Latest / Deep Values: Value Investing Deep Dives / Why Alphabet Is Warren Buffett's New Railroad
Transcript
- 0:00You know, for decades, if you read like any finance textbook
- 0:05or listen to any market analysis, there was basically
- 0:09one universal absolute truth when it came to Warren Buffett.
- 0:13Right, the man does not invest in technology companies.
- 0:15Exactly. I mean, it was treated almost
- 0:17like a law of physics. He famously stayed away because,
- 0:20well, he simply said he didn't understand their game.
- 0:23Yeah, he liked railroads. He liked insurance.
- 0:26You know things you could physically drop on your foot,
- 0:28right? Tangible stuff.
- 0:29And it was a very highly defined, very comfortable
- 0:33boundary for him. The underlying philosophy was
- 0:36just don't put capital into a business model that you can't
- 0:39accurately predict 10 years down the line.
- 0:41Which makes total sense. But then you wake up one day,
- 0:44you take a look at Berkshire Hathaway's portfolio, and
- 0:46suddenly there is a massive, glowing $31 billion position in
- 0:51Alphabet. Alphabet.
- 0:52Yeah, the parent company of Google. 31 billion just sitting
- 0:56right there. It is.
- 0:57It's the ultimate financial paradox.
- 0:59It completely shatters the conventional wisdom about how
- 1:01Berkshire operates, and honestly, it forces us to
- 1:04completely reevaluate how they're looking at the modern
- 1:07economy right now. And that is exactly our mission.
- 1:10For today, we're doing a deep dive into two incredibly
- 1:13revealing transcripts from CNBC interviews, one featuring Warren
- 1:17Buffett himself and another with his successor, Berkshire CEO
- 1:20Greg Abel. We are going to unpack this
- 1:22massive strategic pivot, decode Berkshire's entirely new
- 1:26definition of a you know, a good business, and reveal the hidden,
- 1:30intensely physical bottleneck of the AI revolution that almost
- 1:34nobody in Silicon Valley is even talking about.
- 1:36It's fascinating stuff. But before we jump into this, I
- 1:38have a quick note for you to access more advanced researches,
- 1:42follow Guru's latest movement in AID and leverage our latest deep
- 1:45values agents. Welcome to download Deep Values
- 1:47APP from App Store to track stock market temperature.
- 1:51Welcome to download iOS app Buffett Indicator from Apple App
- 1:54Store or Bubble Watch from Google Play.
- 1:57Links in the description Disclaimer All analysis is based
- 2:02on publicly available sources and may be subject to revisions
- 2:05or differing interpretations. All contents from this channel
- 2:08is for research and entertainment purposes only and
- 2:11does not constitute any investment or financial advice
- 2:13or service. Conduct your own due diligence
- 2:16before making investment financial decisions.
- 2:18OK, let's unpack this, because the way this Alphabet deal
- 2:22actually went down sounds way more like a Hollywood script
- 2:25than, you know, a corporate merger.
- 2:27Oh, absolutely. The execution of the deal defies
- 2:29almost every expectation you might have at how a massive
- 2:32global conglomerate makes capital allocation decisions.
- 2:35So let's set the scene here. Buffett recently celebrated his
- 2:3996th birthday on a Sunday in Omaha.
- 2:41He's there with family, friends and Greg Abel is there.
- 2:45Too right, His designated successor.
- 2:46Right. They spent a great afternoon
- 2:48together, and then Abel literally gets on a plane and
- 2:52flies to Tokyo to check on Burscher's Japanese investments,
- 2:55which, by the way, Buffett absolutely loves she.
- 2:58Really does. But that detail, you know,
- 3:01flying out right after the party, it highlights the dynamic
- 3:04between these two leaders. Buffett openly says Abel is the
- 3:07decider now, but they still talk every single day.
