At the beginning of 2025, Shawn reviewed Alphabet as one of his first few ideas for The Intrinsic Value Portfolio. And since then, the position has performed very well, growing into our second-largest position in the portfolio. But we haven’t revisited the idea on the podcast, and I thought it’d be a good idea to look at how the original thesis has played out, as it’s already generated a 96% return.

I had a few burning questions about Alphabet in 2026. Given how much money they’re now spending on AI Data Centers, the first question was: will these large capital expenditures yield good returns?

Alphabet has been a business with exceptional capital efficiency numbers over its history, but investors are questioning whether it can maintain these numbers while spending unprecedented sums on AI infrastructure.

Mostly, though, I just wanted to see if the investment thesis was still on track, so let’s get into it!

— Kyle

Shawn’s original thesis rested on a simple idea: Alphabet isn’t reliant on any one part of its business. It’s a collection of incredibly profitable world-class assets with billions of regular users.

You got Search, YouTube, Cloud, and Android among others, which, if spun off from Alphabet, would be giants on their own. YouTube, which Google bought in 2006 for $1.65 billion, now generates over $40 billion in advertising revenue and is the largest streaming service in the world. Then you have the investment arm that owns assets like Waymo, SpaceX, and Anthropic, having made substantial investments in these businesses while still in their infancy.

Shawn concluded that in early 2025, the market was pricing Alphabet as if Search were its only business unit. And Mr. Market felt that search was ripe for disruption. The fear at that point wasn’t completely crazy. It was one of the largest threats to Google’s monopoly in years, and I have to admit, I felt it personally. I was a heavy ChatGPT user at the time, and even I could see my own Google usage starting to dwindle.

But Shawn’s question wasn’t whether AI was a threat. It was whether that threat within a diversified business justified cutting the stock’s multiple in half. His answer at that time was an emphatic “no”. And so far he’s been right.

The theory the market seemed to be running on to justify this multiple compression was that AI would disrupt Alphabet’s search function. Oddly enough, I’ve noticed my use of Google Search increasing quite a bit lately. Over the last year, I’ve gone from ChatGPT to primarily Claude. And even now, I find myself using Google’s AI summaries more and more often as I like the quick replies and links to the original source to make sure the information is correct.

So when I need a quick answer I can trust, I’m still heavily reliant on Google.

And apparently I wasn’t alone on this insight. Google’s Search and Other segment has grown at a 15% CAGR over the last decade and isn’t showing much in terms of a slowdown. That's not the kind of clue a dying business unit leaves behind.

Some of the newer growth drivers for Alphabet’s search are coming from two sources, which now reach over a billion monthly users.

My first question when thinking about this was: how is Google making money here? With Claude, it’s pretty obvious. I pay them a subscription fee, and I’m off to the races. But with Google, I’m not paying anything to read an AI Overview.

And the answer is ads. When I search for something like “what are the best waterproof shoes,” which is relevant for me being a Vancouverite with the winter rain coming, I’m hit with a bunch of ads based around the AI Overview. Both above and below are ads from Vessi, Salomon, and On. And even within the AI Overview, there are links to brands paying for the right to appear.

And if I want to dig deeper and make sure they are good for walking in the city or hiking, I’ll refine the search further, allowing Google to gather more data about me and show more relevant products.

A few other details from the most recent earnings call really reinforced their strength in this area. First, Google sends billions of clicks to websites each week from its AI features, which puts a pretty big dent in the whole “AI will starve publishers” thesis. Second, the cost of serving its AI Mode is at its lowest since launch, indicating that Alphabet is able to leverage its scale benefits. And lastly, fewer than 25% of searches even carry ads. This means they aren’t being overly annoying with showing them.

Beyond search, a major overhang could have meant large-scale changes for Alphabet: the regulatory environment. Alphabet continues to defend itself in court globally and currently has $15.6 billion in accrued short-term legal fines and settlements on its books.

