Alphabet 是一家拥有持久护城河和强劲增长的卓越企业，但由于资本密集度上升以及 AI 回报的不确定性，其估值如今比一年前更接近合理水平。

About 18 months ago, Shawn pitched Alphabet as the cheapest of the Mag 7 back when the market was busy pricing in the death of search. But since that episode, the stock has roughly doubled and the death of search never actually showed up.

What did show up as a company that now plans to spend around $200 billion in a single year on data centers, has stopped buying back its own stock entirely, and has raised over hundred billion in capital to fund this spending.

So, the business got better and the balance sheet got a lot stranger over time. So, I'm really excited about this one as Alphabet is actually the second largest position inside of the intrinsic value portfolio with a waiting of around 14%.

Now, Alphabet has always been a business that interested me simply because I just use so many of its products daily. And as a user, I just don't really actually see much of a reason to switch to anything else.

So, on a daily basis, you know, I'm using things like Gmail, I'm using their search engine, their browser, their web-based office applications, and even Google Home.

And yet, I've never actually owned any Google shares. So, it's a bit of a strange conundrum. But what really struck me when I was prepping for this episode is that the alphabet that you added to the portfolio 18 months ago is actually evolved pretty dramatically in that time frame.

And I don't just mean the stock price. The business has made choices that I just didn't really expect, at least in terms of scale. I won't give you too much of a hard time, Kyle, for for not owning Google.

But, you know, the beautiful part of Google in my view is that it's not going anywhere. I mean, this is a company that is a real tight end and that's to say in 5 to 10 more years, I think Alphabet, which is the parent company name that we'll use interchangeably today for Google, will have a much higher intrinsic value than it does today.

But I've also got to know some of your preferences as an investor, Kyle. And that is to say, you typically are not one to invest in large caps, certainly not mega caps. So, is it fair to say that if maybe we imagined a scaled down version of Google with, I don't know, maybe a $10 billion market cap instead of this being this $4 trillion behemoth, you probably would own shares in it. Is that fair?

as Google has gone up nearly 100% since you first added it to the intrinsic value portfolio I think that it's just a simply a really really good example that you can still make multibagger like returns from businesses with multi-t trillion dollar market caps

and Google is an interesting one because when you look at the top of the tech titans in the S&P 500 it's usually a business that seems to be the most reasonably priced maybe you know Meta has had periods where it's a little bit cheaper or similarly priced But Google always seems to have a multiple that I think is not necessarily cheap, but also just not that expensive when you consider the quality and growth of the business.

So, I'd say the thing that surprised me the most when digging into Alphabet was how their capital allocations changed over time. With them set now to spend somewhere around $200 billion on data center just this year, I really wanted to better understand these investments and how are these investments going to impact their impressive capital efficiency numbers going forward.

Yeah, I think most people are are pretty familiar with Alphabet, but just to to get everybody on the same page, the bulk of the thesis was built upon the fact that Alphabet is a collection of worldclass businesses with literally billions of users.

Not exaggerating. And whether you're looking at Google search, YouTube, Google Cloud, or Android, you were and still are looking at some really incredible assets with very high standalone value.

And so YouTube, for example, was purchased for $1.6 billion many, many years ago. And now it generates something like $36 billion in annual revenue. And then you get to the venture capital portfolio in their moonshot bets.

And Google Ventures manages over $10 billion in assets with roughly 400 active portfolio companies. And so that includes, you know, the less popular Google Glass that came out of Moonshot Labs, but also their crown jewel, Whimo.

And at the time when I pitched Alphabet last February, the market was pricing Google as if search was the only business segment that mattered and search was in trouble because of the incoming threat of AI.

And I think to be fair, this was really truly the first major threat to Google's search monopoly in probably more than a decade and maybe ever. And that's a business that we have referred to in the past as perhaps even one of the best in the history of capitalism.

And yet chat g I think created this genuine uncertainty around what the future of search would look like, especially back in 2024 and early 2025. I would say things are much foggier and in the time since it has become clearer how AI can actually drive more search volume on Google and that things like AI overviews and search results could be monetized similarly to traditional search, but we'll probably get into that over the course of the episode more.

So the general consensus was that investors were generally kind of fearful because it was no longer clear that people would use Google search engine to answer their questions anymore.

I personally was a very heavy chat GPT user back then and I actually cut my Google usage very very significantly. And so looking at it through the lens of the business of Google that kind of concerned me as you know I just didn't really feel like I would ever really have the need to go back to Google the way that I used to use it.

But a few months ago, I realized there was a bit of a shift in how I was using some of these AI search engines as well as Google. So I basically noticed that I was going back more and more towards using Google compared to now I'm more using Claude compared to ChatgBT.

And with that difference being when I would go to use Google or through Gemini, if I just wanted a very very quick factual answer to a question, I would get that answer nearly instantly and I would also get a bunch of sources that came with that answer.

So, I feel like when I use Google, at least to some degree, again, I don't want to get too much in the weeds of how I'm using this, but it makes it easier, simpler, and sometimes a lot faster just to get these answers.