- 3:09Every day. And it turns out Buffett is
- 3:12actually the one who initiated the alphabet position about 15
- 3:15months prior to this. And that daily communication is
- 3:18basically the bedrock of what happens next, because the
- 3:21purchase wasn't just some slow methodical accumulation of
- 3:24shares on the open market over several months, which is how we
- 3:28normally picture these big institutional trades happening.
- 3:31Yeah, not at all. So in late May, on a Sunday
- 3:33morning, Abel gets a phone call. Just out of the blue.
- 3:36Yep. The caller is gauging interest
- 3:38in an upcoming equity offering for Alphamet.
- 3:41But they don't have a set dollar amount, they don't have set
- 3:43terms. They're literally just testing
- 3:45the waters. Just seeing if Berkshire is
- 3:47hungry. Exactly.
- 3:48So Abel says. I'll get right back to you.
- 3:50He calls Buffett, they talk it over, and they decide right then
- 3:54that they want a massive block of shares, $10 billion worth.
- 3:59Ten billion on a Sunday. On a Sunday morning.
- 4:02But they don't just take whatever prices offered, Abel
- 4:05recommends. They ask for a 6.5 percent
- 4:07discount. So they ask for it, they get it,
- 4:09and the transaction is consummated just like that.
- 4:12What's fascinating here is the sheer absence of bureaucracy.
- 4:16I mean, you have one of the largest companies in the world
- 4:18acting with the agility of like a two person startup operating
- 4:22out of a garage. I.
- 4:23Have to pause you there because the mechanics of this are just
- 4:25wild to me. How does a $10 billion
- 4:28transaction happen on a Sunday morning?
- 4:31Right, it seems impossible. I mean, if any normal Fortune
- 4:34500 company gets that call, they need a month of committee
- 4:38meetings, risk assessments and like 20 PowerPoint presentations
- 4:42just to reply to the e-mail, yeah.
- 4:43That is the magic of the governance structure they've
- 4:46built at Berkshire. There are basically no layers of
- 4:49middle management. But, you know, let's look at the
- 4:51mechanics of the trade itself, because that explains why the
- 4:55caller needed Berkshire in the first place.
- 4:56OK, laid out for us. If a major shareholder tries to
- 4:59sell $10 billion of stock on the open market on a Monday morning,
- 5:04what happens? The market panics.
- 5:06Exactly. The algorithmic trading systems
- 5:09will instantly spot that massive influx of supply and the stock
- 5:12price will plummet before they can even sell a fraction of it.
- 5:15OK, so they need a whale, someone who can swallow the
- 5:18entire 10 billion in one single bite completely off the public
- 5:22market like a block trade. Precisely, they need a buyer
- 5:25with massive cash reserves who won't flinch, and in exchange
- 5:29for providing that instant certainty and stability,
- 5:32Berkshire gets to demand a 6.5% discount from the current market
- 5:36price. Wow.
- 5:38And the seller just agrees. The seller agrees, because
- 5:41taking a guaranteed 6.5% haircut is way better than risking a 15
- 5:46or 20% crash on the open market. It's essentially 2 liters.
- 5:51Buffett and Abel deeply understanding their own internal
- 5:54logic, looking at a math problem on a Sunday morning and solving
- 5:58it instantly. It's literally like walking into
- 6:00an exclusive invite only garage sale, spotting A pristine
- 6:04vintage Ferrari and somehow negotiating A multibillion
- 6:08dollar discount before you've even had your morning coffee.
- 6:10That's a great way to put it. But look, agility only matters
- 6:13if the target is right. So what does this all mean?
- 6:16Why act at all? Why did the guy who notoriously
- 6:19they ignored theentire.com boom suddenly decide he needed $31
- 6:23billion of a search engine and AI company?
- 6:25Well, to understand that pivot, we really have to dig into
- 6:28Buffett, its core philosophy, which was heavily influenced by
- 6:31his late partner Charlie Munger. For them, a good business isn't
- 6:34about changing the world or having the flashiest product.
