Latest / Deep Values: Value Investing Deep Dives / Intrinsic Value and The Hidden Financial Risks of AAPL, MSFT, NVDA, MU, GOOGL
Transcript
- 0:00What if I told you the single biggest threat to Microsoft
- 0:02right now isn't a competitor releasing a faster AI model or
- 0:08even some sudden shift in consumer habits, but actually
- 0:12its own accounting department? I mean, it sounds totally absurd
- 0:15on the surface, right? Because we're talking about a
- 0:16company that is essentially printing money.
- 0:18Right, exactly. But, you know, when you look at
- 0:21how standard accounting handles the hundreds of billions of
- 0:25dollars currently being poured into artificial intelligence
- 0:27infrastructure, well, that threat becomes very, very real.
- 0:31Yeah, and welcome to the Dee dive, everyone.
- 0:33If you've been watching the stock market, you've undoubtedly
- 0:36seen the charts. I mean, everything is just
- 0:37moving up and to the right. Fueled mostly by the seemingly
- 0:41unstoppable gravitational pull of AI.
- 0:44Right. AI is everywhere, but today our
- 0:46mission is to cut through that daily market noise and really
- 0:49dissect the world's biggest tech Titans.
- 0:52We're looking past the hype to ask a pretty fundamental
- 0:55question. Are these companies actually
- 0:57sound investments at their current valuations, or are they
- 1:01hiding massive systemic risks behind all this AI buzz?
- 1:05Yeah. And to get to the bottom of
- 1:07that, we are stripping away the daily stock price fluctuations.
- 1:11We are looking at fundamental stock analysis, calculating
- 1:14intrinsic value and examining the real economic modes of five
- 1:18absolute behemoths. Right, we're talking Apple,
- 1:21Google, NVIDIA, Micron and Microsoft.
- 1:24OK, let's unpack this, starting with a company that almost
- 1:28everyone listening probably relies on every single day.
- 1:30And that's Apple. We tend to view Apple as this, I
- 1:33don't know, impregnable fortress.
- 1:35Oh, absolutely. And if you just glance at their
- 1:38recent earnings reports, I mean, everything looks fantastic.
- 1:41It really does, and it establishes a perfect baseline
- 1:44for how current market optimism can just completely mask the
- 1:47underlying realities. Yeah, Apple is fascinating right
- 1:50now because there's this massive disconnect between their core
- 1:54business growth and their actual stock performance.
- 1:57I mean, over the last five years, Apple's actual underlying
- 2:00business, you know, selling iPhones, Macbooks, services,
- 2:03that's grown at roughly 7% annually.
- 2:06Right, 7%, but the headline number everyone saw recently was
- 2:09this massive 22% spike in diluted earnings per share over
- 2:13the last quarter. Exactly.
- 2:15And on the surface a jump from 7% historical growth to 22%
- 2:20earnings growth. Well, it looks like a massive
- 2:23acceleration. It makes you think they just,
- 2:25you know, invented some revolutionary new product which.
- 2:29They haven't. That 22% figure is heavily
- 2:33engineered. I mean, it's a master class in
- 2:35capital allocation really rather than any sort of organic
- 2:38business expansion. Yeah, let's breakdown the
- 2:40mechanics of that engineering. Because it relies on stock
- 2:43buybacks, it's essentially like going to a classic car auction.
- 2:48OK, I like this analogy. Right, so the price of a vintage
- 2:50Mustang keeps hitting record highs week after week, but it's
- 2:54not because they suddenly dropped a totally new like 3
- 2:57times faster engine into the car.
- 2:59The car is exactly the same. Right.
- 3:01The asset hasn't fundamentally changed.
- 3:03Exactly. The price is going up simply
- 3:05because there's an aggressive bidding war.
- 3:07And in Apple's case, they are the ones bidding.
- 3:10They are injecting roughly $100 billion into massive stock
- 3:14buybacks, which acts as this relentless passive flow of
- 3:18capital. So mechanically, when a company
- 3:21buys its own shares and retires them, there are just fewer
- 3:25shares available on the open market.