So, you know, I think the market and even me back then was thinking that search was dead, but to me, at least today, it clearly isn't. I remember actually being a bit of a Google search truther back in 2024 telling folks that chat GBT was obviously going to disrupt Google and then actually my thinking came full circle on that and I I think it's continued to be my my belief that you know you you had this narrative that AI would kill search but it's just really the opposite is what has manifested and I think that is what gave us this really special opportunity in the first place to be able to invest in Alphabet, as big of a company as any in the world, at a discount to the broader S&P 500 index.

When we first looked at Alphabet, it was trading at around 17 times earnings. And, you know, a year before, investors have been paying 30 times earnings for shares in Alphabet.

And the question to me was whether the core search business faced a serious enough threat of disruption within the context of Alphabet's already pretty diversified business model to justify a rerating of the valuation multiple that dramatically.

Right? Whenever you do have fundamental increases in uncertainty, the valuation multiple should decrease because the earnings quality has declined. The future is maybe less predictable than it seemed.

But again, you're sort of trying to weigh that against your assessment of reality. And you know, I again, I felt like the market had seen its sentiment swing a little too dramatically toward pessimism about Alphabet overall and in particular the search business.

And interestingly, Alphabet today again trades at 17 times earnings. But I think the circumstances are very different. We've seen the stock double since the last time it traded at 17 times earnings.

And I don't think it's because of concerns on competition, but really more about the uncertainty around the returns on what their massive AI investments will yield.

That is really the question of, you know, what degree will they be able to justify this spending. And on that point, one of the more tangible impacts of AI is that as Alphabet invests to integrate its LLM, Gemini, across its suite of products, this may help to defend their market share, but due to the cost of AI compute, which lots of news articles talking about how expensive that is.

And AI just being more expensive than traditional Google search, the company could actually become structurally less profitable as they integrate AI more and more into the business.

And so I do think that is to me a risk that I take more seriously and definitely more seriously than the idea of just everyone abandoning Google search.

But in reality since I looked at Alphabet about a year and a half ago for the first time really seriously margins have definitely improved across the board except for one area and that's in free cash flow.

And the reason so as you know Kyle is that that is a metric that reflects operating cash flow minus capital expenditures where capital expenditures are investments in the future of the business.

So given all the ongoing commitments to constructing and leasing space at data centers, it's really not a surprise to see free cash flow being dramatically lower and and actually swinging negative for some of the hyperscalers and or at least projected to in the coming years.

I think it was Bank of America that had this really incredible chart that went viral on finit showing sort of this profound shift in markets where the hyperscalers like the Alphabets and Amazons of the world.

They were collectively seeing their free cash flow turn negative while semiconductor companies that are designing and manufacturing the chips powering this AI revolution. They're the beneficiaries and their cash flows have correspondingly skyrocketed.

So the capital intensity is scaling up. That'll be a theme in today's episode, I think. And you know, much more capital is being required for Alphabet to maintain its business than even was just the case 18 months ago.

And it remains to be seen whether that spending is being done defensively, which would be less positive for shareholders as they try to prevent seeding ground to competitors, or if it's more offensive, where they're looking to capture new markets and new verticals that will unlock an even longer runway for Alphabet to keep growing profitably, which is sort of mind-boggling to think about for a company with a $4 trillion market cap.

Yeah. And you know, when I first kind of looked at this, I was actually really really surprised at just how well Alphabet's margins have held up during this kind of entire expansion phase.

But another major theme for investors with Alphabet has been this kind of regulatory environment. With Google being quite clearly, I would say, a monopoly in many, many different areas.

It's a business that's just basically under constant regulatory scrutiny.

The hardest part about evaluating the real risk under this scrutiny is just the sheer number of regulators that you kind of need to understand at least to some degree because you know it's not just a US issue.

It's a global issue and well usually it's more more in Europe and since Google reaches the entire world they're constantly constantly defending themselves in court.

So as of the latest quarter they have short-term acred legal and regulatory fines and settlements of about $16 billion. Well, if a regulatory body like the Department of Justice decided that Google had to spin off ownership of one of its business segments to reduce Google's monopolistic power, I mean, that would obviously be a bad thing for Google's shareholders and the areas most at risk a year ago was their web browser, so Google Chrome.

And then there was also some chatter around the Android operating system.

But structural breakups have been exceedingly rare in US antitrust enforcement over the past four decades. I would say and regulators have more often ended up relying on restrictions for how dominant companies can behave than actually going for breakups.

And so that's exactly what we saw with Google. The Justice Department actually sought to force Google to sell Chrome, but the court rejected that remedy and instead it placed restrictions on Google's distribution agreements.

And so Google can still pay Apple to make Google the default search engine in Safari, but those agreements can no longer lock up Apple's distribution in the same way. So they're limited to one-year agreements.

They can't tie Google's default status across different devices and access points. And they have to allow Apple to promote competing search and AI products, even if it's uh just done so in theory,

right? And I think we should probably spend a little more time here because I think understanding regulations is quite important for understanding the Google thesis. Now, do you think it's fair to say that Google is largely safe for now compared with the more maybe acute regulatory risks that they were facing last year?

And if so, that would actually maybe be an argument for Alphabet potentially deserving an even higher PE multiple with some of the more dense regulatory fog now clearing up.

It's a really great point and at a high level, as I was mentioning, big tech companies and shareholders in those companies have broadly benefited from regulatory enforcements that are usually pretty far from the worst case scenario to put it nicely.

