SpaceX’s Tesla Tie-Up, Muse Momentum, and AI’s 1% 10/21/26

0:00This message comes from Wise Business, the only account your business needs to scale globally. You're building a business with global ambitions, but hidden fees are eating into your margins. With Wise Business, you can pay overseas suppliers without exchange rate markups. Have customers worldwide get payment details in 22 currencies, so you get more of each hard-earned payment without hidden conversion fees. Be smart. Get Wise Business. Download the app. T's and C's apply. Introducing Fidelity Trader Plus. With customizable tools and charts you can access across all your devices. Try our most powerful trading platform yet at fidelity.com slash trader plus.

0:40Investing involves risk, including risk of loss. Fidelity Brokerage Services, LLC. Member NYSE SIPC. You're listening to The Exchange. Here's today's show.

0:51Thank you very much, Scott. The first big earnings event of the week is a massively positive one. And the next one looms after the bell today. I'm Kelly Evans and welcome to The Exchange. Shares of SpaceX are soaring 10 % on the nose now after strong third quarter results. And only its second report is a publicly traded company single-handedly reviving the AI trade today. The Nasdaq is leading the rest of the markets, as you can see there, popping 1.2%. We'll talk about the results, the potential Tesla merger, and if geopolitics could derail the company's momentum. And of course, we'll look ahead to Meta's results after the bell, another key test for the AI trade.

1:28The shares are down 9 % from their all-time highs last month as Musemania has faded somewhat, although you can see the pop today up 2%. Plus, some new data from Ramp. The firm tracks AI spending among companies and says we've had zero growth in token spend over the last few months. We'll talk through the takeaways as we await Anthropix mega IPO. But let's begin with the mover of the day, SpaceX, of course. Gene Munster of Deepwater Management said the results and forecasts last night were so strong, the shares should be up 30 % today. But our first guest says there's a liability risk in the stock that investors may be underestimating.

2:06Olivier Blanchard is the research director at the Futurum Group. Isn't there an Olivier Blanchard who was like the head of the IMF? Anyway, welcome to the program. Hey, Kelly. Yeah, you're right. We both have the same name, very different practices, though. That's hilarious. Okay. So you're not related. You're not his offspring.

2:27I think the shot just froze. Anyway, Olivier, if you can hear me, talk through what you think is the liability risk hidden in SpaceX shares. Well, the main one for me, I mean, aside from the IPO and the strange circumstance around how that came about is actually fairly recent. And it's focused on Starlink. And to me, it's a demand friction problem. The issue is that Starlink has to be reliable. And with any kind of new technologies, especially those that are basically just high stakes, you have to be able to trust the technology that you're invested in is going to work for you when it needs to.

3:05And my issue, and I think a lot of countries and a lot of organizations may have woken up this week realizing that Starlink access could be throttled or cut off at any moment for any reason outside of their control. And I feel that that puts an undue amount of stress on the value of Starlink and the risk. To me, it's a very high risk of investment for a lot of parties that would have the funding to really invest in it. And I think that needs to be addressed. We can show again the results kind of across whether you look at revenue, earnings, the growth in certain businesses like Starlink that you're talking about, Olivier.

3:46I mean, Starlink, I think it was up 70 percent. So that's that's a great result. But even if that growth tapered off or leveled off, and there are some of the figures, as you can see, it feels like SpaceX would still have a number of other levers to pull. I mean, a lot of people today, and I think the point Gene Munster was making, was actually much more about the compute. And even with neoclouds in question, I mean, Musk has been on X going back and forth about this, but basically saying they have the compute. They can deploy it, kind of questioning whether other neoclouds really have the same resources that they do.

4:16And Munster's point was if they deploy what he says they're going to deploy, that there should be even more upside than what he kind of suggested could be their guidance. Sure, it's possible. And there's a good story there, right? I have questions about the math and honestly the ROI of doing this. Putting data centers in space is, to me, a solution to a problem as opposed to just the primary mode of operation. Generally, there are two principles that I like to follow. One is that the closer the processing is to the data, the better. And so putting data centers in space puts the processing as far from the data as you can get it, which isn't necessarily the best way to do things.

4:58It's also more costly and more risky. Those satellites aren't going to be in orbit forever. You have to launch a tremendous amount of satellites for this to work. And just the numbers, I think, still require a little bit more scrutiny. And I'm still skeptical about it. It's not that I'm bearish about it. I just want more proof. I want more evidence. And I just need a little bit more time to really be convinced about this, especially when there are so many other options. And one of the trends that I'm tracking primarily is the expansion of AI to the edge, which I think is more important and more interesting than the expansion of AI to space.

5:37And I feel that it's a lower risk, higher reward and just more scalable option than putting data centers in space. When you say expanding to the edge, does that mean on a PC? Yeah, yeah. So it's PC, it's homes, we're seeing it in automotive. of obviously the next big inflection point is going to be robotics, not just humanoid, but also delivery robots and all kinds of other form factors. So we're seeing these two things happening simultaneously, right? We're putting data centers in space. And at the same time, we have much more high performance hardware and more efficient models being able to get smaller and smaller and be more mobile.

