Episode 80 | Inside the Mag 7: Who is Really Spending to Win?

In Episode 80 of “Market News with Rodney Lake,” Professor Lake, director of the GW Investment Institute, delivers an analysis of the Magnificent Seven as massive AI-driven capital expenditures reshape the competitive landscape. He explains why outsized CapEx commitments, particularly from Amazon, Alphabet, and Microsoft, reflect a deliberate strategy to “play to win,” even amid short-term market pressure. Lake examines the growing infrastructure demands of AI, including power, water, and data center constraints, and explores the potential for space-based data centers as a solution by Tesla and Amazon. The episode contrasts differing approaches across major tech firms, from Apple’s platform-centric strategy to Tesla’s real-world AI ambitions and Meta’s push into wearables and advertising. Lake concludes by highlighting Nvidia as the purest AI technology play, emphasizing its central role in the global compute buildout and the enduring importance of rising demand for computing power across the AI ecosystem.

Transcript

Thank you for joining Market News with Rodney Lake. This is a regular program for the GW Investment Institute where we talk about timely market topics. I'm Rodney Lake, the director of the GW Investment Institute. Let's get started.
Welcome back to Market News with Rodney Lake. I'm your host, Rodney Lake. On today's episode, we're coming to you from Foggy Bottom, the George Washington University School of Business, right here in the heart of Foggy Bottom. Duquès Hall, Duquès family studio. This is a production by the GW Investment Institute. Educational entertainment purposes only. What are we going to talk about today?
This is going to be a general overview, as we do in class when we kick off market news with Rodney Lake. Today is sort of closing in on mid February here 2/10/2026 when we're recording this. It would come out a few days after that. However, there's a lot of things going on that we're going to be talking about and some of the things I that I want to focus on here at the top of the show, the general market news is the Mag 7.
So last year the Mag 7 had really drove the market, and this year they are not. They are actually putting downward pressure on the market for some of the companies. And especially like Amazon, for example, when they announced earnings and had a fairly steep sell off. Now I'm going to give you my hot take on these companies and what I think is going on with these companies.
So let's dive right in. So the Mag seven for the uninitiated, you’re talking Amazon, Alphabet, Meta, Microsoft, Tesla, Apple and Nvidia. So that's the mag seven, The Magnificent Seven. And so what I want to talk about first up first I want to talk about the capex budgets. And so when you saw Amazon sell off after it announced earnings and they did fairly well.
But what was the sell off from? Well people at least in the market. And you know sort of what's happening. The reaction to it's the capex budget for Amazon. So those who didn't see what the CapEx budget for Amazon was, they announced for 2026, this is not over the next ten years, by the way, for 2026, this year, the CapEx budget is $200 billion.
That is one year, $200 billion. Now, a lot of people view this as, you know, over the target, what was estimated and very concerned and so you had a fairly steep sell off, in the company after it announced earnings and after it announced this very large CapEx budget, and even ahead of what people had expected, a very large CapEx budget in any case.
But beyond that. So what's my take? The hot take for me here is you have companies that are now playing to win in the AI game, and I think you have to decide, not necessarily decide, but you have to do the work and then decide what is your take on this situation. And at least in my view, opinion.
Here remember not investment advice. Opinion here is that the companies that are playing to win and not all of them, but these companies that are playing to win. And I'll mention mentioning Amazon first here. I think it's the right move. And why do I think it's the right move? Well, I think there's going to be winners and losers for sure in this game.
Like any other game. But the payoff here is quite significant. So if you think about even some of the productivity gains that maybe people are already getting now just being more efficient, those productivity gains, you can multiply by quite a lot. So the multiplier in the multiplication effect here is going to be quite high. Excuse me. For these companies that are at the center of this. Now Amazon does not have their own frontier model.
They are an investor in Anthropic. And Claud, but they're supporting all of this. And so they have their trainium chips and they have, you know, a full tranium data center that they launched not that long ago, which we talked about on the prior episode, but they are playing to win on the AI infrastructure, which is Amazon Web Services, AWS, AI chips, robotics.
And the other piece which we have mentioned before is the Kuiper satellites. And so, that I think it's going to be super important. And I think we bring it back. Data centers in space, we talked about on this show before, lots of people in the market, we're talking about this. And if you listen to somebody like Elon Musk, who's running SpaceX and Tesla, they're talking about having data centers in space in three years and 36 months.
And so, you know, what's the reason for that? Well, if you think about the challenges for data centers right now. And so if you're here in DC and you drive out to northern Virginia and you see all the big boxes that are data centers and one car parked outside, well, some people locally are not as excited to have maybe one job.