- 6:38It is a strict mathematical reality centered on a metric
- 6:41called return on capital. Return on capital, Yeah, the
- 6:45sources mentioned that metric. Specifically, Buffett contrasts
- 6:48a regular, run-of-the-mill bank with a company like American
- 6:51Express. Right, the American Express
- 6:52example is perfect. He points out that a normal bank
- 6:55might earn, say, a 13 or 14% return on its capital, but
- 7:00American Express earns 30% or more on its capital, and they do
- 7:05it without taking on any more risk than the bank.
- 7:08Let's unpack the mechanism behind that, because return on
- 7:11capital is basically the Holy Grail of investing for
- 7:14Berkshire. OK, let's hear it.
- 7:16Imagine you build a factory for $1 million, and every year that
- 7:21factory generates $100,000 in pure profit.
- 7:25That's a 10% return on capital. Simple enough.
- 7:27But if another company builds $1,000,000 factory and generates
- 7:30$300,000 in profit, they have a 30% return.
- 7:35The higher the return, the faster the business generates
- 7:38real, tangible cash that can then be reemployed to build even
- 7:42more factories. Oh, so it's the compounding
- 7:44effect? It's exactly yes.
- 7:46In the interview, Buffett brings up John D Rockefeller, pointing
- 7:50out that Rockefeller was perhaps the greatest historical investor
- 7:53purely because he just kept compounding his oil and gas
- 7:56profits at a massive rate of return over a huge stretch of
- 8:00time. Like a snowball rolling down a
- 8:02really long hill. Exactly.
- 8:04The profits fuel more growth, which fuels more profits.
- 8:08But wait, for decades the tech industry was defined by
- 8:10software, right? Which is like the ultimate high
- 8:13return business. It's asset light.
- 8:15You write a piece of code once in your dorm room, put it on a
- 8:18server and you can sell it a billion times with almost 0
- 8:21marginal cost. Right, it's incredibly
- 8:23efficient. And Wall Street merchandise that
- 8:25exact model because the profit margins were astronomical.
- 8:28So why did Buffett avoid that era of tech if the returns were
- 8:31so high? Because asset light businesses
- 8:34often lack a deep physical competitive mode.
- 8:37If it only takes a laptop to build a product, well, someone
- 8:40else with a laptop can potentially replicate it.
- 8:43Buffett prefers deploying massive amounts of capital into
- 8:46tangible assets, things that are incredibly difficult for a
- 8:49competitor to just copy. But the rise of artificial
- 8:51intelligence changed the rules of the game entirely.
- 8:54And this is where my brain broke a little bit reading these
- 8:56transcripts because Buffett explicitly compares the spending
- 9:00of companies like Google, Microsoft and Amazon on AI to
- 9:04Berkshire building the BNSF railroad.
- 9:07It's a striking comparison. I have to push back here though.
- 9:10A railroad. Google writes software and
- 9:13search algorithms. How on earth is a tech company
- 9:16comparable to laying 19th century steel tracks across a
- 9:19continent? It all comes down to CapEx
- 9:21capital expenditure. For a long time, tech companies
- 9:24focused heavily on OpEx operating expenses like paying
- 9:28salaries and renting server space.
- 9:30Right, keeping the lights on. But AI models are not just
- 9:33software. Training A large language model
- 9:35requires thousands of highly specialized, incredibly
- 9:39expensive microchips like GPU's running continuously for months.
- 9:43So. They can't just rent some extra
- 9:44cloud space anymore. No, not at this scale.
- 9:47They are being forced to lay out hundreds of billions of dollars
- 9:51to build massive bespoke data centers from the ground up.
- 9:55In the AI era, the data centers are the physical steel, and
- 9:59those microchips of the locomotive engines is incredibly
- 10:02capital intensive. So there really are railroads
- 10:05now. They are.