So a year ago, the scariest outcome regulators could have handed down was a forced breakup. The Department of Justice wanted Google to divest itself of Google Chrome. But the judge in charge concluded that wouldn’t be necessary. Instead, he restricted Google's distribution deals. Google can still pay Apple $20 billion a year to be Safari’s default search engine, but the judge cut contract lengths to one year, barred exclusivity, and required Google to share certain data with qualified competitors.

Just to give you an idea of how powerful Alphabet is within the tech ecosystem, the judge noted that forcing a sale of Chrome would hurt Apple and Mozilla, which depend heavily on revenue generated by Alphabet. Structural breakups have been quite rare in U.S. antitrust cases over the past four decades. And even if regulators forced a spinoff, shareholders would still receive a proportional stake.

After the ruling on Chrome, Alphabet's shares have increased by more than 50%. Who knows if all of that gain was a result of the case ending. But I think it quite clearly shows that markets can be manic. Shawn saw this event as an opportunity to start the Google position, and so far, it’s worked out very well!

While I’m not going to say AWS is going to die anytime soon, I think it was clear that Amazon Web Services (AWS) had a massive lead over Alphabet for quite a long time. The reason was scale.

For most of Google Cloud’s life, it lost money as Google had to fight Amazon and Microsoft for market share. Last year, the segment generated about $58 billion in revenue with a 24% operating margin. This made it roughly 2/3’s as profitable as AWS. In the original thesis, Shawn viewed AWS as the roadmap for Google Cloud’s future, but it’s closed that gap much faster than he or I could have imagined.

So, with Google Cloud’s margins improving and its incredible growth so far in 2026, Google certainly has a very strong growth driver here. Just for context, Google Cloud’s year-over-year revenue growth was 63% in the latest quarter, while its backlog nearly doubled quarter-over-quarter.

While the backlog growth is impressive, I’ve always been apprehensive about putting too much stock in backlog for revenue recognition. This could just be my own biases, though. The fact that Alphabet’s revenue and backlog have exploded together definitely gives me confidence that it will remain a trustworthy metric going forward.

Another major tailwind for Alphabet is in the chip sector. Google has designed its own chips; they call them Tensor Processing Units, or TPUs. They’re now on their seventh generation. For years, Google used these exclusively in-house to avoid buying NVIDIA chips. But now Google is selling TPU capacity, and sometimes the chips themselves, to external customers.

Another area I’d nitpick in Alphabet’s Cloud backlog is the customer concentration risk. I generally prefer that no single customer accounts for more than 10% of a business’s revenue. Anthropic recently announced a five-year commitment to Google Cloud. With the backlog over $460 billion, Anthropic accounts for around 40% of it.

To be fair, if we look at Google Cloud as just one part of the overall Alphabet business, then Anthropic’s revenue contribution easily falls below that 10% mark.

One of the best parts about Alphabet is that not every one of its business segments needs a budget in the hundreds of billions of dollars to grow.

And subscriptions are a super-sticky business with high switching costs. If my Google Drive or Gmail account gets full, I can try to fix it, maybe by deleting old files or emails. Or, I can just spend a few extra bucks per year to free up even more space and just forget about it. Most of us just end up paying. Same goes for YouTube Premium.

They’re both cheap enough that most users don’t really bother thinking about cancelling. But those subscriptions, overall, add up, with Alphabet accruing over $25 billion in subscription revenue in just the first half of 2026. They also reached 350 million subscribers.

As a daily YouTube user, I can’t help but be impressed by its underlying business model. It’s the largest streaming service on the planet, yet it has a superior economic model compared to its rivals. Netflix and Disney spend billions of dollars producing content. YouTube lets creators do all of that work, then pays them 55% of the ad revenue without having to finance the content.

And YouTube clearly has stand-alone value. If you wanted to value YouTube similarly to Netflix, it would be worth between $300 and $500 billion.

When Shawn first pitched Alphabet, its autonomous driving service, Waymo, had raised money at a $50 billion valuation. In February, it raised money with a $126 billion valuation.'