And I'm exaggerating here a little bit for effect, but it doesn't produce a ton of jobs. It uses quite a lot of electricity and uses quite a bit of water as well. And so that's putting pressure in that, not my backyard type of situation for building new data centers. And not that they don't have good economic output for some of these communities.
But then you have to call into question, okay, well, what's the net impact? Do we think that we're going to get net plus benefits from this data center? And some people are saying that they're not sure. And so that's putting pressure on how to cite these things, and what communities to do it. And you have the Trump administration say, well, they have to figure out their own power set up for those data centers.
So that's going to help in the short term. But longer term. And I don't even mean very long term. I mean, the next three years, that pressure is probably going to get higher and higher with communities that don't necessarily want these, in their backyard. And it's gone. And the power is going to be another issue. So what does that do?
Well, it drives people to find other solutions. And so one of those solutions is data centers in space. And so how does that connect to Amazon. Well the Kuiper satellites that they're that they want to have up in space, they're low Earth orbit. They're going to have their partner Blue Origin for launch. And hopefully they can get that launch cost down.
So maybe they're going to be taking rides on spaceX in any case in the short term. But I do think it's reasonable, to consider that at the margin in three years that the pressure for the data centers will, will be, in such a place, both on the power side, the siding, the water, that they're going to end up, heading to space.
And you got to get that launch costs down. But the pressure demand is going up for those resources here on the on the ground. But as they go into space, you know, the sun shines. This gigantic fusion reactor in the sky called the sun is out all the time. There are no clouds, there's no atmosphere. And so you get much better power efficiency.
From the from the solar panels, you have to radiate the heat away, from that. But possible as long as you're not facing the sun. And so there are absolutely ways to do it. Physics, does say that it is possible. Now, the cost is the important piece. How can you do that at a lower cost?
And again, I think that that, you know, line is somewhere around three years. So Musk has put that out there. And that seems quite reasonable. 36 months to say the pressure, for the terrestrial, the Earth bound data centers is going up with demand on new power. Because even if these places all, by the way, get the ability to build these power plants, you have to build them.
And so where is that going to come from? So, that's pressure on turbines. That's pressure on transformers. The joke is, you know, you got to get Transformers to to get the Transformers to work. And so that puts a lot of pressure on Earth here. And so I think that's going to be that pressure is going to continue to mount and that supply demand imbalance is going to continue to grow.
And that's going to push, those marginal data centers, not marginal, let's say those data centers, to start being considered. Well, the economics are starting to play out where we need to get this in space. Launch costs is coming down. The reliability of the chips is going up, and there's already been some tests, by the way, for GPUs and space Nvidia chips.
And so, that's already happening. So you got some early testing going on there being resilient from the radiation. And so some of these problems are going to get solved more quickly than others. But let's say in three years I think that that's going to happen. So the Kuiper satellites for Amazon is a way to play that.
So who else is in this mix? And so we're going to move on to a few other companies in the Mag seven. We're not maybe going to cover every single mag seven company. Today. Amazon had the biggest CapEx budget. Who is up next with the next largest CapEx budget in the Mag seven. That is alphabet. Also under some pressure.
Now alphabet also playing to win. You know, saying 175 to 185 billion for the CapEx budget. And they just announced that they're going to do this 100 year bond, to help finance a lot of this build out. So that'll be interesting. I think that's worth watching. Investor. Business person analyst. You should definitely be watching what's going on there.
Quite, quite an interesting setup. 100-year bond, from Google. And so here, you know, when you talk about 175 to 185 again playing to win. Now what's the AI play for Google. Well unlike Amazon, Google has their own frontier model Gemini and Gemini 3 we've talked about on the show before is quite good. So I think it's quite interesting to pay attention to that.
So you have the other aspects of Google. You're talking Google Cloud and they're building that. That's growing. You have the custom silicon and their TPUs, the tensor processing units on version seven. And they have been doing a good job for years on that. And now really just getting credit for that. And so they have the ability to sidestep at least partially, the GPUs from Nvidia as an example.
And because they have their own custom Asics chips, called TPUs, tensor processing units to, to build their models, train their models, to run inference for Gemini 3 and again, Gemini 3.0 Very good. We'll pause here and talk about because they have a frontier model, the frontier models. This is another hot take here. Amazon again does not have their own model.
Alphabet has their own frontier model. So that's a big distinction between those two companies. Both are playing to win, in a very different way. Now, maybe Amazon at some point will come up with their own frontier model. Currently, they don't have one. Alphabet does Google. And so when you talk about this, it looks like also that these models, the ones that I think are the real frontier models.
So who are the frontier models, at least in my mind. This is Gemini, this is grok. This is ChatGPT, and this is Claude. Then you have a bunch of those are close source, then you have a bunch of open source, particularly Chinese, models like DeepSeek, which was, you know, put a lot of pressure, last year on Nvidia.