- 10:07Wall Street is still trying to analyze these tech giants as if
- 10:10they are asset light software companies, but they are rapidly
- 10:14mutating into the new heavy industry of the 21st century.
- 10:17They are laying digital steel tracks and what really stands
- 10:20out in the transcripts that these tech companies don't even
- 10:22necessarily want to play this capital heavy game.
- 10:25Buffin points out that these giants are essentially being
- 10:27forced to spend spend this money just to survive.
- 10:29It is an existential arms race. In the interviews, Buffett
- 10:33brings up IBM in the mid 20th century.
- 10:35IBM would have loved to just keep playing their old,
- 10:37comfortable, highly profitable game, but the market forced them
- 10:41to adapt to new computing realities.
- 10:43And then he brings up an even better historical parallel A&P
- 10:46grocery stores. A perfect example.
- 10:49I remember my grandparents talking about A&P, but they
- 10:52aren't around anymore. Right, because they didn't
- 10:54adapt. In the 1930s, RMP was the
- 10:57dominant grocery chain in America.
- 10:59They were a massive national powerhouse, earning high returns
- 11:02on capital. But they failed to adapt.
- 11:06The larger supermarket model emerged.
- 11:08They clung to their old way of doing things.
- 11:10Exactly, and their dominance just evaporated.
- 11:13Because the customer isn't dumb. If someone else offers a better
- 11:16result, the customer leaves. So Google and the other
- 11:19hyperscalers are spending hundreds of billions on AI data
- 11:22centers because if they don't, they become AMP grocery stores.
- 11:26They become obsolete. That is the brutal reality, and
- 11:30this is exactly where Wall Street's flaw gets exposed.
- 11:32How so? Well, Wall Street analysts are
- 11:35notoriously obsessed with short term free cash flow.
- 11:38They want to know what the earnings will be next quarter.
- 11:40When a tech company announces massive capital expenditures,
- 11:43Wall Street panics because that money isn't going into dividends
- 11:47or stock buybacks, right? Now, but Berkshire operates on a
- 11:50totally different timeline. Exactly, they are looking at the
- 11:53internal rate of return over a decade, not 1/4.
- 11:57They realize that AI has turned tech from a light, fluffy
- 12:01software game into a brutal capital heavy infrastructure
- 12:05game, and Berkshire understands infrastructure better than
- 12:09anyone on the planet. Which completely shatters the
- 12:11myth that Buffett just doesn't get technology.
- 12:14I mean in the interview he casually mentions that he
- 12:17actually started a tech company back in 1958.
- 12:20Yeah, data documents. Right, it was built specifically
- 12:23to capitalize on an IBM antitrust ruling.
- 12:25He understands the mechanics of tech perfectly fine, but he only
- 12:28invests when those underlying business mechanics matches
- 12:31philosophy of building a massive capital heavy Moat.
- 12:34But here's where it gets really interesting for me.
- 12:37If AI is the new heavy industry and these companies are building
- 12:40massive digital railroads across the country, what is the fuel
- 12:44powering those trains? That is the multibillion dollar
- 12:47question. We treat AI like it's magic.
- 12:50We sit on our phones, we ask a chatbot to write a funny poem
- 12:54about a dog, and we think it's just weightless data floating in
- 12:57a cloud somewhere. But somewhere a real river has
- 13:01to cool down a real server farm to generate that poem.
- 13:05Are we completely ignoring the physical footprint of the
- 13:07digital revolution? The broader public is certainly
- 13:10ignoring it, but Greg Abel is hyper aware of it.
- 13:13Remember his professional background.
- 13:15He's not a software guy. Not at all.
- 13:17Abel isn't a software engineer from Silicon Valley.
- 13:19He's been decades in heavy infrastructure, running
- 13:22operations at Kiewit and Berkshire Hathaway Energy.
- 13:25When he looks at the AI boom, he doesn't see lines of code.