6:16And again, going back to my initial principle of bringing the processing to the data, as opposed to taking the data all the way to the processing in outer space. And so to me, when I look at both options, it feels like data centers in space is the least valuable, the most high risk and the lowest return of all the other options for AI that are on the table. Would you, just to kind of put a point in it, say that the quarter still impressed you? Yeah, the quarter impresses me. I'm not concerned about now. I'm concerned about three, four or five years from now. But for now, yeah, it looks solid. Yeah.

6:48All right. Olivier Blanchard, thank you. Really appreciate you making the time. We'll let you get back to the bike joining us from the Futurum group. After those strong results, we now turn our attention to Meta. It reports after the bell with the shares almost in correction mode from their September highs, down about 9%. Yusuf Scully is head of Internet and Media Equity Research at Truist. Yusuf, it's great to see you. Do you have anything that you were kind of taking away from what we learned from SpaceX first before we turn our attention to Meta? Well, first of all, thanks for having me, Kelly.

7:19So we don't cover SpaceX, so I'm not going to comment on it, other than to say, obviously, AI is all the focus right now. And that's going to be true of what's going on with Meta as well tonight. Meta is going at it, obviously, from a totally different perspective. They're focusing on consumer super intelligence, which doesn't, you know, it's only a fraction of what SpaceX does. But the net of it is there is still continued interest across the investor base. And frankly, in the case of Meta, what we're going to start seeing tonight is proof points that all these investments are starting to bear free.

8:04Do you think their launch of Muse was extremely successful? Yes, there are some questions about its stickiness and staying power. But I think more broadly, I wonder how well a consumer-oriented app can do when you look at an Anthropic, which is just, you know, entrenched already, it seems, or beginning to be really entrenched in the enterprise, an area that seems much more deep-pocketed in the long run. How would you say, how big can Meta's muse get, or do they need kind of a product for businesses as well? Yeah, so the short answer is I think can be very, very large. We published a note a couple of weeks ago where we're actually trying to answer that exact same question, which is over the next two, three, four years, how big MetaMuse can become from a variety of sources, subscription being one, T-Crader e-commerce being another, enterprise AI being the third.

8:56and we estimate that by 2030, which is around the corner, honestly, it's about three, three and a half years from now, we think Meta could be generated anywhere between$28 to$30 billion of incremental revenues and that could actually, and very high revenues as well, and that could continue to compound over time. You saw that they went from literally zero users to over 10 million, almost 11 million users within 44 days. Today, they're giving consumers a ton of compute. They're basically giving you 100 million tokens a week, which is a lot of compute to just play with it. They want people to really engage with it.

9:36I think that's the first thing. But the second is you are going to start seeing a clearer and clearer path to monetization. But make no mistake about it. The way Meta is looking at monetizing AI is very different from the way OpenAI or Anthropic is looking at it. I think there is a market for boats to thrive over time. Presumably, advertising is what they do already on their other platforms would kind of be the thing to do. For the earnings later,$5.75 a share,$63 or so billion in revenue. What are your expectations? What line items are you most focused on? The CapEx expectation, you know, I think they raised a little bit last time to 130 to 145 for the year.

10:22Their free cash flow almost went negative last quarter. But again, this is the quarter where at least we've seen the launch of Muse. So we kind of know what all that spending is for. And the CapEx figure for next year is around 196 billion. So what are you going to be most focused on? So two things. One, obviously, is the top line. We're looking at about 25 percent year on year growth in revenues. We're a little higher than the street. I think the street around 23, 24. So, you know, we think the fundamentals of the story on the advertising side remain really, really strong. On the bottom line, you're going to have a lot of noise because, remember, they just settled with 48 AGs and they're taking$10 billion.

11:01Most of that is going to actually hit from the county standpoint this quarter. But to your point earlier, CapEx is going to be front and center. I don't think they're going to necessarily quantify CapEx for 2027 yet, but they'll probably speak to it to say it's going to be, again, significantly higher, and the devil is in the details. So clearly, everybody's focused on their ability to start normalizing that CapEx spend. And unfortunately, I don't think we're really necessarily going to get a satisfactory answer yet. But in the meantime, if we're able to see that Meta continues to grow faster for longer, right, that to me will give them the right to continue to invest.

11:44Just before getting on the call with you, I kind of did a quick look back at our estimates back in 2024, what we thought they're going to do in 2027. And we thought the revenue growth was going to be mid-teens, 14%, 15%, 16%. That's where consensus was, by the way. Fast forward the tape two years later, we think this year they're going to have about 27 % growth. That built up between growth in mid-teens and growth in the mid to high 20s, that's the benefit of all these investments that they've been making. And I think that's really impressive. That's about$50 billion of incremental dollars, just to put it in perspective.

12:21And obviously, the equity investors are forward-looking, Yusuf. But is there anything as they had that big multi-state settlement and the payouts there and the legacy of their social media days? I mean, they're still very much in that business. And is that an overhang at all on the stock? I think it'll always be an overhang. And honestly, remember, Google or Alphabet had a big overhang until, you know, there wasn't. And I think here you still have a bunch of lawsuits. I think the big lawsuits with the AGs are effectively behind them. And there are some other lawsuits, particularly at the local level.