- 3:27Right. So even if the company's total
- 3:29overall profit stays completely flat, the profit per share goes
- 3:33up because, you know, the pie is being sliced into fewer pieces.
- 3:35Which is wild. It mechanically pushes the stock
- 3:38price higher completely independent of how many iPhones
- 3:41are actually moving off the shelf.
- 3:42Exactly that. And what fundamental analysis
- 3:45deals here is that Wall Street's price targets for Apple seem to
- 3:49just be trailing this mechanically inflated stock
- 3:52price upward. Instead of predicting it based
- 3:54on real business fundamentals, right?
- 3:57So let's apply a strict value investing lens to this.
- 4:00Let's assume Apple Score business growth normalizes down
- 4:03to a more realistic 5% going forward.
- 4:06Which makes a lot of sense given the heavy market saturation and,
- 4:09you know, fierce global competition.
- 4:11Right, So what does the math actually look like then?
- 4:13Well, when you take those projected future cash flows at a
- 4:165% growth rate and you discount them back to today's value, the
- 4:20intrinsic value of Apple stock is only about 40% of its current
- 4:25trading product. Meaning there is a potential 60%
- 4:28downside risk just to get back to a fair fundamental value
- 4:33based on single digit business growth that that fundamentally
- 4:37changes how you look at their stock chart.
- 4:39It really does. It certainly puts recent
- 4:41institutional moves into perspective, too.
- 4:44I mean, when you see Warren Buffett significantly reducing
- 4:47Berkshire Hathaway's massive Apple position, it's a glaring
- 4:50signal. Yeah, the ultimate value
- 4:52investor is recognizing that the price has just become totally
- 4:55detached from the fundamental cash flow it generates.
- 4:57Exactly. Which brings up a really
- 4:59interesting contrast, actually, because while Apple's valuation
- 5:03looks transparently expensive once you dig into the buybacks,
- 5:06Google presents an entirely different kind of puzzle.
- 5:08Yeah, Alphabet, Google's parent company, looks deceptively cheap
- 5:12right now. I mean, they are trading at a
- 5:13price to earnings or PE ratio of around 26, right?
- 5:17And compared to the rest of Big Tech, a 26 multiple feels, you
- 5:22know, incredibly reasonable. Yeah.
- 5:24But what's fascinating here is how standard accounting rules
- 5:27can completely skew our perception of a company's core
- 5:31profitability, because that PE ratio of 26 is largely an
- 5:35illusion. Let's walk through this because
- 5:37the mechanics behind this illusion are just wild.
- 5:41So APEE ratio is simply the stock price divided by the
- 5:44company's earnings, right? Google reported trailing 12
- 5:47month net profits of roughly $160 billion.
- 5:51So price divided by 160 billion gets you that attractive 26
- 5:56multiple, but the catch is inside that 160 billion number.
- 6:00Right, exactly. You have to look at the line
- 6:01items. Of that 160 billion, 55 billion
- 6:05came from a category called other income.
- 6:07And to be completely clear, other income does not mean
- 6:10selling more search ads or getting more businesses onto
- 6:14Google Cloud or selling YouTube Premium subscriptions.
- 6:17No, not at all. It's strictly non operating
- 6:19income and in this specific case it's tied to an accounting rule
- 6:23regarding investments. See, Google made a massive early
- 6:27investment in a generative AI startup called Anthropic.
- 6:30Oh, right. And recently, Anthropic went
- 6:34through a new private funding round that bumped its valuation
- 6:38to a staggering $940 billion. And because of mark to market
- 6:43accounting rules set by the SEC, Google is legally required to
- 6:47record the increased value of their Anthropic shares on their
- 6:51own income statement. Yeah, think about it like owning
- 6:53a house. Imagine you bought a house for
- 6:55$300,000 and this year the market value jumped to 400,000.
