or so? That's a correction in the chip names in a bit more than a week. So would you lean into that weakness?

Yeah, selectively, you know, we have 300 analysts on our platform and they're playing the AI trade very selectively. They're looking at all the bottlenecks one after the other, right? Chips, memory, so there might be chip makers that they might like, memory that they might like. something called packaging, where you put it all together, all the way to the construction phase, the energy supply, everything around this trade. You really need active management and selectively pick your spots. Winners and losers, I think, in all these phases of the AI trade, John.

Are you surprised, though, that TSMC's upbeat outlook, the fact that Anthropic told prospective investors that its third quarter revenue hit close to $ 17 billion, that that didn't stop what we're seeing in the technology stocks?

You know, it's interesting. Our team went to Taiwan and came back and, you know, they reported that there's still a lot of demand for the basics, the components, the beginning of the chain of the AI trade for the chips. And, you know, when I look at Anthropic, it's a great example, Anne-Marie, because like a year ago, let's just remember, they had about $ 5 billion in annualized revenues. And if they're at 70 now and talking about 100 by the end of the year, that's just a massive illustration of the money that's being spent to implement AI of the demand itself.

Think about it from $ 5 billion to $ 70 billion. I think like Nike and Coke's revenues are like in the $ 40, $ 50 billion range. So they've created that much revenue. in one year and that shows real end user demand and we all know it because we're all end users nowadays, especially in enterprises but also in our own lives. We see the impact this is having and how the models are getting incrementally better and the age intake phase of this is quite powerful. You know, I don't want to sound like an uber bull, like pounding the table. It's never going to go down. And all the players are all going to be collectively successful. There's definitely overinvestments in many spots of that trade. And there's more leverage than

there was. And, you know, incrementally, the demand acceleration might accelerate a bit less, you know. But overall, I'm looking at what I would describe as a healthy trade for the next six to 12 months.

You're a boomer, not a doomer, but how much is basically this market and this trade reliant on Anthropic going public before the end.

Of the year?

Yeah, that's a big risk. You know, it's a big risk in terms of the sentiment around the trade. And again, this creates volatility. This is not for the faint of heart. It does create volatility, winners and losers. And we hope the IPO is going to go well. And the expectations are very high. But overall, again, I keep going back to the end demand. The end demand is there. Now, will it be Entropic? Will it be OpenAI? Will it be another player on the frontier? The horse race at the frontier tends to shuffle between the players. The end product is very helpful for automating processes. And basically, it's quite transformational. I think we've talked a lot and maybe enough about this. But so... You know, this is the bellwether, though,

for sure. And, you know, Claude is really transformational for a lot of enterprises. For example, we're using it. We announced a partnership with Entropic here at Tiro.

Seb, just to wrap things up, if this was 8.30 yesterday morning, we'd be talking about the bond market selling off. That's not the story this morning. Guilds are down across the curve. We've just had a 25 basis point range. at the front end of the curve, off the back of hotter than expected CPI, pushing yields up to about 495.

They've come down to about 470 as of this morning, Seb. What do you think is happening in the bond market? And do you think, are you willing to call a bond market bottom and a top in yields?

Well, yesterday was interesting because you just talked about this, but the 10-year reacted to the inflation news. So the 10-year went up really quickly following the higher-than-expected inflation. But then it became a growth scare, quote-unquote, or just concerns about the AI trade. And then before the end of the day, what was driving the 10-year shifted towards growth or risk-off concerns. And that's the thing about rates, Jonathan.

I'll just end on this. There are so many factors driving rates. The famous book by Agatha Christie— it' s a really old book, but the punchline, A Murder of the Orient Express, is that there are 12 suspects. And you read the entire book trying to figure out who the murderer is, but they're all guilty. So when you look at rates, you have inflation, you have growth, you have issuance, you have deficits, you have international rates and oil prices. But the bottom line is these factors, right? They all generally, as a trend, pointing in the same direction of higher rates.

We have closed about half our short duration position a few weeks ago. And we're looking at closing more, but we're not sure yet.

Stay with us. More Bloomberg surveillance coming up after this. Concerns over the AI debt binge hitting Corweave, a sell-off in the corporate bonds triggering a sharp drop in the stock. The company's CEO calling the market reaction irrational. Ted Morrison of Baird writing, the financing of the superintelligence CapEx trade is now under scrutiny. Total AI debt issuance is projected to grow to about 25% of US dollar non-financial investment grade bond issuance. Ted joins us now for more. Ted, good morning.

Good morning.

Maybe the most dangerous phrase of the last month or so that has been repeatedly used mentioned on this program is that this build-out is rate insensitive. Are we about to find out just how rate insensitive or rate sensitive this might be?