But I think this is going to continue to happen. The close models seem to be ahead, let's say, on functionality and custom, the ability to customize, for, for clients. And so but each of the models, when they come out, when you hear, Claude Sonnet 4.6 come out, it leaps ahead of Gemini 3.
And, and so, grok is doing very well in some of the benchmarks, for example. And so it looks like the models each continue to get better. Now at some point as an investor business person and analyst, you're asking yourself, well, is this becoming more commoditized? That's something that we have to think about and see, okay, what are the differences?
If each version of the frontier model that comes out is just slightly better than the version of a model that came out just earlier than them? Well, you know, does that really have somebody move from one model to the next or they're just going to continue use all these models, especially if it's an enterprise set up.
So for example, that would have an advantage then for alphabet to say, well, I'm using Gemini 3 and I use Google Workspace, for all of my stuff. And so yeah, it's maybe the, the ChatGPT, you know, newest model at the time. You know, fast forward a year or two years from now is slightly better, but I have all my stuff built into the Google Enterprise set up here, and I'm not really willing to change that for a model that's slightly better.
Or the other thing is, people are going to continue to use all of the models and try to figure out a way to integrate them. So that puts pressure on pricing for different, in different ways. But and you can have somebody else, come in. So we'll have to see how this plays out. I do think it's worth paying attention to.
I do think if it becomes more commoditized, that gives, I think, an advantage to Google in this, in this way, because it's an enterprise set up already and a lot of people use it, businesses included use it that way. And that's where Claude, that's not a publicly traded company. That's where Claude has really been focused as opposed to ChatGPT.
I've really been focused on the enterprise, in the enterprise clients, for that. So we'll have to see how that all plays out. All right. Some other companies that I think worth mentioning, before we run out of time in this show, but moving on to a few others. So Microsoft, Microsoft also announced big CapEx, over 100 billion, you know, could be closing on 150 billion.
And that's Microsoft Azure. That is the cloud, for the AI scaling. That is the copilot, which, you know, if you've used copilot, I don't I, I didn't mention it as a frontier model early. I don't think that it is. I definitely think it is behind. But they're certainly in the AI game. They have a big investment in openAI and partnership.
And so I think Microsoft is key to pay attention to also again, big on the enterprise world. And so, the cloud based setup for Microsoft Azure I think is important and obviously will continue to be important or it obviously has been important. I think it will continue to be important. Copilot this is just something that I think Microsoft is going to have to figure out, and they're playing to win to.
So they're they're putting real CapEx dollars and I think more than people expected also. And so that number, is, I think higher than, than what people expected and maybe higher than they're comfortable with. But I do think this is going to be a winners and losers game. And I definitely think that these companies if you talk about Amazon, alphabet, Microsoft, they are absolutely, playing to win.
And so we'll see how that plays out. Another mag seven here, which I think is in a very different situation. But let's talk about Apple. So if you look at Apple, you know they have pretty modest and I'd say 20 billion or less in CapEx plan for 2026. That pales in comparison. That's not a small number overall, but that pales in comparison to what, you know, all these other companies are spending.
And so even if you use the 20, you know, Amazon's a 10x, this year and I don't think they're going to even hit 20, but let's say they're 20. The rounded up the Amazon is a 10x. So that those are two different leaks in order of magnitude difference there, between the CapEx budgets, between Apple and Amazon. And this is where sometimes Apple gets the criticism that they don't have their own
AI play in place now. Tim Cook has been out and saying that, well, we're going to be, we’re the best position to deploy AI that our customers choose across our devices, the iPhone. And as an example, iPhone 17, which I have a new one, iPhone 17 Pro. And so I upgraded. And so yes, the platform still is in the billions.
And they have loyal customers. And people typically don't switch their phones. When I survey people in our classes and our students and they say and I say, hey, if we raise prices by 10%, who's switching to Android? No hands go up better than the iPhone users. And so it's fairly interesting. You definitely have that pricing power. You definitely have that networking effect.
And so they're locked in. Now maybe that thesis is going to play out for Apple. But it it's a different thesis than the game that Amazon is playing. It's a different game than alphabet is playing. It's a different game than obviously Microsoft is playing too. And so I think it's going to be interesting to see how this plays out.
So services business did really well for Apple. If you listen to the most recent earnings. And so and the iPhone 17 is doing much better. We've talked about that before. Then I think people anticipated you definitely had the the big upgrade cycle coming there. And Tim Cook is saying, hey, I do think we're best positioned to deploy AI against our network, our set up, you know, our customers are going to benefit from this, whatever this is.