13:00We don't really have great visibility into those, but I think those are more limited in scope. But I think the stock, the way where it's trading, has already been reflecting that. But I think ultimately the stock could still work very well, even with those concerns, if they're able to continue again, put up the numbers to show that they're growing faster for longer, all driven by these AI investments, which at some point are going to normalize and start allowing free cash flow to start, you know, pivoting back up again materially. It's like the invert, you know, when Charlie Munger would always say invert.

13:32And if you said, well, where would Meta be without AI? The answer is pretty clear. And it's it's kind of a slower growing. All we're talking about are the lawsuits. I mean, that that would be the stock. And it's such a different story today with these bets. Exactly. Probably in the routines, maybe. Yeah. You know, by now, by by next year, certainly. All right, Yusuf, thanks very much. Appreciate it. Yusuf Scully with Truist Securities. Speaking of meta, our next guest called the Mag7 stupid cheap six months ago, and he's been pretty right. All of the mega caps are higher since then. Microsoft up 43 percent, Apple up 32.

14:07We'll look at how much room is left in that next. Plus, new numbers showing that 1 % of AI customers account for 80 % of all spending. Is that a benefit? Is it a risk? We'll break it all down ahead on The Exchange. This is The Exchange on CNBC.

14:48in 22 currencies so you get more of each hard-earned payment without hidden conversion fees. Be smart. Get wise business. Download the app. T's and C's apply. Weren't agents supposed to take things off your plate? Building another to handle the others is a bit much. Don't make more work out of work. Gemini is the one agent that knows all parts of your business and actually gets stuff done. You give it a project, close your laptop, and when When you come back, the job's finished. No matter your industry or role, your agent has the skills to help every team stay in their flow. Seriously, make work less work with Gemini on Google Cloud.

15:25Check response, subscription required. Introducing Fidelity Trader Plus. With customizable tools and charts you can access across all your devices. Try our most powerful trading platform yet at fidelity.com slash trader plus. Investing involves risk, including risk of loss. Fidelity Brokerage Services, LLC. Member NYSE SIPC.

15:45Welcome back. With less than two weeks to go until the midterm elections, Maine has become a flashpoint in this year's Senate toss-up race, with incumbent Senator Susan Collins trying to fend off Democrat Troy Jackson, who says she should have been doing more to protect Mainers from the high cost of tariffs and energy. Megan Casella is live in Portland, Maine, with more. Megan? Kelly, that's right. Terrace, very much in focus here in Maine because no state was as vulnerable in the recent U.S.-Canada trade war as this one, and especially its iconic lobster industry. So lobster was left off the retaliation list at the last minute, but everything else that these lobster men rely on is getting more expensive.

16:23The bait, the rope, the diesel fuel, and especially the lobster traps themselves. So Mark Brooks is a third-generation lobster trap manufacturer and supplier. He says he's selling his traps now for about$180 versus$130 just two years ago. And all of that extra money, he says, is going to the government in the form of tariffs. Take a listen. Prices have changed so much so rapidly that it's hard to quote prices. We don't know what product it's going to cost us by the time we get it here. It's a nightmare to try to do business under these conditions. Now, these tariffs are piling on top of an already high and rising cost of living.

17:06Home heating oil, for example, Maine is more dependent on that than any other state. That's up 79 percent year over year. Just this morning, though, Senator Collins announced that the Department of Energy will be releasing some 300 ,000 barrels of heating oil from the New England reserves, something that she says will directly help Mainers with their energy bills this winter. And that's exactly the kind of thing that she's been pointing to all campaign to show what she can do for her home state on the tariffs, too. Collins has been critical of the White House, and she's pushed for some key exemptions for her state.

17:36But her challenger, that's the Democrat Troy Jackson, still says that she should have been doing more. And so Maine then is becoming a key test of something that we've started to see play out nationally. And that's whether a Republican candidate can separate themselves from some unpopular pieces of the president's agenda. But here in Maine, the folks that we're talking to say they believe Collins can do it. Take a listen. She has been like a guardian angel for the lobster industry for many years. And I can't oppose her because she's in the same party as somebody I don't like. Now, Kelly Collins is the only Republican incumbent running this cycle in a state that voted against President Trump in 2024.

18:20And the polls as of now say that it's essentially tied. But I will mention that back in 2020, the last time she ran, the polls at this point had her in a very similar position. Collins went on to win that race by almost nine points. Kelly. All right, Megan, thank you very much. And again, a new announcement there from Senator Collins releasing those heating oil reserves. Megan Costello. Meantime, back to the markets. You might think that only growth stocks work in this environment, but none other than value investing poster child Berkshire Hathaway is also making new highs today. Let's bring in Morgan Stanley's Dan Skelly.