It's not rate sensitive. The hyperscalers can charge anything they want because demand is way outstripping demand. So, I think they can show a return on invested capital on anything they want. They're fine. The hyperscalers are. But the amount of debt put on, we do an extensive list on free cash flow. And I'm seeing clients basically go to anything with strong free cash flow. The names that don't have free cash flow, like Oracle or NBIS or CoreWeave in this area, area is just being thrown out at this point.

When those leveraged names get into trouble, and we're seeing them get into trouble in the credit market this morning, the company will have its own opinion on what's happening here, but the market certainly is pricing these companies differently. How do you think about those companies as a source of demand for some of the chip stocks that got in a bit of trouble as well in the last week?

I think it's related, but your previous guest was basically saying how strong the cycle is. It is strong, but we're going to have these potholes along the way, and one of these is financing. My clients, they hate the bank of AMD, the bank of Nvidia, and the bank of Broadcom. They hate it. They hate special purpose vehicles.

What do they hate about it?

Well, they just don't like the implied guarantee. It's a soft guarantee. But the other thing they don't like is these special purpose vehicles that are pushing debt off the balance sheet of some of these companies and moving it to another entity. The demand for debt is so high that the equity market window also has to stay open.

And in the last three weeks, it hasn't. We've lost five really high-profile IPOs that have postponed. And I think what has to happen to fund this debt, both windows have to open up.

How much has the anthropic IPO and open AI IPO being pushed back been an issue as well?

It's an issue, but I think everybody has kind of digested all the safety issues that both companies are addressing. I think where there is a rub is this mix between frontier models and open source models. This is a big deal. NVIDIA is three years ahead of it, three, four years. Jensen's three, four years ahead on on infrastructure and software, what he's doing with Nemotron, what he's doing with Palantir. There's a rumored investment on NVIDIA from Revelation.

These are the three guys who popped out of Google DeepMind. They have a new open source model called Beam. There is a movement to open source, so lower token costs.

You think there's a risk that the application layer becomes commoditized?

That's a concern. It's a race to zero.

That's a problem, isn't it?

Well, we'll see. You know, everything's moving so quickly that I don't think anybody knows, to be honest with you, but it's on everybody's radar screen.

If you're Microsoft or you're Google and you're looking at what's happening to some of these neocloud firms in the last 24 hours, Are you feeling good about things?

I'm clicking my heels. You know, if you look at Google, they raised, what, $ 85 billion?

They got through the.

Window, you know, before anybody got through the window. They're sitting on a lot of liquidity. So they're in great shape. If you look at Microsoft, I think they're going to have a blowout number. I mean, I think Copilot is, Nadella has done a great job on Copilot, and Azure is off to the races. I mean, 40 plus percent.

And their free cash flowing, I don't think they can spend their free cash flow. Normally on a quarterly basis, Microsoft derives about $ 25 billion of free cash flow. That's a lot. So, you know, if you look at Google and you look at Microsoft, they're in the catbird seat on free cash flow.

So the individuals that were calling you, you said you walked in this morning and you were on the phone very late last night. It wasn't about those companies. What is the angst on the other side of the phone?

The angst is attribution performance. We're at a bad time of the year.

Because of politics?

No, well, if you're a PM, you're trying to protect your performance. Everybody's been jammed into this infrastructure play. I mean, you know, at the peak, semiconductors were up 87%. That was only two or three weeks ago. Now they're up only 65%. You've got to look at a five-year chart here. And the attribution that these PMs are seeing, it's hurting their year-end performance. And if you look at, they're worried about their shareholders, but I think they're also worried about their year-end compensation. So, everybody's kind of, the rats are jumping off the ship, so to speak.

So, what do you think? Chase it into year-end or, as I've joked a few times in the last month, twos and snooze? You know, sit at the front of the yield curve and park it there.

Yeah, I gave Anne-Marie a term that I think is really appropriate. You don't catch a falling knife until it hits the ground and stops spinning. That's the key theme. And I think one of the names you should watch is Micron. One of the best balance sheets now in tech, $ 33 billion of free cash flow per quarter. They're raising prices.

It's down and out. And I think you've got to watch when the stock bases and when you see an uptick. Nobody knows when that happens.

Where are we in terms of the trajectory of the falling knife?

I would say maybe the third or fourth inning of a baseball game.

Yikes.

How many innings are there in a baseball game, just to be clear?

Nine.

Okay, just to clear that up, you know, just in case people weren't aware.

And it could happen very quickly. It's going to happen very quickly.

You're expecting a sharp move lower from here?

Yeah, I just say, you know, everybody thinks it's going to take weeks. I think if you look at the market and liquidity, we've gone from over-exuberance, speculation, and greed. Now we usually go to concern. We're kind of in the fear phase right now. If we get to the panic phase, it's going to happen real quick.

Will some companies go bust because of this?

No, no, I don't think so. The macro is way, way too strong. It's just, it's price discovery.