And whether Alphabet is forced to eventually spin off some of its businesses or chooses to do so on its own, whether that be with Chrome, YouTube, Android, or whatever it is, shareholders in Alphabet today would of course get a proportional stake in these spin-off businesses.

So the risk for shareholders is not that one of Alphabet's subsidiaries worth hundreds of billions of dollars is just going to disappear from the picture entirely, but it's that if these businesses are forced to operate truly independently, whether they will see their moat shrink because they can no longer benefit from the data and relationships that Alphabet has across their entire enterprise.

And for me, when I first began really looking at Alphabet, I came to terms pretty quickly with the reality that these hefty legal expenses and fees are just a cost of doing business when you operate at the scale that Mag 7 companies do.

And so that really is overall though a pretty small percentage of their business, at least for Alphabet.

And so another way you could think about this too is that the fact that they're subjected to this constant litigation is really a sign that they are truly an extraordinary business.

And so Peter Teal talks about that famously in his book 0ero to1, one of our favorites to recommend. The more time a business spends in court fighting over antitrust concerns and all the efforts they go through publicly to try and downplay some of the monopolistic benefits that they enjoy.

That is actually a sign of an incredibly dominant business, right? It's the insecure companies that are trying to brag about their competitive dominance. Those are the ones that are ironically the least likely to yield the benefits of monopoly for shareholders.

Yeah. And I completely agree with your point there on Teal. If a company is defending itself as much as a business like Alphabet is, I think it's a very clear signal that there's something going on and they probably have some very, very strong competitive advantages.

And of course, they're going to just play them off like there's some sort of minor issue. But in reality, I think they know exactly how strong they are. So, you know, they have to kind of try to address their positioning in a way that appears as least threatening as possible.

But I think in reality, it's very clear that Alphabet does have these monopolistic benefits. And as of now, it kind of appears they're just continuing to strengthen and not actually weaken.

Just to continue the conversation here about the regulatory enforcements around Alphabet and Apple's relationship in particular, I think that's a good area for us to focus.

Alphabet pays Apple something like $20 billion a year to have Google search be the default on Apple devices.

It kind of alluded to that earlier. And once the conclusion in court was reached about the validity of this, Alphabet shares actually did really well and increased by more than 50% in over just a year's time.

And so it's crazy how much opportunity could still be baked into these kinds of overhang investments uh for what you would expect to be the most efficiently priced company in markets basically.

And so in our intrinsic value portfolio, which we'll have a link to in the show notes for anyone who wants to see our holdings and keep up with their portfolio, uh we ended up buying more shares when the price dropped down toward about $150 where our average cost basis was low enough that I felt the downside was pretty well protected.

And so in other words, we had much more to gain than we'd likely lose. Or that was the thinking at least. Yeah. And the interesting thing about the potential divevestature was the judge that was overlooking it, judge Amit Ma, actually ended up with this decision because

he actually felt that forcing Alphabet to divest Chrome would have been bad for the entire ecosystem and not just for Google.

you mentioned that Alphabet pays a steady stream of about 20 billion in cash annually to Apple as a revenue share based on the amount of advertising revenue that Google generates from those searches on Apple devices using Safari that their browser specifically.

the judge decided that the benefits of allowing Chrome to stay under Google's ownership actually outweighed the second order impacts on competitors of forcing a divestature of that business segment.

And so, Alphabet doesn't quite enjoy the same level of contractual guarantees anymore around their relationship with Apple and how far that will extend into the future because now they're basically required to renegotiate that deal every year.

having this arrangement with Apple is very much a win-win for both sides. And it makes it all the more difficult for a challenger to disrupt Google's dominance overformational searches.

how do you beat the fact that Google search is baked into every iPhone that people buy?

So, at least for the time being, I think Google seems to be vastly more efficient and cost advantage in answering most inquiries, while more of the complex questions tend to get routed to paid LLMs, which is really a separate business model from the volume game that Google's playing in its core search business.

Now the company doesn't break out the search business completely cleanly for us but still you know their reporting segment Google search and other revenue has continued to grow at a really nice 14% compounded annual growth rate over the last 2 years and doesn't really show any signs of slowing down.

So I think the initial fear that at least I had was that users would use less and less of Google search and their search revenue would suffer as a result and that just clearly as the numbers say is not panning out.

And in hindsight, I think it makes sense because Google is still very much the best search engine on Earth.

it's really just not a great model compared to Google search because you don't have the same targeting at scale, right?

And looking at the growth in search for Alphabet, it's kind of come from two main areas. So, we got AI overviews and AI mode. I've heard that Google's model Gemini 3 and a half isn't as good as Claude or Chat GBT, I've actually found the answers to be a pretty equal quality just delivered in a much shorter time.

First, I think Google has probably done the best job of meshing the monetization of search with AI. Google, I think, has figured that out very, very well. So, whenever you do a Google query for some sort of question, you'll get an AI overview.

Google has ads that will show above or below the AI overview as well as just inside of the AI overview.

And with Google's distribution advantage, it's sort of their game to lose because they could immediately just roll out AI mode to over a billion monthly active users, allowing them to collect even more data on how consumers are using AI for things like shopping, which obviously is a great great help for advertisers.