18:54He's head of Wealth Management Model Equity Portfolio. Dan, it's good to see you. And here, by the way, here's a look back year to date, S &P 500 up 13, MAG 7 up 11. I mean, Berkshire at all-time highs today. That kind of tells you there are a number of different ways you can make money in this market. Absolutely. It's great to see you again, Kelly, as well. And look, that's been the story of this year. It's been a year of rotation. It's been a year where you've gone from AI infrastructure back to the Mag7. You've gone into some of the AI adopters. The whole AI trade has gone from fear to greed and back again.

19:29And most recently, Kelly, per your Berkshire example, you've seen a flight to quality as some of the macro pressures, including higher rates and higher oil, have weighed on more average stocks. So that is a good quick recap of what you say is going on. And again, And a lot of people have been talking about this flight to quality. They say, look at health care and so forth. But flight to quality, that's what you think is going on here? Yeah, absolutely. You've seen quality metrics, be it better balance sheets, higher profitability, higher cash flow, working really in what's been an intensifying macro backdrop.

20:03We've seen higher rates, higher oil. And no doubt we've seen rising uncertainty related to the midterms coming up in a few weeks. Having said all that, stepping back, Kelly, I would just highlight the market multiple has come in all year. It's been offset by strong earnings. And I think the market multiple is appropriately and frankly already discounting some of these macro risks, but also this handoff that we're going to see in the coming year from the AI infrastructure winners, the semis, to more of the AI adopters longer term. And frankly, that's creating a lot of opportunities underneath the surface.

20:36You know, there are those who would still say the opportunity for semis is so massive. Is it too soon to get out of these trades? If we're talking about just the very tippy top of companies and those starting to deploy AI, and that's already generating this massive demand, we're short of all these components. We're going to be even more short as more adoption continues. I can see the case for still being in some of the infrastructure, the build-out names. Yeah, really great point, Kelly. And I agree. In fact, in the beginning of June, we cited risk to the semis trade because we didn't see a lot of those fundamental or technical risks discounted at the highs in late June.

21:16But now, of course, we've seen rising risk and question marks around open weight competition, safety, and of course, ahead of the next several weeks and a big event in early November, not in my backyard data center, political pushback. So I would argue with semis well off their highs and to your point earlier, still positive earnings revisions on the horizon, you have to step back and acknowledge that they're becoming a bigger and bigger part of the index. So what we've been doing selectively is buying back semis on some of these pullbacks. The AI CapEx trade is like the cat with nine lives. Every time you've tried to kill it in the last six months, 18 months, remember deep sea going back to early 25.

21:55It's been very difficult to disprove this thesis that we're going to need a lot more spending in the years to come. Well, because we're so early in adoption, and if there's any signs of adoption slowing, and that's why it'll be interesting to talk with Ramp in a little bit, but if we start hearing from companies, and there are reports out there sometimes that say they're not sure if they're getting the ROI on AI, but for people to also know and differentiate between the parts of a company that might be deploying AI to code, and the parts of a company that might be deploying AI to replace services, there's a lot of different ways to deploy it.

22:27But in those areas where it's coding and doing that support of like a CIO, it seems like it's just only beginning to make inroads. So all of that said, the application layer, what would be the names in that trade too, if you think we're starting to head in that direction as well? So all well said. On the application layer, look, this is typically the last and typically longest lasting stage of any tech cycle, be it mobile going back to 07, being via the Internet in 2000 and beyond. And look, we think the application phase could last a decade and it could surprise many people. So health care is a case in point in terms of drug discovery and revenue opportunities.

23:05Also, on the cost side related to insurance, analytics and distribution. Secondly, I would cite financials in terms of middle office, back office and more kind of operational work. And look, we've been in an environment where the blue odds have increased. And despite a positive fundamental backdrop, the financials have sold off. We think you can get back involved in the financials post the election and into next year. And AI adoption is going to be yet another tailwind at their back. All right, Dan, always great to see you. Thanks for making the time. We appreciate it. Dan Skelly of Morgan Stanley.

23:38Coming up, we've got more big names reporting results this week, including Western Digital. Speaking of the chip trade, memory names, those shares have more than doubled since January, along with the memory names. Tesla is the worst Mag 7 stock this year, and IBM is coming off its worst quarter in eight years. We'll look at how to position in all of these names ahead of their results next.

24:02This message comes from Wise Business, the only account your business needs to scale globally. You're building a business with global ambitions, but hidden fees are eating into your margins. With Wise Business, you can pay overseas suppliers without exchange rate markups. Have customers worldwide get payment details in 22 currencies so you get more of each hard-earned payment without hidden conversion fees. Be smart. Get Wise Business. Download the app. T's and C's apply. Weren't agents supposed to take things off your plate? Building another to handle the others is a bit much. Don't make more work out of work.

24:40Gemini is the one agent that knows all parts of your business and actually gets stuff done. You give it a project, close your laptop, and when you come back, the job's finished. No matter your industry or role, your agent has the skills to help every team stay in their flow. Seriously, make work less work with Gemini on Google Cloud. Check response, subscription required. This episode is brought to you by Schwab Market Update, an original podcast from Charles Schwab. Schwab Market Update, an original podcast from Charles Schwab, provides an information-packed daily market preview on stock updates, U.S.