And Waymo is really starting to scale. They’ve now driven 127 million autonomous miles, with 90% fewer serious injuries than human drivers. In 2025, it completed 15 million rides, triple the prior year. The current run rate, based on 500k rides per week, will bring them to 26 million rides per year sometime in 2027.

The big question about Alphabet is its current spending forecast. The expected amount for 2026 is just below $190 billion, nearly topping the total for the past four years combined. Historically, with Alphabet, their internal spending wasn’t really a big question mark because they had so many great places to invest. Or they could rely on share buybacks.

But the big topic here is that the bulk of that money is going into data centres. And Alphabet will use these data centers to power Search, YouTube, and Gemini internally, and also to rent capacity to third parties. And I don’t really think the market has figured out whether this spending will actually turn a profit, and if it does, how much it will be.

I’m naturally skeptical, so let me start with the bear case for this buildout. Alphabet’s current return on invested capital is pretty close to its historical lows. Keep in mind Alphabet is a very good company, and that number still sits at “only” 26.4%.

So why has this number dropped? Operating earnings and margins have actually risen, so the recent decline is more due to the influx of new capital into the business. Ironically, the biggest software companies in the world are now becoming less digital and more reliant on physical assets. Five years ago, this event would have been completely unheard of.

There are four reasons why this spend concerns me:

Not all of its assets are producing revenue. Alphabet carries about $122 billion of assets not yet in service. And these assets aren’t depreciating on Alphabet’s balance sheet. This means the current Cloud margins aren’t covering their full future costs.

What depreciation assumptions are correct? Michael Burry argued that hyperscalers are collectively underestimating their depreciation by about $175 billion dollars over the next five years. If data centre components like servers need replacing every three years rather than the six years they are scheduled for in the depreciation schedule, then Alphabet's reported earnings overstate the true economics.

Bye-bye buybacks. Alphabet has hit the gas pedal on buybacks, never spending less than $45 billion since 2020. But over the last two months, they haven’t spent a penny.

Negative Free Cash Flow. In the latest quarter, Alphabet generated $39 billion in cash from operations but spent $45 billion on capex. This resulted in free cash flow turning negative for the first time in more than 20 years.

Understanding how much money Alphabet needs to generate from this capex spend is not easy. But given how much they are spending, I thought it would be an interesting experiment to at least try and see what kind of revenue Alphabet needs for these investments to make sense.

We treat the first year of spending as $200 billion for simplicity. Running the data centre will cost about $30 billion, with 60% going to servers and 40% to buildings and network gear. If Alphabet wants to generate a decent profit, I calculate they need to earn about 40 cents for every dollar spent, meaning revenue will need to reach around $80 billion. Think of that as the bar they need to surpass for these investments to make a reasonable return.

But is this number doable? It’s hard to find out exactly how much these AI data centres charge. That’s because most hyperscalers aren’t in any rush to disclose what they charge, at least not to non-insiders. The industry is already saturated, and inviting potential competitors in is a strategy they’re trying to avoid.

This is where businesses like Nebius and CoreWeave come in handy. They’re pure-play businesses renting out AI computing capacity.

The problem is that their pricing isn’t the clearest. We know Nebius has doubled its pricing over the past six months. CoreWeave increased prices by about 25% in July alone and noted that near-term capacity is sold out. To add to the fact that demand is high, Nebius noted that they could sell out their entire 2027 capacity today if they wanted to. When you put it all together, it’s clear that demand is high and suppliers are exercising pricing power because of the supply shortage.

But here’s the problem. The actual economics of both businesses are still hard to come by. The standard price per MegaWatt is what we want to know, but it’s really hard to determine because prices are fluctuating so much. And even though the number is increasing now, once the hyperscalers bring more supply online, the price per megawatt will be very hard to predict.

So with all that said, it’s tough to conclude these investments will be a cinch for Alphabet. As of now, there is a supply gap, but it’s impossible for me to know when it will close.

You may be thinking I’ve been trashing Alphabet all this time, and to be honest, that’s not my intention. I still think Alphabet is a great business, and will continue to be great in the future due to its monopoly on search. But even when it comes to the AI data centres, it still has an edge over businesses like Nebius and CoreWeave. Unlike these neo-cloud businesses, Alphabet actually turns a profit, helping to protect its downside.