And actually on the latest earnings call, Alphabet CEO Sundarpai really focused on just how successful this has been. He noted that Google is actually sending billions of clicks to websites each week from the AI features embedded in search.

of that traffic that no longer needs to go to the underlying website. I mean to some extent that is a real phenomenon but the scale of which I think could be exaggerated. And so the other detail I would probably flag is the cost side.

They've dramatically reduced the cost of what it takes for them to generate a response in AI mode to solve a query down to the lowest level since their launch. And that's very promising.

one of the most surprising things I learned from you about Alphabet is that the majority of searches basically carry no ads at all. So you mentioned that only about 20% of searches have ads according to data from Google in 2020.

And more recently, Google's vice president of search went on a podcast and confirmed that less than 25 percentages have ads still.

So Google has spent a long time just learning how many ads it can show without losing users and bothering people or pissing them off. So, I think it seems like using AI summaries and AI mode will just help them to continue to leverage these findings longer into the future.

Well, there's also another twist that I didn't anticipate when I first looked at Alphabet because it wasn't totally clear yet that Apple would take a more passive role in the AI race and choose other companies models.

But for context, in January, Apple announced that it would pay Google something about $1 billion per year for access to the technology that underlies Gemini to use that to power Siri for the next few years.

And what that also means is that Apple is going to rely more on Google Cloud, too. And so, obviously, that $1 billion in cash annually is not material to Alphabet's bottom line, as crazy as that is to say.

But I I think it's revealing in what it shows about Apple's AI strategy and also in how the two companies are becoming even more intertwined and in different ways.

Yeah, I would say it's a pretty impressive turn of event for Google. So the cloud division recently inflected into earning positive operating income showing that while growth has been off the charts, this is actually a segment that Alphabet can make real money on because they actually lost money on cloud for nearly its entire existence that they've scaled it and they've tried to take share from Amazon and Microsoft.

We're talking about $58 billion in cloud revenue last year at a 24% operating profit margin. And so these are huge numbers, but what's even more promising is that compared to Amazon, Alphabet's cloud business was about twothirds as profitable last year.

And that doesn't necessarily sound like a good thing at face value, but Amazon has been operating in the cloud space for two years longer than Alphabet.

But we actually do have some clues as to where revenue is going to show up specifically for Google Cloud in the next few years because Alphabet gives us a very, very important figure.

Yes, that figure is the backlog. I figure that's what you're assuming to Alphabet has more than $500 billion of remaining performance obligations. That's another way to say backlog for their cloud segment.

And so for those unfamiliar with what backlog means, it reflects the contracts Alphabet has signed with customers for future deals where the work has not yet been delivered. And so therefore, the revenue has not shown up yet on their financial statements, but we know with fairly high confidence that it will come in the future thanks to these order requests.

And so be sort of like a bakery maybe having a bunch of orders for wedding cakes next year. they haven't been paid for those yet and they haven't delivered the cakes yet either.

And unless the weddings are cancelled, that business will come through. And so maybe that's a crude proxy for how to think about Alphabet's order backlog. And and just to give you an idea of of how fast this is growing, Google's cloud backlog was $106 billion in Q2 of last year.

So we're talking about five times growth in the order backlog in about 12 months. I mean that is that's just absurd Kyle.

So just to take the other side of this argument. So when Sean and I were talking about comfort systems I mentioned that I personally am not the craziest about taking too much data from backlogs simply because of the revenue recognition issues that I've seen before in other businesses.

So an example of that would be where backlog might be realized in let's say a year or two and then maybe you try to extrapolate okay well how much of that backlog is going to turn to real revenue and then you can kind of back into a number.

But for my personal experience at least, I've kind of gotten into a little bit of trouble doing that. But in Alphabet's case, the fact that cloud revenue is exploding while the backlog grows is still a very strong signal.

Alphabet says they expect about 50% of cloud revenue growth over the next 2 years. So that's another 130 billion or so over the next year, which would be higher than the current run rate is showing.

Then you have to layer on the fact that the demand for Google Cloud's products doesn't appear to be going away. The fact that they're investing so heavily into that area of their business is a pretty good suggestion that Alphabet believes that the demand for their cloud services is just improving with time.

The other thing I probably missed in my original pitch for Alphabet is where a chunk of this demand was coming from and that's specifically for their computer chips. And so Google designs its own AI processors called tensor processing units or TPUs.

And they're already on their seventh generation of the technology. And so for much of TPU history, they were used as an internal cost-saving tool, specifically for Google. And so that has allowed them to bypass buying chips from Nvidia or or to at least reduce their dependency on Nvidia.

But since the TPU has proven successful, you've had select companies that have actually been offered the opportunity to purchase Google's TPUs outright from them.

And so that point that you just made about the TPUs going to other companies, I think actually kind of helps with diversifying the backlog because it's actually the concentration inside of that backlog that if I'm nitpicking isn't really my favorite setup specifically for Google.

So you know if I had to choose between a backlog with a diverse customer base where no one customer is making up more than 10% let's say of the total backlog, I'd take that any day over a backlog that only has a few key customers where let's say one customer is making up 50% of that backlog.

Now, it's impossible to say exactly what the structure of Google Cloud's backlog looks like. It's likely very concentrated given that they just announced a 5-year 200 billion deal with Anthropic to use Google Cloud.