- 7:00OK. Under Mark to market accounting,
- 7:02it's as if the IRS forced you to claim that $100,000 increase as
- 7:06actual income on your tax return.
- 7:08Even though you haven't sold the house and you haven't seen a
- 7:10single dime of actual cash, you're just wealthier on paper.
- 7:14Exactly. So Google didn't actually
- 7:15generate $55 billion in cash from that investment.
- 7:19It's an unrealized paper game, but it legally inflates the
- 7:22reported net income. Wow.
- 7:24So when you strip away that 55 billion accounting bonus,
- 7:28Google's true core business operating earnings are actually
- 7:32around 105 billion. Right.
- 7:34So what does this all mean for their valuation?
- 7:37Well, if we recalculate their multiple based on the real cash
- 7:40generating business of 105 billion, Google's true PE ratio
- 7:44leaps from 26 all the way up to 40.
- 7:48Oh wow. So the core business, the search
- 7:50engine, the cloud infrastructure, YouTube, it's
- 7:53priced significantly higher than the headline numbers suggest.
- 7:56Significantly higher, yeah. And this is happening at the
- 7:58exact moment their free cash flow is facing massive downward
- 8:03pressure. Because they're suddenly
- 8:04spending 10s of billions of dollars on capital expenditures,
- 8:07right? Exactly.
- 8:08Which begs the obvious question. I mean, where exactly is Google
- 8:10sending all those billions in capital expenditures?
- 8:13It's flowing directly into the second phase of our analysis,
- 8:16which is the hardware boom. Google, along with pretty much
- 8:19every other major tech company, is frantically building out AI
- 8:23data centers. And that money is landing
- 8:25squarely in the pockets of the semiconductor giants, primarily
- 8:29NVIDIA and Micron. OK, let's tackle the $5 trillion
- 8:33elephant in the room, NVIDIA. I mean a 36% compound annual
- 8:38growth rate since 1993. It's insane.
- 8:40They are the ultimate pick and shovel play for the AI gold
- 8:43rush. But I want to push back a little
- 8:46on this narrative of invincibility.
- 8:48OK, go for it. Because NVIDIA manufacturers a
- 8:51piece of physical silicon. It's hardware.
- 8:54And historically, the life cycle of highly profitable hardware is
- 8:58brutal. Someone eventually figures out
- 9:01how to build a, you know, slightly less powerful chip for
- 9:04half the price, and they undercut your margins.
- 9:06Aren't they incredibly vulnerable to commoditization?
- 9:09If they were just selling silicon, absolutely I would
- 9:11agree with you, but NVIDIA has engineered a wide and Whiting
- 9:15economic Moat that actually has very little to do with the
- 9:18physical chips themselves. Really.
- 9:20Yeah, their real power lies in their software platform known as
- 9:23CUA. OK.
- 9:24Let's demystify CUA for a second.
- 9:27How exactly does a software layer create an inescapable
- 9:30hardware monopoly? Well, CDA is the programming
- 9:33interface that allows developers to write software that actually
- 9:37utilizes the immense parallel processing power of Nvidia's
- 9:42GPU's. NVIDIA saw the writing on the
- 9:44wall and launched CGA way back in 2006.
- 9:48Wow, about long ago. Yeah, and over nearly two
- 9:51decades, they've cultivated an ecosystem of roughly 5.9 million
- 9:55developers globally. So these developers have spent
- 9:58years, sometimes their entire careers, building their AI tool
- 10:02sets, their mathematical models, and their workflows specifically
- 10:06using the CUDA language. Exactly.
- 10:08It creates an insurmountable switching cost.
- 10:11Imagine you are a major tech company.
- 10:13Even if a competitor like AMD comes to you tomorrow and offers
- 10:16a chip that is just as fast as Nvidia's, but let's say 30%
- 10:19cheaper. You can't just buy it and swap
- 10:21it out. No, because none of your
- 10:22software would run on it. Your 5.9 million developers
- 10:25would have to rewrite millions of lines of code to speak the
- 10:29new hardware's language. Right.