25:13and global economic news, monetary policy decisions, and key results and statistics that may impact your trading. Download the latest episode and subscribe at schwab.com slash marketupdatepodcast or find Schwab Market Update wherever you get your podcasts. Welcome back. And we are seeing the Nasdaq lead the way today, up 1%. Dow's up 200 points. S &P about 0.6%. Russell up 0.5 % as well. Netflix shares are hitting a two-year low, though, after giving weaker-than-expected guidance last night and lowering its ad revenue outlook. They're down about 4 % today. Marks the sixth straight quarter that shares have dropped after reporting results.

25:53And Netflix is down 50 % in that time. Evercore's Mark Mahaney is saying most of these challenges are addressable, but Needham's Laura Martin is throwing in the towel, downgrading the stock to hold and saying it'll take new content chief Peter Rice at least two years to fix the company's$20 billion content slate. Elsewhere, Kroger shares are slightly higher after they unveiled a new $5 billion plan over the next three years at their investor day yesterday, trying to lower prices and improve the store experience. They also recently announced plans to cut a third of their corporate roles as they lean more into AI.

26:27CEO Greg Foran joined Squawk on the street this morning, his first TV interview since taking over in February. He didn't comment on Kroger's ongoing trial with Albertsons over that failed merger, but here's what he said about the impact of higher costs on the company. The customer out there at the moment is resilient, but probably more intentional. And, you know, they're really focused on where every dollar goes. We're starting to see a little bit of inflation starting to kick through, and that's not surprising input thing, you know, characteristics around fertilizers and diesel, those sort of things are making their way through.

27:05But actually, we're controlling that, and I'm reasonably happy with how we're managing our way through that. Showing some optimism there. Let's get to Pippa Stevens now for a CNBC News update. Pippa? Hey, Kelly. The New Mexico attorney general today announced a lawsuit against the co-executors of Jeffrey Epstein's estate to stop them from distributing money. The state AG alleged the two men helped finance and conceal Epstein's sex traffic operation, including the abuse of girls and young women at Epstein's New Mexico ranch. Both men have previously denied the allegations. FCC Chairman Brendan Carr warned broadcast networks they could violate their regulatory obligation by airing, quote, entirely fake polls and suppression polls.

27:48He did not provide any information about which polls he thinks are fake. And NFL Commissioner Roger Goodell today said the league welcomes an investigation into the way the Philadelphia Eagles handled blood costs discovered in quarterback Jalen Hertz. The Players Association asked for the probe. The NFL's chief medical examiner defended the Eagles medical staff after criticism that Hertz was allowed to play in a London game and then fly home. He's expected to miss the rest of the season after the clot traveled to his lungs. Kelly, back to you. All right. Thank you, Pippa. Coming up, the startling stats behind the AI spending binge and whether it means trouble ahead or is something of an all clear for the AI trade.

28:32We'll dig into that next.

28:37Welcome back. There's been a lot of talk about circular spending when it comes to the AI build out. And some new figures show the spending has also been very concentrated. Kate Rooney has more in today's tech check. Kate. Hey, Kelly. So we are seeing some evidence that spending in AI is coming from a small crop of companies out of Silicon Valley. That may be an underrated risk in the AI demand picture. Some new data out from Ramp highlights a group of what we're sort of describing as power users. So it is the top 1%. They account for about 80 % of total AI spending. So it's extremely top heavy.

29:12If you look at that 1 % that I mentioned, they spend around$10 ,000 a month on average on AI. They're driving demand. And then you have even the top 10%. They're spending less than a thousand bucks and then the average spender. So the median here is below 20 bucks a month. So revenue concentration is a risk. It's also showed up in different ways in the leaked S1 for Anthropic that was reported by the Wall Street Journal. There's also a lack of diversity that we're seeing in this small group of power users. So the same report does note that some of those big spenders, or most of them, I should say, 80 % are venture-backed private companies.

29:48So we're not talking about the AI labs necessarily. It's startups. So names like Replit, you have names like Sierra, Lovable, Cursor, before it was bought by SpaceX. These really are the power users. And the big picture here, if there is a crack in this Silicon Valley ecosystem out here, you could start to see a ripple effect. So all of it has bigger market It impacts demand, impacts revenue with the model makers. You have Anthropic looking to go public, which would be a big one, OpenAI as well. It's also key for the cloud providers. And then eventually, if something goes wrong, it then affects the chip companies supplying them.

30:18So that's why market participants should be paying attention here. Apollo, at the same time, pointing out the growth in spending has stalled. So that could also be a result of some price cuts, lower prices overall at the top AI labs. You can see almost no growth in the last month so far. We should note that even if the demand is narrow, the demand from what we're seeing is still there right now. SpaceX results did highlight some bullish demand in the overall picture, but it does come at a time when other parts, Kelly, of the AI trade have proven to be a bit more fragile. Corweave has been the big example, but some jitters around the AI trade, I would say, lately.