So, that implies that Anthropic makes up somewhere around 40% of that backlog. So, you know, don't get me wrong, you know, cloud backlog growth is incredible, but it comes with this trade-off in customer concentration where, you know, let's say something were to happen to Anthropic. say their business falls off since you know after all they're still in the very early innings of AI as an emerging industry and so if an event like that were to happen well then a lot of that backlog growth would just appear to be kind of fugazi to quote Matthew McConna from the Wolf of Wall Street it's a great movie and a great quote and uh to keep using that bakery metaphor from before if one customer made up 40% of the value of your wedding cake backlog that would be very concerning right so it's a sort of extreme example because Enthropic is right now on path to do a multi- trillion dollar IPO and will be raising lots of fresh capital.

So, it's not like they're going anywhere anytime soon. But ideally, your biggest customer would not be burning billions in cash while making up such a hefty percentage of your backlog.

But if we can pivot to highlighting a part of the company that I I found to also be very promising, even if it's not as breathtaking as the cloud business, that would be the Google subscription segment.

And so basically, Google has a bunch of different licensing and advertising fees that it earns on top of paid subscriptions like Google 1 that allow you to, for example, increase your storage space on on Google Drive and and in Gmail.

But Google subscriptions are a very, very strong segment, and I use it very heavily personally. So when you're using something like Drive or Gmail and your files and emails accumulate one day you might get a notification saying okay well you're running out of space.

So you have a couple of options here. You can delete things which you can certainly do but I've done it before and believe me it's timeconuming and offers a lot of friction or you can just take the simple route which is to just pay Google a little more money for that space that you can fill up again into the future.

So you know another example would be YouTube. You know, if you're sick of watching ads, well, then you can purchase YouTube Premium and just do away with ads while also being able to do offline downloads and background playback.

I personally like the background playback a bit as it allows me to watch content while answering text messages, for instance. And the cool part about the subscription business is that they're all relatively low priced.

So, when you think of switching costs, they're generally low enough where you just don't really give it that much thought when you think about cancelling or switching. And yet, this business generated over $25 billion in revenue for Alphabet in the first half of 2026.

YouTube is now effectively the largest streaming service on the planet where they also pay much less for the most popular content on their platform than say maybe Netflix where they either have to make the hit shows themselves, produce it themselves or pay a premium to license them.

And so YouTube is is bigger and probably more profitable inherently with monetization being split across ads and paid subscriptions. And so YouTube ads are see nowhere near the explosive growth that other areas of the business are.

But I don't think it requires much spending to grow this part of the business either. So as long as YouTube attracts more content creators, that in turn attracts more users, there'll be a steady stream of people to advertise to.

And so that's really the beauty of the flywheel behind that YouTube business model.

>> No, not not bad at all. And and YouTube is now also testing out a new feature called Ask YouTube. And so it uses Google's Gemini models to let people ask questions about individual videos.

I guess the idea being you get some fast maybe takeaways or you can use it to help you find the most relevant moments in a video be able to filter through the video more quickly.

I actually think that is a pretty interesting value ad. And so Sundar Pachai noted that more than 140 million people have used that service in June of this year alone. And so that doesn't obviously drive revenue directly, but in theory it makes the user experience better.

Maybe people spend more time on the YouTube platform and so that increases monetization by allowing them to run more ads and really just illustrates the rationale behind how some of these AI investments can improve the entire business beyond what's just obviously directly attributable to AI where somebody's paying, you know, X dollars a month for a Gemini subscription.

So this piece of alphabet which at one point could have been seen as you know just a rounding error albeit with a lot of optionality has nearly tripled in value over a year and a half.

So, as an Alphabet shareholder, to the extent that they can capitalize on Whimo by maybe monetizing their stake partially or fully from a Whimo IPO down the road, I think that's sort of all gravy for us.

It does. I haven't got a chance to use it yet, but Alphabet made another big bet in 2015 for about $900 million in a growing space company. Now, I bet you can guess what that company was.

>> I think so. Yeah, I think most people know that it just IPOed. SpaceX. >> That's the one. So, today, Alphabet's 5% stake is worth $95 billion. So, between businesses like Whimo and SpaceX, Alphabet has some monster winners in its other bet segment.

But, you know, I also totally agree with your points from the original research, especially your point that it's just really difficult to evaluate some of Alphabet's more speculative technology bets.

But, you know, at least with a business like SpaceX, we can get some idea of what the market thinks, which can maybe provide a little more clarity on some of its other bets, such as Whimo.

There is one part of Alphabet that I don't think I fully wrapped my head around and that I still find fascinating and to be fair I don't think the market has really wrapped it head around this either and and that's what the ROI from their data center commitments is going to be.

And so if Alphabet is going to part ways with hundreds of billions of dollars to directly and indirectly invest in data centers, I think it's very timely to try and figure out what kind of returns this spending is likely to generate, especially since the magnitude of this estimated spending seems to just get revised higher every quarter.

It's a little complicated to disentangle because the data centers are not only profit centers when we look at Google Cloud's backlog, but they also are a big internal investment inside of Alphabet, helping to power nearly every aspect of the business's overall productivity.

the interesting thing about this data center is that it's actually the third largest in the world and it's actually owned by Google and it's actually located in Council Bluffs, Iowa

Back in 2024, Google data cent's power use nearly equal the entire country of Ireland. And Google's data is much more bigger than that today.