- 10:31And the labor cost, the delay in product launches, the sheer
- 10:34operational headache of rewriting that code that
- 10:39completely wipes out the 30% savings you got on the physical
- 10:42chip. Totally wipes it out.
- 10:44NVIDIA essentially acts as an inescapable toll booth for the
- 10:47entire AI industry because everyone is already locked into
- 10:50their Rd. system. That is fascinating, and that
- 10:53structural dominance leads to one of the most counterintuitive
- 10:57statistics we've seen. Usually when a stock price
- 11:00rockets upward as aggressively as Nvidia's has over the last
- 11:03few years, the PE ratio just goes completely through the
- 11:06roof, right? Right, making it wildly
- 11:08overvalued. Yet Nvidia's PE actually
- 11:11dropped. In 2020, their PE was hovering
- 11:14around 80, but recently, despite the stock price going parabolic,
- 11:17that multiple fell to around 32. Yeah, their stock got more
- 11:20expensive to buy, but the company actually became cheaper
- 11:23relative to its earnings. Because their earnings growth
- 11:26was so absurdly explosive, it just outpaced even the wild
- 11:29stock rally. Exactly.
- 11:31We're talking about them generating A staggering $48
- 11:33billion in free cash flow in a single quarter.
- 11:37In one quarter. But you know, no company
- 11:39operates without risk, and the analysis highlights a pretty
- 11:43glaring vulnerability in Nvidia's armor, which is extreme
- 11:46customer concentration. Yeah, this is the big 191% of
- 11:50Nvidia's revenue is now generated by their data center
- 11:53division, and a massive portion of that revenue relies on just
- 11:58four individual hyperscaler customers.
- 12:00Hyperscalers being companies like Microsoft, Amazon, Google
- 12:04and Meta. You know the companies building
- 12:06data centers the size of small cities.
- 12:08Right, So if even one of those 4 hyperscalers decides to tap the
- 12:12brakes on their AI infrastructure spending,
- 12:14Nvidia's growth profile takes an immediate massive hit.
- 12:17Which brings us to the other critical hardware player in this
- 12:20story, and that's Micron. Because you can't train a
- 12:23massive AI model using processing power alone.
- 12:26The processor needs to be fed data at lightning speed, which
- 12:29requires specialized high bandwidth memory.
- 12:32Right. And Micron's cloud memory
- 12:34business is currently growing at an explosive 200% rate. 200%.
- 12:38Yeah. And Micron operates in a
- 12:40fascinating market dynamic. It's essentially A3 player
- 12:44oligopoly alongside Samsung and SK Hynix.
- 12:48OK. Together those 3 control roughly
- 12:5095% of the global market for this type of memory?
- 12:53So that's a massive barrier to entry, but this introduces A
- 12:57crucial concept for anyone evaluating these semiconductor
- 13:01companies right now. It's the difference between a
- 13:03valuation bubble and an earnings bubble.
- 13:05Yeah, and most people instinctively understand
- 13:07evaluation bubble. I mean, think back to the.com
- 13:10boom in 2000. With a company like pets.com,
- 13:13the stock price goes to the moon based on hype, but there are
- 13:16absolutely no fundamental profits or underlying revenues
- 13:19to support it. Right.
- 13:20It's just pure speculation, a complete financial mirage.
- 13:23Exactly. But Micron is not an evaluation
- 13:26bubble. Their profits and revenues right
- 13:28now are very, very real. The danger they face is an
- 13:31earnings bubble. Meaning the profits exist today,
- 13:34but they are being driven by a temporary, unsustainable panic
- 13:37buying spree from their customers.
- 13:40Right. Historically, the memory chip
- 13:42market is violently cyclical. It operates on this massive boom
- 13:46and bust cycle. When a new technology emerges,
- 13:49cloud providers panic about running out of memory chips, so
- 13:52they double or triple their orders just to secure supply.
- 13:55Which creates a totally artificial demand signal.