30:51All right, Kate, thank you. Kate Rooney, appreciate it. Ramp, as she mentioned, also reporting a spending slowdown, calling it significant and unusually stubborn. We're talking about token spend. Let's bring in Ara Karazian. He's RAMP's lead economist, and he's here on set with me. Welcome. Thank you. Thank you for having me. It's great to have you here. Can we just start with what do we know? Be as granular and specific as possible. First of all, about who are these 1 % users of AI? What do we know about them? Well, this is a finding that we start to see in our data set. We've been tracking AI spend now over two years.

31:22We have a business spend data set of 85 ,000-plus businesses, $275 billion of annual spend every year. And it's here that for the most part, we have observed month over month increases in AI spend. That has then fueled the AI trade as this expectation that there will be continued demand for AI, particularly driven by the top 1 % of spenders. But who, like, give me some, like, who are we talking about? That's exactly the latest crack in the AI thesis is that top 1 % of businesses is now starting to slow their spend. And they tend to be highly venture backed. They're more likely to be tech companies.

31:56They also drive 80 % of open-eyed and anthropics business revenues. Now, although most of the AI trade has been predicated on the idea that they have unlimited demand, they do so far because you look at that top 1 % and they continue to use AI more and more. We can see token volumes continue to increase month over month. And yet spend is what's starting to slow down. And that poses a risk to the model companies. Exactly. So it's a business model risk more than it is a sign that demand for the AI services is flailing. And there's also a different way you could look at this 1 % figure, which is to say, we're just getting started.

32:28Well, it's also a product issue, though, too. Now, if you look at the median firm that spends maybe$15,$16 per employee per month, a lot of people may look at that, particularly if they're bullish, and think, well, it's going to at some point end up similar to the top 10, top 1 % of businesses that spend$600 to upwards of$10 ,000 per employee per month. I think that's actually unlikely. We likely cannot, we likely will never see the median employee rise to that level of AI spend, particularly as you get outside of the tech sector. But let me ask you why, because this is so new, so early days that a lot of people are struggling to even understand what we're talking about.

33:02Not the people in your world, they know. I'm talking about the broad public, lots of companies out there who have maybe experimented a little bit with this, but it feels like the true demand for this is a little bit more in the future than it is even right here today. I mean, wouldn't you agree? Don't you think that this is still like the dawn of the adoption of this technology more so than the end of its adoption cycle? Well, it depends. I mean, one of the main findings about AI today is that its spend ends up being highly concentrated amongst a small group of businesses. Now, those businesses do end up driving outsized returns from AI's usage.

33:35We can see that they tend to be faster growing. They tend to hire more after adoption. We found that in a paper last year. At the same time, they are highly concentrated in certain sectors. They tend to be VC backed. And there is still no evidence to suggest that the gains from AI will be able to proliferate in as much quantities outside of the tech sector. All I'm saying is it makes sense to me that the people using this the most are people who are in the computer world. Right. It makes sense to me that it's not the Midwest or the South or even, you know, the traditional financial services industry that would be the early adopters of this.

34:10It would make sense to me that it's tech-heavy firms, the kinds of things that would be venture-backed, the kinds of things that would be, right? I mean, that's where people understand and can quickly implement this new technology, I would think. Yes, but that then has then driven this growth in the AI trade. But then what happens when that top 1 % starts to turn down its spend? Now, it doesn't mean that the AI trade is over. There are still areas for growth, including at the median, including at greater proliferation across the Fortune 500. But if that top 1 % starts to slow its spend and they end up not only being a highly concentrated source of spend for the AI trade, but also a highly correlated one, you end up having a potential risk of cascading effects that affect other parts of the AI trade, including compute.

34:51Absolutely. Can we talk harnesses for a second? Can you explain to everybody what a harness is? And also one of the things I think you guys really uniquely do well is to show, you know, is AI a commodity or not? People often switch between all of these different platforms, not necessarily because they're perfect substitutes for each other, but because one's a little bit better at coding and one's a little bit better at reading PDFs and one's a little bit. And so what do we know about the stickiness with any given platforms like OpenAI, Anthropic and the like? AI has ended up developing very differently from other software paradigms that we've observed, especially in business spend, in that as firms become more advanced users of AI, they tend to distribute their spend across multiple providers.

35:35They also then end up getting some market power that allows them to negotiate better rates with OpenAnthoropic. They also end up offering more model providers to their end employees that then end up doing their own switching or implementing routers to get more cost optimization for it. That's an unusual structure for this kind of market because it means advanced users actually end up being quite picky. They don't lock into a specific vendor. Great for the actual business that's using AI, but a new risk for the model companies that believe that there could have been outsized gains that were everlasting from that top 1 % of businesses, which actually ends up being quite choosy.

36:12Sure. And finally, just any moment now, any week now, we're going to get Anthropics IPO. and where do you see this all going a year from now, especially if token prices keep falling? Are they going to come up with other levers and ways to monetize their business? Well, I think they're going to have to. One of the main problems with AI today is that yes, there are outsized gains to have from it. Yes, businesses do gain from it, but they end up being in these highly concentrated, highly correlated sectors. So the question facing the AI companies is that product issue. Can they develop AI into something commercially advanced that also delivers outside returns outside of the tech and software sector.