So I think there is no doubt that the investment will bear fruit for Google as it requires this kind of growing amount of compute power just to operate the business of Google. Not to mention that, you know, all the capacity that they're leasing out to other businesses is obviously bearing fruit as well.

But the problem is understanding, okay, well, what kind of economics is Google getting from these investments and once they're all complete? So, if I'm looking at Alphabet's return on invested capital, that number is kind of trending in the wrong direction.

in their last 12 months, their returns on invested capital is now the second lowest that it's been since the business went public.

But the interesting part about this equation is that Alphabet is still making a ton of money. You know, their operating profit margin has actually gone up. But the reason their capital efficiency numbers are trending downwards is actually this increase in invested capital.

Given that Google's capex this year is estimated to be six times more than what they spent in 2023, I guess it's not all that surprising that the incremental returns on some of that capital are not yet manifesting or clearly higher than in the past.

And so Alphabet is going from a company built around the digital world with its software to increasingly being anchored in the physical world to an extent.

another thing that kind of offiscates Alphabet's numbers and cloud in particular is that Alphabet is carrying some of its assets on the balance sheets that are yet to depreciate in value.

So this actually serves to inflate Google Cloud's margins and as of their latest quarter, Alphabet carries assets not yet in service valued at about $122 billion.

To me, this is probably the biggest black box at Alphabet

I already discussed how Anthropic appears to be taking up a large share of Google's backlog. So just to give you an example here a real example of Anthropic deal with Alphabet.

So Alphabet committed up to about 40 billion into Anthropic as part of the deal. So you know to put it simply Google invests in the customer the customer buys Google compute then Google then books the backlog.

So in looking at Alphabet, they've traditionally been a business that has compounded their per share value. And part of this was completed through buybacks, which ran between say 45 billion to about $60 billion between the years of 2023 until 2025.

But as of the first half of 2026, they're literally zero.

And additionally, Capex has eaten up almost all of Alphabet's free cash flow. And in Q2 of 2026, they actually had their first negative free cash flow quarter in a very, very long time.

Well, now we're now we're bashing Alphabet a little bit here, but I do think that it's a productive exercise to do that because after all, we're pretty bullish on on Google, otherwise it wouldn't be the largest holding in our portfolio.

But just as maybe some more context on the composition of this AI spending, Alphabet's CFO Anat Hashenasi has said the mix of the AI infrastructure investments that they're undertaking is approximately 60% in servers and then about 40% into the data centers and and networking equipment.

So, in Google's case, it has a 1.09 PUE. And this is supposedly much lower than the industry average of around 1.3, meaning that Google is just basically running much more efficiently than the average data center is.

Now, it's great that Google is more efficient here, which gives them a number of scale advantages. These are advantages including cost savings, greater computing capacity per megawatt, and a reduced carbon footprint.

And so Alphabet is supply constrained with with power and land and supply chain, but everything else is irrelevant if the energy needs cannot be met.

So, you know, even if you are making a complaint about Google's appreciation being artificially low, they're still probably much more profitable than either Nibbius or Coreweee is today.

So you mentioned here earlier Sean that Alphabet has developed many of its own chips including the tensor processing units. Now because of this it skips to some degree this kind of Nvidia tax that pretty much all of these other companies are forced to pay.

Now Google doesn't necessarily have that same issue. So Google's tensor processing units come in at about a 40% discount to the NVIDIA equivalent.

And as we've seen with Anthropic deal with Google, it just makes more sense for many of these AI businesses just to rent compute power from Google rather than building it out on their own.

Another simple advantage that Alphabet has as well is they have the ability to fill up their capacity themselves.

So Alphabet really has very little capacity being wasted. They have a lot of flexibility and if there's a brief shortfall in customer demand that compute can be redirected internally and it's sort of like a hotel that doesn't have any room occupancies.

and as of now this probably is not a problem for Neoclouds like Cororeweave and Nebius but if Alphabet's advantages continue to grow there may come a time when these businesses find it difficult to keep the lights on with too much unused capacity

Now, for Alphabet, owning one year's worth of data centers runs about $30 billion in expenses once it's all switched on. And that composes of about 25 billion in chips that are slowly wearing out and the rest in power and upkeep.

So if we work backwards from a decent profit, you land somewhere around here, which is one year of spending has to bring in somewhere between 70 and hundred billion of brand new sales every single year.

If you simplify that, the rule is about 40 cents of new annual revenue for every dollar spent. That's kind of where the bar sits.

So as of now I think Alphabet is far from guaranteed to making good returns here. and that's obviously a condition that Google feels it makes sense for them to justify this amount of spend.

Makes sense and and clearly as an Alphabet shareholder it's it's great to see this.

And that is a company for context that specializes in the maintenance of servicing data centers. So the stock has gone up in a straight line, but we really don't know how good the business will be when things normalize and and to a lesser extent, the same is true for Alphabet.

Yeah. And as Alphabet has become more capital intensive, it really actually surprises me that Warren Buffett, who's kind of known for preferring businesses that don't require much capital to generate these cash flows, has now chosen now to be the time to invest in Alphabet after he first discovered the business, you know, 20 years or so ago while running search ads for Geico.