- 13:27He sees concrete, steel, water rights and massive power lines.
- 13:32Power, that's the ultimate bottleneck, Abel says.
- 13:35It flat out in the transcript. The limiting factor for the
- 13:38entire AI build out isn't about whether we can manufacture the
- 13:41microchips fast enough or write smarter algorithms.
- 13:44It is energy, absolutely. Specifically, it is the physical
- 13:47time it takes to prepare the geographic sites and get the
- 13:50power grid in a position where it can actually serve these
- 13:53colossal data centers. Let's define who is building
- 13:56these centers. Just to be clear, the industry
- 13:58calls them hyper scalers. These are the massive tech
- 14:01companies, the Amazons, Microsofts and Googles of the
- 14:04world, who need to scale their computing power exponentially to
- 14:10train AI. And the energy demand for that
- 14:12is staggering. It's beyond staggering, Abel
- 14:15notes, that last year, in the state of Iowa alone, a massive
- 14:198% of Berkshire Hathaway Energy's total electricity load
- 14:24came from data centers 8. Percent.
- 14:26Just for data centers and they're seeing incremental
- 14:29requests coming in constantly. Wait, if they need the power,
- 14:31why can't they just buy some land and, you know, plug the
- 14:34data center into the grid? Because the existing power grids
- 14:37were not built to handle a single facility drawing the same
- 14:40amount of electricity as a medium sized city.
- 14:42To plug in a hyperscaler data center, the local utility often
- 14:46has to upgrade substations, build new high voltage
- 14:48transmission lines, and drastically increase power
- 14:51generation. Wow, that's heavy
- 14:52infrastructure. Yeah, and all of that costs
- 14:54hundreds of millions of dollars. Which brings up a huge issue.
- 14:58If the utility company spends half a billion dollars upgrading
- 15:01the grid for a Google data center, who pays for that?
- 15:05Does my grandma's electric bill go up just so a tech company can
- 15:08train its new language model? That is exactly the friction
- 15:11point and it's why Berkshire isn't just handing over the keys
- 15:15to the power grid. They have a very strict 3 part
- 15:18policy for dealing with these hyper scalers.
- 15:20Oh really? What are the rules?
- 15:22Well, they sat down with state governors and utility regulators
- 15:26and laid down the law rule #1 there has to be a net benefit to
- 15:30their everyday customers. Meaning the hyperscaler has to
- 15:33foot the bill for the infrastructure upgrades
- 15:35protecting the local residents from rate hikes.
- 15:37Precisely without that rule, everyday citizens would
- 15:41essentially be subsidizing the tech industry's multibillion
- 15:44dollar AI race. That makes total sense.
- 15:46And the second rule is about water, right?
- 15:48Why do these data centers need so much water?
- 15:51Heat. When you have thousands of
- 15:53specialized AI microchips running at maximum capacity 24
- 15:57hours a day, they generate an immense amount of heat.
- 16:00Right, like a laptop burning your legs, but scaled up a
- 16:02million times. Exactly, and if you don't cool
- 16:04them down they literally melt. Cooling towers require millions
- 16:09of gallons of water, so ABLE states that the hyper scalers
- 16:13have to strictly manage their water usage with the best
- 16:16available technology. They cannot be permitted to
- 16:19drain local aquifers dry, especially in areas that are
- 16:22prone to drought. Which brings us to the third
- 16:25rule, which Able implies might actually be the most difficult
- 16:28one. The community must actually
- 16:30welcome the data center. Berkshire expects these
- 16:33companies to be good neighbors. Yes, if the local community
- 16:36doesn't want a massive, humming industrial complex in their
- 16:39backyard, Berkshire isn't going to force the project through.
- 16:43But are people really fighting back against these sites?
- 16:46Absolutely. Abel notes that communities
- 16:48across the US are starting to organize and push back against
- 16:51these developments. You have noise pollution from
- 16:54the massive cooling fans, visual blight from the new transmission
- 16:57lines, and real concerns about local resource depletion.