First off, since they have their own chips, they don’t have to rely solely on NVIDIA for sourcing. They can also produce chips at a much better price than NVIDIA offers. It’s no surprise that Anthropic chose to rent out capacity from Alphabet rather than build its own.

Second, Alphabet can utilize its own capacity. Nebius and CoreWeave do not have this advantage. When supply is more in line with demand, if Nebius or CoreWeave has underutilized capacity, it will end up burning a hole in its pocket. Alphabet can use its compute on its own business.

And finally, Google’s power usage effectiveness exceeds industry standards. This means they are more energy-efficient, allowing them to operate at better margins by keeping energy costs lower than competitors. And power matters a lot. It’s only 6% of annual AI capacity cost, yet it’s the biggest bottleneck in bringing more capacity online.

When Shawn originally pitched Alphabet, one of the issues he saw with the business was that it had too much cash and nothing to do with it. That’s why the strategy was to use the cash for buybacks. This helped reduce Alphabet’s share count.

But the days of share buybacks seem to be over now, or at least on hold. Alphabet has raised well over $100 billion through equity financing and bond issuance. Long-term debt is now $98 billion, about nine times the level at the end of 2024. But on the positive side, they are also currently holding $242 billion dollars in cash and cash equivalents.

But if they intend to ramp up their capex spending to the $200 billion-plus range annually, and potentially higher in later years, the chances are high that they’ll need to refinance in the near future.

When I assessed Shawn’s initial modelling of Alphabet, I was impressed with how spot-on he’s been. While Alphabet's growth story still seems on track, its balance sheet has definitely changed. Because the business is so good, I don’t think they’ll have any trouble finding suitors willing to give them money when they need more.

The question is, will they focus more on tapping debt markets, or will they turn to doing even more dilutive financing deals? Because Alphabet still generates so much cash flow, I think they’re in a good position to figure it all out. Even if the AI data centres don’t offer the best returns to third parties, they will still enhance Alphabet’s own product offerings, making them even more valuable.

Shawn’s sum-of-the-parts intrinsic value puts the company’s share price at $205. His assumptions about growth rates, profit margins, and multiples all seem realistic to me, so I don’t see much reason to adjust his model.

Because Alphabet has done so well since it was added to The Intrinsic Value Portfolio, our decision now is whether to add more or sell a partial or full stake. We will continue to maintain the position, but we currently have no interest in adding to the position. And since Alphabet has a multitude of compounding characteristics, we don’t think trimming is the right move either. With that said, our decision was simple: do nothing.

To listen to our discussion of Alphabet, or for more company Deep Dives, check out our podcast here.

Updates on our Intrinsic Value Portfolio below 👇

To discuss stocks daily with Shawn, Kyle, Daniel, and all Intrinsic Value Mastermind Members, apply here.

"AI is probably the most important thing humanity has ever worked on. I think of it as something more profound than electricity or fire."

— Sundar Pichai, CEO, Alphabet

📺 WATCH: How to Live Off Your Portfolio w/ guest Ian Cassel on The Investor’s Podcast

🎧 LISTEN: AI Kills Everybody or Doomer Psyop? by the All-In Podcast

📖 READ: The Stories We Tell Ourselves by Adam Wilk on the difficulties of success and failure in investing

You can also read our archive of past Intrinsic Value breakdowns, in case you’ve missed any, here — we’ve covered companies ranging from Alphabet to FICO, Transdigm, Perimeter Solutions, PayPal, DoorDash, Crocs, LVMH, Uber, and more!

Enjoy reading this newsletter? Forward it to a friend.

Was this newsletter forwarded to you? Sign up here.

Apply to join our Intrinsic Value Mastermind Community of investors.

Read our full archive of Intrinsic Value Breakdowns here.

Keep an eye on your inbox for our newsletters on Sundays. If you have any feedback for us, simply respond to this email or message [email protected].