36:50There are early signs that they'll be able to do that, but it's not a given. All right. And again, you guys catalog all of this so well. Ara, thanks for making the time to join us. Thank you for having me. Ara Karazian of Ramp. Quick note, Ramp also ranks fifth on CNBC's 2026 Disruptor 50 list. Coming up, near-term options in Western Dig imply a nearly 10 % move in either direction on earnings. Tesla has beaten revenue estimates in 12 of the last 20 quarters, kind of a coin flip, while IBM has missed on earnings estimates only once. We'll get the trades on all three ahead of their upcoming reports.

37:23Earnings exchange is next. Season continues and we'll hear from Tesla and IBM after the bell today and Western Digital tomorrow. For more on how to position ahead of those results, let's bring in our trader, Jeff Kilberg, KKM founder and portfolio manager and a CNBC contributor. Jeff, it's great to see you. So much to talk about. So let's dive in with Tesla, which is perfect. It's all the news already today. After Elon Musk floated the idea of a potential merger with SpaceX last night, Musk saying both boards have created specific committees to discuss what would be the largest merger in history.

37:56If that weren't enough, analysts will also be listening tonight for any hints on the future of that flying roadster, the latest robo-taxi numbers. What do you do with it here? Well, I own Tesla. I think you have to get excited. And if you just go back to, I watched that Roadster unveil live. That was remarkable. It seemed like it was AI, I guess we're supposed to call it SI now. But nonetheless, what we are seeing, and I don't know if I could actually use the Roadster. It would be kind of cool to do some flips around the neighborhood. But nonetheless, the fact that we have that proof of concept, that we have this kind of new normal in Tesla, I think there's opportunity.

38:30And if you think, what is Tesla right now, this car company? And I know they're talking about deliveries being about 486 ,000, probably a little bit more than anticipated. but it's just a data input net. It's a distribution mechanism for all this artificial intelligence or super intelligence, we should call it. At the end of the day, I think Tesla is going to continue to move higher, but it's interesting, Kelly, to your point, after that SpaceX, which SpaceX kind of set the table last night for everything AI. We're seeing the QQQs up today about 100 % because of the enthusiasm that maybe this AI train is back on.

39:00But I think the SpaceX, Tesla, they'll stay connected hip to hip, and I think they're going to march higher. All right. So that one being kind of easy for you. Yes. I guess the only remaining question I have, though, about SpaceX and Tesla, Kilberg, is. Are both shares getting upside from the idea of combining and yet what actual value is created by combined in the near term and maybe fine in the long run? Are we pulling forward a lot here? I mean, Tesla's still down year to date. So it's been kind of sideways for the last year is trying to price in all this cap backspend. And so I see your point, but I don't think it's fully priced in yet.

39:37I know Brad Gerstner was talking earlier about what would that buyout rate potentially be in that's closer to$5.50 or even$5.50. So I don't think it's priced in because there's so much uncertainty. And typically, as much as I support, and I'm long, Tesla and SpaceX, his timelines, Elon's timelines are typically not on time. I guess, look, and I'll get off this point, but it feels like if you're just taking money from one pocket and putting it in another, right? The higher you pay for Tesla, it feels like the more that should ding SpaceX. The lower you pay for Tesla, the more that should benefit SpaceX.

40:08So I don't know why SpaceX is benefiting from the idea of paying more for Tesla when the money's coming out of their pocket. Well, I could see how it's juxtaposed, but it's a$2.5 trillion company now in SpaceX, and it's$1.5 trillion in Tesla. And I think the two combine. There's a ton of synergies. You talk about TeraFab. You know, this kind of goes back to the conversation we've had quite some time on Intel, right? When all of a sudden, Terafab stepped into Intel. And everything is, no pun intended here, but everything seems to be tagged long to this rocket ship of SpaceX. So I'm long Tesla. I'm optimistic.

40:38It has not been for the faint-hearted. You saw a dip just back in July under$300,$298. Didn't feel great then. So yes, we are in a low euphoria right now. But we have to see how today's call goes. All right. Let's move on. Western Digital. Speaking of ups and downs, the shares have more than doubled since Jan 1. But they're down 53 % from the all-time highs. Bank of America sees upside ahead, raised price target to$800 from$720 on strong hard disk drive pricing. They are flagging some risks about their digital storage shipments, though. B of A and Wedbush keeping an eye on profit margins and growth.

41:10What are you looking for and what do you do with Western Digital? Well, you want to talk about beta. Western Digital is a high-flying, high-beta stock. And if you think about just back in June, that Ford P.E., Kelly, was trading at 75 times. That has been hair-cutted more than the drawdown of dropping 50%. That 4PE now is 19 times. So there's opportunity. We don't own it right now. As you know, in our Mango Growth ETF, Gary, we're very tactical, but we don't own it right now. But it is interesting to see. About 80 % of all data centers still have their data on hard drives. So there's still a need.