I remember you highlighting in the original episode how Alphabet just had so much cash that they didn't really know what to do with it. They had so much cash that they were just returning it to shareholders through buybacks, which I think felt right for a company at that time with excess capital.

To your point on Buffett and Birkshshire, not only have they used their own cash piles, but they've increasingly tapped the debt markets for more financing, and now they're turning to selling equity to raise enough capital.

Yeah, that's right. So, they said they were going to spend about 200 billion on AI data centers this year. Then they said, "We need more money." So, they went out and raised many, many billions of dollars.

And I think this shows that Alphabet isn't focused on just managing their excess cash anymore.

So, here's what I find noteworthy about all this, though. So, the share count has declined pretty steadily since 2018 because of those buybacks that I just discussed. But as of Q2 2026, they have 12.3 billion shares outstanding.

And that's actually the highest number since 2024.

That kind of shift shows you how much they're prioritizing this data center spent. If you think about it, assuming Alphabet is going to continue to need funding for its AI data centers, there is a good chance that they'll use all avenues available to them, which likely means more net share issuance.

That increases the number of shares outstanding. And whether that is technically dilutive depends on how effectively the capital raised from selling more shares is deployed. But at a really simple level, you've spread the business's intrinsic value across more shares and shareholders.

And that is at a minimum going to put pressure on the stock in the short term because you're increasing the supply of shares to the market.

And so while Birkshshire invested $10 billion into Alphabet directly, Alphabet's total equity raise earlier this year was for $85 billion. And so the market was happy to fund that evidently.

But it's the first major equity raise we've seen from them since they IPOed.

And in theory, Alphabet should be at a scale and a maturity where they don't need to rely on selling stock to raise money. Right? That's something you associate with startups and unprofitable tech companies because it's a very costly form of financing.

>> That's right. And the fact they haven't had to issue equity for two decades, I think is a pretty obvious signal of just how good of a business alphabet is and speaks volumes about their ability to generate cash.

But the share issuance is actually just the beginning of their capital raising effort. So as of Q2 2026, long-term debt is now $98 billion or about nine times since the fiscal 2024 year ended.

>> Yeah. With how conservatively financed Alphabet has been, even I was a little surprised at how much this figure has grown. >> Yeah. And part of the increase in debt has been from raising money in both domestic and global bond markets.

So I guess you know if you can't raise it all from home, why not look elsewhere, right? And the appetite for these bonds is very very high. So in February they issued about 20 billion of US dollar denominated bonds ranging from three to four years.

And this was upsized from 15 billion. And I read the offer drew more than 100 billion in orders.

And then outside of that, they've also just looked around the world. They've raised over $50 billion in other currencies such as the sterling, Swiss Frank, euro, Canadian dollar, Japanese yen, and even Australian dollars.

>> Yeah. So they currently have $242 billion in cash and cash equivalent. So, even with all this new debt, there's still obviously net cash, but that 242 billion is probably going to get depleted very quickly given their current spending rate.

And based on what management has said, they're not planning on slowing that down anytime soon. What are like the the backlog and with Alphabet alone, we're talking about more than $800 billion in offbalance sheet liabilities over the next few years.

And I'm not misspeaking when I say that. Literally almost a trillion dollars in liabilities that are not at present reflected on the balance sheet that weren't even conceivable a few years ago, right? they weren't on anybody's radar.

So, of course, the business is riskier today than it was in the recent past because there's just no way to get around the fact that a tremendous amount of financial uncertainty has been injected into the business and their future financial prospects because of these different liabilities.

But also, I think you could argue that if AI is even remotely as revolutionary of a technology as Silicon Valley thinks it is, then Alphabet's ability to do all this spending is is buying them maybe several more decades of dominance and being one of the world's biggest tech companies.

And so, just kind of looking at the facts, plain and simple, the risks have increased. They've changed also, right? a little bit less of a regulatory discussion and and competitive discussion and more about how the returns on this capital will look like.

But on the flip side of that, Alphabet's corporate life cycle has potentially been reset pretty dramatically. And so all of a sudden, Alphabet looks to be a much younger company with more dramatic growth possibilities ahead instead of being some kind of stagnant mega conglomerate that was just slowly in the process of of hardening.

So, Alphabet reported 112 billion in profits in its second quarter. But when you dig in, 99 billion of that was a non-cash gain from marking up its equity stake in private companies like SpaceX and Anthropic.

And that single gain added over $6 in EPS, but it's zero actual dollars coming into the business. And with Anthropic planning to IPO at some point in 2026, we're probably going to continue to keep seeing these large non-cash gains on the income statement.

It all just kind of serves to add complexity when analyzing this business. So, I think the flag here is simple. Don't get too excited by the headline numbers. Focus on operating income or cash flows when you're evaluating Alphabet.

But beyond the income statement and depreciation accounting, I would come back to the reality that the cloud backlog sort of epitomizes both the upside here as well as the risks facing Alphabet.

So half a trillion dollars in revenue is being committed into the backlog is really great, but Anthropic and OpenAI have to actually be able to afford to pay that money in cash to Alphabet regardless of whatever they promise today.

And so the cash has to come from somewhere. And I'm nowhere near optimistic enough about either of those companies to say with conviction that they will be solid customers of Alphabet for many, many years to come.