And so you saw them partner officially with Google, which will be the Gemini. And so that'll be interesting to see. And so look they're taking a different strategy now. It doesn't mean the strategy will work or it won't work. I do think it's interesting. And in some cases you might have to say, look, if you didn't have a frontier model and you think you're that far behind, a different strategy is what you need to win.
You certainly don't want to compete against competitors who are way ahead of you. And you, you know, to replicate their setup probably means you just fall further behind them. And so we'll see how this goes. And that doesn't mean they won't launch something in the future. But it is interesting to see how that's playing out now. So another one here, is Tesla there.
You know, let's say that close to the $20 billion CapEx. And I think they're raising that to 60. You know, they're going after they're talking about building fabs to build their own chips. And really this AI5 chip, AI6 and all the things to follow are really the future. And that's that's autonomous driving. And that is that the cyber cab as an example.
And so we'll see. And power is a huge part of that. They're going to build a lot more solar. They have a factory up in Buffalo, New York. And are we going to be pumping out more panels. They just have a new version come out. So and then they have the, the, the batteries that are grid scale and home scale as well.
And so they're playing to win also. And so it will be interesting. They're playing in a different part that's, you know, the real world AI which is, you know, how do you get a car to move around, you know, with a passenger in it, without, you know, having them have to drive, or at least right now you can.
You're basically have an assistant with full self-driving. But they're very close to solving that if it's not solved already. And so it's going to be interesting to see how that plays out. That is real world AI application. They're still spending a significant amount of money, not quite as much as the others. But you're deploying big dollars.
They have been efficient in the past, and I definitely think that they're the leaders, in autonomous driving. You know, Waymo is part of alphabet or majority owned by alphabet. So we'll see how that plays out. Next up, we can't close the episode without talking about Nvidia here. So, before I mentioned Nvidia, what is quickly talk about meta.
So meta is spending, you know, 135 billion, for example, plus and they're focused on wearables. And you know llama is there open weight, not fully open source, open weight model. There are some limitations when you get to a certain point. But they're absolutely playing to win and playing to win, obviously in the advertising, they are building out their AI ecosystem and they, they have these wearables, the Ray-Bans, if anybody's seen them or use them and have the commercials if you've seen them, you know, meta, you know, tell me this.
So I think that's interesting. I think they're in a slightly different game, but they're playing to win in the AI game for sure, with big CapEx dollars being put to work. And so again, before we close here, let's, talk about Nvidia. So Nvidia is going to continue to spend and Nvidia keeps coming out with the next version of the chips.
And I think different than the other AI companies or AI plays in the mag seven. Here are the ways that these Mag seven companies are playing AI. What's an Nvidia? Nvidia is the purest technology company. So if you talk if you hear Jensen Huang, the CEO of Nvidia, talk about Nvidia, it's a very different set up. Right?
So the other companies make their money differently. If you pick all of these companies, you know, each one of them makes money in a different way. And some are similar. Nvidia is probably the purest play tech company. Specifically, they really sell, technology to their customers. Leading edge technology and the form of GPUs. And you hear, Jensen talk about this, that they're selling leading edge technology that are inputs, to their customers, and they really don't then in a way compete with their customers.
And not sometimes they are competing with their customers, but not always. But they're really selling technology. So I think that's an interesting set up. And I think that's something worth thinking about, about their business model. So they're really the the purest play technology company in this group. And when you think about Cuda, the network they've built around that, that's the software wrapper here that we're talking about.
They continue to spend, a lot of money there, Ruben coming out even more powerful than black. And they keep making the chips more and more powerful. And so if you think about, even if you have your own TPUs by Google as an example, or the Trainium chips at Amazon, you're probably going to want to continue purchasing these Nvidia chips.
As you know, it seems like compute is just going in one direction, which is up or demand for compute 9t's just going one direction up. Which positions Nvidia well. Now people can to place it. You have all these other things which maybe they still want the chips, but the pricing gets downward pressure. That hasn't happened yet. So we're going to be paying close attention to this.
Nvidia for the investment institute and our students is a big position in our portfolios. And so we're going to keep a close eye on what's happening at Nvidia. But again different than the other AI plays here. That's a more more excuse me pure technology play. Then I think all the rest now their network can be eroded like anyone else, but they continue to invest heavily.
The cadence once a year coming out doesn't seem to be slowing down. And Jensen Huang has, you know, 40 direct reports as the CEO, founder and co-founder, he does not seem to be slowing down as an investor. We really like that. So founder led businesses that are working super hard, we appreciate so that's the mag seven.
That's this episode for Market News with Rodney Lake. We look forward to seeing you on the next episode. Thank you. See you next time.