41:43They're rolling out a 44 terabyte in 2027. I think it's going to 100 terabyte hard drive in 2030. So I think there's opportunity. But when you see oversold conditions like you just did see in Western Digital, RSI went down to 30. It's had a little bounce, but I don't know if it's a dead cat bounce or it's an opportunity to really buy. But we're going to find out tomorrow on earnings. And we'll speak with the CEO, Irving Tan. He'll join us for an exclusive interview after that report on Friday. Really looking forward to that. Quick one on IBM, Kilburg, before you go, down 17 % this year. We're expecting, I guess, UBS is in line.

42:15The company cut its software outlook, warns that supply chain tightness across servers, storage, memory, is forced to ship in CapEx in the second quarter, didn't improve in the third. You like this one, right? It is growth at a discount. And we actually own IBM in our Essential 40 ETF. We think it's a critically essential name. A word that we don't say anymore, Kelly, it's called quantum. And I think the quantum computing opportunity with IBM, I think you can buy it at a discount here. So, yes, we're going to really be looking forward and hanging out every word. But when you talk about that delayed, when they talk about that delay, we're going to see how their business is actually doing.

42:49But I want to own IBM here. And if you don't own it, I think there's a great opportunity when it's 25 % from its all-time high. Now's the time to own Big Blue. Jeff Kilburg, thank you. Appreciate it. See you, Kelly. Good to see you of Kilburg, KKM Financial. Coming up, home builders on pace for their third negative session in four amid elevated mortgage rates. We'll get the latest read on home buyer demand after this.

43:15Welcome back. September's housing data showing, I don't know how to describe this, a deep freeze, if not worse, I guess. Pending home sales fell to a 25-year low as mortgage rates sit near 7.5%. Diane Olick is here on set with me. We've got mortgage applications today. Any glimmers where no rates are down? No. Okay. Well, not surprising. In fact, the hope was, okay, we saw last week's average on the 30-year fix actually drop back for the week. That's a weekly average. We also have daily averages, which move differently. But last week they fell. Did homebuyers get in there for more demand? No, they did not.

43:47In fact, demand dropped again and is down close to 16 % year over year. And we might have thought it'd be more responsive, but I guess it shows that when rates are this high, it just takes more than a couple basis points of a drop to get people going again. Yeah, and it's not just this kind of emotional feeling about, okay, rates are over 7%, whether it's 7.5 or 7.38, which I think they hit today. But it's more, a lot of people don't qualify at this level. It's not just about paying that monthly payment, but qualifying by the lenders, who, as we know, are much more strict since the Great Recession and the financial crisis.

44:17But it's also home prices are just not giving any help at all. They continue to be higher year over year, and that's because we still have this supply issue not enough. We were talking to the Compass CEO about that the other day, and a few different people who are saying they're starting to see more prices being cut. Do you think that that's going to get things to move a little bit, more price cuts, or the inventory that's sitting there is starting to come down? Well, so there is more inventory on the market now. We're getting into the season where you see more inventory, And there's more inventory simply because homes are sitting longer on the market.

44:48So that builds it up. But overall, we're still short several million homes for the demand that's out there right now. Now, are some sellers cutting prices a little bit? Yes, some are. Are there more deals to be made? It depends on what market you're in. All real estate is local. We know that. But a lot of buyers are just sitting there saying, maybe I'll wait. They're saying, maybe rates will come down. Or they're saying, maybe I just can't afford saving for the down payment because of how much I'm paying for everything else in the economy. Also, we asked about this back and forth. that we've seen amongst the realtors lately about these private listings, which Compass specializes in.

45:19And some are saying this is unfair of him. So we asked him about these. They're engaged in some legal and industry battles with the multiple listing service. Take a quick listen to what the CEO told us the other day. They actually do have an important role of how you can share your listings with the brokerage community. It is an important marketing channel, but it's not the only marketing channel. And it's definitely not a channel that should tell you, dictate how you market outside the channel. Right now, they fine agents up to$15 ,000 if they market outside them less. And because of that complete control, they've been able to take the agents' listings, put all these negative insights on it, sell the data to over 900 financial institutions, including 20 federal agencies, 300 capital markets companies, 400 property managers, and 900 mortgage companies and banks.

46:07And they get kickbacks from these entities. So the home sellers listings are getting monetized in countless ways by the MLSs and real estate portals. And all we're giving the home seller is choice. Yeah. So anyway, we're up against the wall here, but I wonder if there's going to be more to come. I didn't realize what a battle there was behind the scenes. So there have been several suits. Three of them have already been settled or just have stopped because they've come to agreements. You have the one in California that's still up. Again, it's just about buyer choice, which is what he's arguing, or whether everybody or seller choice or whether buyers get to see everything that's out there when it gets out.

46:39Yes, there's so much change in the housing market, even on this piece of things in recent years. Diana, thanks. Glad to have you here. Diana Olick. And that's it for us. Thank you for watching The Exchange. I'll join Brian Sullivan for Power Lunch right after this quick break.

46:54you've been listening to the exchange make sure you're subscribed to get each episode every day same time same place introducing fidelity trader plus with customizable tools and charts you can access across all your devices try our most powerful trading platform yet at fidelity.com slash trader plus investing involves risk including risk of loss fidelity brokerage services llc member NYSE SIPC.