- 13:57Exactly. The manufacturers see this huge
- 14:00demand, They ramp up production massively and they generate
- 14:04record profits. But eventually the cloud
- 14:06providers fill up their warehouses, they stop ordering
- 14:09and suddenly say the market is flooded with excess supply.
- 14:12And then prices crash and those astronomical profits just
- 14:15evaporate overnight. Precisely so.
- 14:17The core risk for Micron right now is that they are at the
- 14:21absolute peak of an over earning cycle.
- 14:24OK, so let's follow the money trail to its final destination.
- 14:27We've established that Google is spending heavily.
- 14:29NVIDIA and Micron are raking in historic profits by selling the
- 14:32hardware. But someone has to be footing
- 14:35the majority of this enormous bill.
- 14:37And that brings us back to Nvidia's single biggest
- 14:39customer, the heavyweed of the software world, Microsoft.
- 14:42Microsoft is truly the fulcrum of this entire AI narrative.
- 14:47I mean, on a surface level, just looking at their current
- 14:50financial statements, they appear completely unstoppable.
- 14:54The numbers are pristine, 15% overall revenue growth, 20% net
- 14:58income growth, and they're trading at a seemingly
- 15:01reasonable PE ratio of 23. Their intelligent cloud segment,
- 15:06which houses Azure, is growing at 30%.
- 15:09It just looks like a perfectly oiled cash printing machine.
- 15:12But OK, here's where it gets really interesting.
- 15:15Hidden beneath those glowing headline numbers is a
- 15:18fundamental structural mutation of their entire business.
- 15:21Yeah, the analysis highlights a truly staggering figure that
- 15:24most retail investors are completely overlooking right
- 15:27now. Microsoft is projecting to
- 15:29invest roughly $190 billion in capital expenditures for AI
- 15:34infrastructure this year alone. 190 billion.
- 15:37Yeah. To put that into perspective,
- 15:38that is almost twice their entire current annual net
- 15:41profit. Twice their net profit spent
- 15:43entirely on servers, chips and data centers in a single year.
- 15:47And this isn't just cash leaving the bank account today.
- 15:50This triggers a long term accounting consequence that
- 15:53fundamentally alters their profitability moving forward.
- 15:56Right. If we connect this to the bigger
- 15:58picture, we have to talk about the mechanics of depreciation in
- 16:02accounting. When a company spends 190
- 16:05billion on physical assets like servers and data centers, you
- 16:09don't just record that entire amount as an expense in year 1.
- 16:12Right, Let's break that down with a smaller example.
- 16:15Say a business buys a $10,000 server that will last for five
- 16:18years. They don't take a $10,000 hit to
- 16:21their profits today. They amortize it exactly.
- 16:24They record a $2000 depreciation expense every year for five
- 16:28years. It spreads the cost over the
- 16:30useful life of the equipment. So currently, Microsoft's annual
- 16:33depreciation cost, the amount hitting their income statement
- 16:36every year for past hardware purchases, is around $30
- 16:39billion. But because of this
- 16:41unprecedented $190 billion AI spending spree, that annual
- 16:47depreciation line item is mathematically guaranteed to
- 16:50explode. So.
- 16:51How high does it go? Well, projections show it
- 16:54leaping from 30 billion today to roughly 160 billion to 170
- 16:58billion by 2027 or 2028. Oh man.
- 17:02Yeah, that is a massive non negotiable expense that will hit
- 17:06their income statement every single year going forward,
- 17:10regardless of how well their software is actually selling.
- 17:12So what does this all mean for Microsoft's bottom line?
- 17:15I mean, they can't just absorb an extra 130 billion in annual
- 17:18expenses without finding new money somewhere.
- 17:21The math is just brutal. To simply justify this level of
- 17:25capital expenditure, to cover those incoming depreciation
- 17:28costs, to pay for the immense electricity and cooling required
- 17:31to run these new data centers, and to maintain their current
- 17:34pristine profit margins, well, Microsoft needs to generate
- 17:38roughly $300 billion in entirely new AI driven revenue over the
- 17:42next few years. 300 billion. That is practically the
- 17:46equivalent of building another massive Fortune 10 company from
- 17:49scratch just to break even on the hardware they're buying
- 17:52today. Exactly.