Now the next risk that I raise here is on the margins. So the margins obviously look really really good right now. Operating margins is sitting in the low30s. But here's where I think many people miss about this massive capex. every dollar that Alphabet spends on these data centers today is going to become depreciation expenses next year or the or many years from now.

So even if their revenue just stays flat, even if they don't grow at all, margins automatically will compress from that added depreciation expense. And when you're spending $200 billion a year, well that depreciation hit is going to be quite substantial.

Now the final risk that I'd flag right now is that Alphabet doesn't really seem to me to have a variant perception as in the stock, not the actual business.

You know, analysts currently love the stock and are issuing strong buy recommendations with an average target price of around $430 versus a current price of about $340.

Now, I think Shawn did the right thing adding this business very, very heavily when it was completely out of favor. You know, this business right now is not out of favor at all at this time.

So without getting into the weeds of valuation too much, I will say it was a really useful exercise for me to go through your model from last year on Alphabet, update it a little bit, and think about whether the market is offering us a really attractive bargain with that stock.

Well, in short, I think we'd both say that we see this as an exceptional business, but one that is much closer to being fairly valued today than it probably was a year ago.

And that's partly because of the runup that the stock has had, but also because of the changes, these new uncertainties that are facing the company that personally I have less confidence in.

I felt more comfortable saying that I disagreed with the market about chat GBT being a real disruptor of Google search. I'm really not sure that I have a strong opinion that differs from the market where I feel strongly one way or another about whether these massive investments that they're making are going to underperform or outperform expectations.

And so if you're searching for something on the internet, at the end of the day, there is a very good chance that you're going to touch one of Google's products. I use YouTube pretty much every day myself and I just can't think of any substitute that would turn my attention elsewhere.

And so Alphabet's not going anywhere. This is an incredible company.

We're very happy to kind of let our winners run here. And I don't want to overthink things, but yeah, I don't see as compelling of an argument that I can make in good faith that that Alphabet stock is as undervalued as it was this time last year, despite ironically having the same PE ratio.

Yeah, I really don't have any arguments here. I think my stance on Google is quite simple. I just don't think it's the right time to add to the position. You know, if we were speaking strictly rationally, then selling is actually probably the right decision due to the fact that we already made nearly a double.

But, you know, we are long-term investors. So, sometimes the most optically rational decision isn't actually the correct move.

And then speaking of maybe potentially trimming, well, we aren't really into market timing. So, I think selling pieces of the stake assuming that we can buy more for later at an even cheaper price just doesn't really make that much sense.

So, in my view, I think the best move with Google, given that we already own it, is to just do nothing and let the position play out. >> Well, I I completely agree with you there.

And uh yeah, you know, it's sort of a weird concept. I remember when I first pitched Alphabet on the show. I concluded that it was about fairly valued. Actually, I I didn't even make the case for being super undervalued.

So, when I say that it was really undervalued, I I say that in hindsight.

And we had a listener write to me and ask me why we would decide to invest in Alphabet if the conclusion was that it was fairly valued. And back then mostly it was because I knew I'd been like very very conservative in the modeling when I said that it was fairly valued.

But again, it kind of goes into this idea of just wanting to own wonderful businesses because they will surprise you to the upside typically and just letting them compound over time and and just sitting on your hands and not doing anything about it.

So, I felt like if I could get a wonderful business at a fair price, that was a great arrangement.

Alphabet has surprised me to the upside a number of times in the last year. And that's why I sort of take this benign and maybe more like tranquil view on the runup in the stock in the last year where I wouldn't want to bet on where the stock price is going to go in the next 30 to 60 days.

But looking at this with a 10-year, 20 year perspective, if you can even fathom thinking that far out, I think these are rounding errors in hindsight. And the simplest thing is to just sit back and watch because you'll drive yourself crazy if you're trying to time every single swing in the market and buy and sell and buy and sell.

It's just not a winning formula. It's not a recipe for success.

And this one is coming from the Oracle of Omaha himself, Warren Buffett, back in 2012. So he said the chances of being way wrong in IBM are probably less at least for us than being way wrong with Google or Apple.

But that doesn't mean that those the latter two companies aren't going to do say far better than IBM. Now this is just such a great quote because I think it showcases Buffett's thinking process.

In just a few sentences you can see how he thinks about opportunity cost and why upside isn't the only thing that matters.

And of course he was completely correct that both of those businesses provided much better returns than IBM. but he just didn't have the competence in his understanding of Apple and Google at that time to make an investment.

But now he's invested in both. And perhaps you can argue that maybe he feels like he's gotten competent enough or that, you know, at the time that he bought them, the businesses just had been derisked to some extent and that downside was well protected.

For 5 minutes, Google was unusable. As a result, global internet usage fell by 40%. We also shouldn't forget that there are over three billion monthly users of the Android operating system.

As Android controls about 70% of the global operating system market for mobile devices, it's not lost to me that as I research Alphabet, I've gone back and forth between Google Docs, Google Sheets, Google Drive, and Google Search, all while using Google Chrome, and watching YouTube videos about the company.

So, that just really speaks for itself. Google is such an integrated part of my life that I had to stop and think about all the different touch points I have with Google services.