Stay with us. More Bloomberg Surveillance coming up after this. Here's the latest this morning. Sources telling us here at Bloomberg that President Donald Trump's super intelligence task force is expected to hold its inaugural meeting today. The White House facing mounting pressure to take action on AI with less than three weeks until the midterms. Libby Cantrell of PIMCO writes the following. The biggest risk to the AI build out is from the states, not D.C. We believe there will be more friction ahead in AI investment in many places.

Libby joins us now for more. Libby, good morning.

Good morning.

There's a big spread right now. Main Street wants regulation. Wall Street doesn't think it's coming. Who wins?

I mean, this is sort of a David and Goliath story, honestly. And I have to think that David may win. You're just seeing so much antipathy for data centers in particular. I think the data centers are sort of a manifestation of all these other concerns, affordability, concerns about big tech, corporate concentration, safety. And they're kind of voting with their feet, so to speak. And so what you're seeing at the municipal level, at the very local level, is that a lot of these voters, even though they are being promised the world, are not sort of voting for or not approving these data centers. And I think you're going to see even more of that come 2027. The reason I say that is because there are 36 governor races that

are happening every in November. Lots of folks are, of course, you know, they're focused on Congress, they're focused on what the White House is doing, but this is really a state story. And so you're going to see many governors who have run on this anti-data center platform. And by the way, a lot of them are Republican. I mean, this is the interesting thing. This is actually much more to do with if you're an incumbent versus anti-incumbent than actually partisan lines. So many of these folks will get elected and they're going to have to do something on this.

And the easiest thing that they can do politically is they just push pause, right? They impose a statewide moratorium, a little bit like what has happened here in New York, and sort of to give time for maybe the sort of the fever to cool a little bit for the state legislatures to provide a framework. But I don't think the market anticipates this. I mean, if you look at all of these assumptions around the AI infrastructure build-out, it's hockey-sticking. And, you know, maybe that does come to fruition, but I'm not sure it's going to come to fruition here in the United States, just given this sort of political sentiment.

It's not just voting for your state representatives or your governor. AI is actually on ballots.

Yes.

There's real measures. Walk us through what you're looking out for.

Yeah. So, I mean, if you, it's fascinating, honestly. And I don't, I don't think any, you know, in my kind of 25 years of hanging around the policy hoop, have I ever seen an issue go from kind of hero to zero in such a very short period of time, or at least neutral to zero. I mean, people, again, have such, you know, sort of disgust almost, such animosity for this, these pull worse than nuclear reactors. So, you know, in Ohio, Ohio is ground zero. So that I would look at sort of a bellwether, many actual referendum on the ballot. And again, they have a governor's race where you have both candidates who are advocating for a statewide moratorium.

I think it's the statewide moratorium that's probably the most pernicious just because. particularly in these states where there is a lot of planned capacity, like Ohio, like Georgia, that just can put kind of a chill on this. So again, we're not trying to be doomsdayish. We are offered a lot of these deals, right? And so we're just being much more selective. And again, I think the politics have just become a premium that you have to sort of price in here, just given, again, just given the- So what.

Does it say that the president's basically all in on this? I mean, he literally had the top tech CEOs for an impromptu press conference in the driveway just a few weeks ago.

Yeah, I think part of the problem here is that there really is just a, was sort of a latent distress going into this of big tech in general, right? I mean, this is, these are the same folks who, you know, told you that your kids would be safe on Instagram. And so I think, and, you know, there's been just over the last decade, you know, I think that people are becoming more worried about just their, you know, addiction to digital devices and what have you. The initial conditions here were bad, and then now we're supposed to trust them on data centers or AI or the average voter.

I think one of the issues here in the United States is there's really no natural evangelizer who's trusted, a trusted source on this. There's nobody on the other side who folks think are credible, and I think that's part of the gap. It's interesting, Annemarie, you've looked at this. If you look at the polling in the United States relative to the rest of the world, The U.S., we are more skeptical of AI. We want to kind of shut it down more than Europe, right? I mean, this is sort of interesting given that, you know, we're usually sort of the culture of innovation and the culture of risk-taking, and you're just not seeing that. So, again, I think the long way of saying is, you know, oftentimes politics

leads policy, and I just don't think the market is pricing in what could be just a lot more friction in this build-out, and then obviously it could affect valuation.

And we've talked about capital constraints, energy constraints. Political constraints is just not something that people are really putting a lot of weight on. They talk about it, but again, there's a massive spread. The people want one thing and Wall Street expects another. 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 6am to 9am Eastern. Subscribe to the podcast on Apple, Spotify or anywhere else you listen. And as always, on the Bloomberg Terminal and the Bloomberg Business Hour.

Bloomberg Surveillance TV: October 15th, 2026

Kapitelklicken springt direkt hin

Shownotes

Transkriptvom Podcast bereitgestellt · Zeitmarken springen hin