- 17:53This is the fundamental mutation we were talking about.
- 17:56Microsoft sent decades as a high margin, low CapEx software
- 18:00machine because the beauty of software is you write the code
- 18:03once and you sell it a billion times with almost zero
- 18:07additional cost. Right.
- 18:08But now they are morphing into a highly capital intensive heavy
- 18:14infrastructure business and historically the stock market
- 18:18assigns much lower valuation multiples to heavy
- 18:22infrastructure businesses. Because they are constantly
- 18:24forced to reinvest their cash just to replace aging equipment
- 18:27and stay competitive. Exactly.
- 18:29So if Microsoft's AI tools like their Copilot subscriptions
- 18:32don't generate hundreds of billions in fresh recurring
- 18:35revenue, that upcoming tidal wave of depreciation is going to
- 18:38severely compress their profit margins.
- 18:40It's just incredible when you piece the whole ecosystem
- 18:42together like this. It really is.
- 18:44So to recap the journey we've been on today, we are witnessing
- 18:47a high stakes, incredibly expensive AI arms race among the
- 18:51tech giants, the hardware sellers, NVIDIA and Micron.
- 18:55They're reaping the immediate, massive rewards as they sell the
- 18:59digital picks and shovels. But if the software giants like
- 19:02Microsoft and Google cannot figure out how to turn that
- 19:05wildly expensive infrastructure into massive sustained new
- 19:09revenue streams, the whole cycle could violently reverse to
- 19:13access more advanced researches, applications and our latest
- 19:16published personal intelligent multi agent Deep value
- 19:19intelligence. Welcome to explore and join deep
- 19:22values dot AI community track total market temperature.
- 19:26Welcome to Search and download iOS App Buffett Indicator from
- 19:29Apple App Store. It really is a precarious loop.
- 19:32I'm going to leave you with one final thought experiment,
- 19:34though. Something that wasn't explicitly
- 19:35modeled in the financial charts we looked at, but it's kind of
- 19:38the logical conclusion of all the spending.
- 19:40What if these massive tech companies end up in a brutal
- 19:42race to the bottom? Well, what if they massively
- 19:46overbuild this AI infrastructure driven mostly by FOMO, creating
- 19:50this massive glut of computing power?
- 19:52I mean, they could bankrupt their own profit margins in the
- 19:54process, but if compute power becomes practically free because
- 19:59of that massive oversupply, the ultimate winners of the AI
- 20:03revolution might not be Microsoft or Google at all.
- 20:07The real winners could be everyday consumers and the next
- 20:09generation of small startups, exactly because they will
- 20:12suddenly have access to practically unlimited hyper
- 20:16cheap AI computing power to build entirely new industries we
- 20:20haven't even thought of yet. I mean, the tech giants build
- 20:23the roads, take on all the debt, and someone else gets to drive
- 20:26on them for free. It's.
- 20:26A very real possibility. We assume because a company has
- 20:30a trillion dollar market cap that its foundation is somehow
- 20:33impenetrable. But as we've seen today, when
- 20:35you dig into the accounting mechanics, the buybacks and the
- 20:38sheer weight of these upcoming capital expenditure as well,
- 20:42that foundation is facing an unprecedented amount of stress.
- 20:45Yeah, it's up to you to decide if the math will hold up or if
- 20:48the cracks are already starting to show.
- 20:50Disclaimer All analysis is based on publicly available sources
- 20:53and may be subject to revisions or differing interpretations.
- 20:57All contents from this channel is for research and
- 20:59entertainment purposes only and does not constitute any
- 21:02investment or financial advice. Conduct your own due diligence
- 21:05before making investment financial decisions.
- 21:07Don't forget to like this video. Much appreciated.