- 17:00So it takes time to get approved.
- 17:02A lot of time getting the proper zoning laws passed, conducting
- 17:06environmental impact studies, and actually physically building
- 17:09the power lines. It all takes years.
- 17:12So the friction is entirely physical.
- 17:14It's the real world slowing down the digital world.
- 17:17Yes, now Abel mentions Berkshire hasn't had a specific site
- 17:21rejected yet, but he definitely acknowledges the growing tension
- 17:25and this whole situation highlights what we might call
- 17:28Berkshire's brilliant 2 prong strategy for the AI revolution.
- 17:32Oh, I see it. On one hand, they are taking a
- 17:35massive $31 billion position in Alphabet.
- 17:38They are investing in the companies building the AI.
- 17:41They are, you know, buying the gold miner.
- 17:43Exactly. But on the other hand, through
- 17:44Berkshire Hathaway energy, they own the underlying utility
- 17:47infrastructure. They own the actual electricity
- 17:50that the AI desperately needs to exist.
- 17:52They are selling the pickaxes. Yes, they own the digital
- 17:55railroad tracks, and they own the electricity that powers the
- 17:58locomotives. It has a master class in
- 18:00strategic positioning. It really is.
- 18:03They recognize that the nature of the game had fundamentally
- 18:06changed. The asset light software only
- 18:08era of tech is closing. We are back in an era of heavy
- 18:12physical capital expenditure and Berkshire is playing both sides
- 18:16of the board. Perfectly said.
- 18:18So let's trace this whole journey back because we've
- 18:20covered massive ground today. We started with a casual Sunday
- 18:24morning phone call that resulted in a $10 billion stock purchase
- 18:28executed in minutes because of a governance structure that just
- 18:31entirely bypasses corporate bureaucracy from there.
- 18:34We unpack the mechanics of return on capital and CapEx,
- 18:37realizing why Buffett finally broke his rule against tech.
- 18:41The AI boom isn't just software, it's a capital intensive arms
- 18:44race. The tech giants are laying out
- 18:46hundreds of billions in infrastructure just to avoid
- 18:49becoming the next obsolete A&P grocery chain.
- 18:52And finally, we hit the physical reality of the situation with
- 18:54Greg Abel. The entire digital revolution is
- 18:57currently bottlenecked by the physical world.
- 18:59The true limitations aren't in the algorithms.
- 19:01They were in zoning board meetings, local water rights and
- 19:05the capacity of the power grid in places like Iowa.
- 19:07For you listening, it's a perfect reminder of how to
- 19:11analyze any new technological trend.
- 19:14Don't just get mesmerized by the flashy software interface on
- 19:17your screen. Look deeper.
- 19:19Look at the physical supply chain, look at the
- 19:22infrastructure required to run it, and most importantly, look
- 19:25at the immense capital required to sustain it all.
- 19:28Because the ultimate magic trick of AI is making you forget that
- 19:32it's actually a heavy industrial process.
- 19:34And this raises a really important question, something we
- 19:37should all be pondering as this race accelerates.
- 19:39What is that? Well, these tech giants are
- 19:42essentially being forced to play this game.
- 19:44They have to spend hundreds of billions on capital expenditures
- 19:47just to stay relevant. But what happens if those
- 19:50massive internal rates of return that Wall Street is dreaming
- 19:53about don't actually materialize?
- 19:56If the AI products don't eventually generate enough cash
- 19:58to justify the massive cost of the infrastructure.
- 20:01Right. If the profits don't match the
- 20:03spending, who? Truly holds the power in that
- 20:06future. Is it the tech companies who
- 20:08raced to build the biggest, smartest AI?
- 20:11Or is it the utility companies who hold the absolute physical
- 20:14keys, the electricity required to keep the lights on?
- 20:17Well, the guy holding the power lines always wins.