Latest / Deep Values: Value Investing Deep Dives / Why Big Tech Net Income Is Deceptive? Who is the real big BOSS?
Transcript
- 0:00So I want you to imagine opening a quarterly earnings report,
- 0:03right, Right. Yeah.
- 0:04And you're looking at it and you see that a company made $112
- 0:10billion in a single quarter. Which is just a totally
- 0:14staggering number. Right.
- 0:16It sounds like this unstoppable juggernaut, just a business
- 0:19operating at a scale that you know completely defies history.
- 0:21Yeah, absolutely. But then you look closer, you
- 0:24check the cash flow statement, you look at their actual bank
- 0:27accounts, and you realize this really crazy truth.
- 0:30The phantom profit. Exactly $98 billion of that
- 0:34profit didn't actually exist in the form of cash.
- 0:36It was like the complete phantom.
- 0:38It's wild, right? It creates this fascinating
- 0:41scenario where the headline numbers you see on financial
- 0:44news networks are just entirely divorced from the physical
- 0:48reality of the business. Yeah, totally divorced.
- 0:50You have this massive divergence between gap accounting rules
- 0:54and, well, cold hard cash. And so today we are basically
- 0:59tearing down these financial illusions hiding inside Big Tech
- 1:03Welcome to this deep dive. Glad to be here.
- 1:05Our mission today is to decode the true underlying
- 1:09profitability of the most powerful companies on Earth.
- 1:12So we're talking about NVIDIA, Microsoft, Apple, Alphabet,
- 1:16Amazon, Meta and Micron. Right, the heavy hitters.
- 1:19Exactly. And we're using these incredibly
- 1:21detailed financial analysis that reconstruct trailing 12 month
- 1:26figures directly from their SEC filings.
- 1:28Which is really the only way to get to the truth.
- 1:30Yeah, because the goal is to move way past those flashy
- 1:33headlines and uncover who actually generates the most
- 1:36cash. You know who has the widest
- 1:39economic Moat and how these bizarre accounting quirks are
- 1:42currently hiding literally hundreds of billions of dollars.
- 1:45We are essentially going to perform financial forensics
- 1:47today. I.
- 1:48Love that financial forensics. Because I mean, relying on a
- 1:52simple bar chart of net income right now is the quickest way to
- 1:55just fundamentally misunderstand the modern tech economy.
- 1:59Totally. But before we begin
- 2:01disassembling these balance sheets, we do have some
- 2:03essential housekeeping we do to access more advanced researches,
- 2:06follow Guru's latest movement and leverage our latest DEEP
- 2:09values agents. Welcome to download Deep values
- 2:13APP from App Store to track stock market temperature.
- 2:16Welcome to download iOS app Buffett indicator from Apple App
- 2:20Store or Bubble Watch from Google Play.
- 2:23Links are in the description. And just a quick disclaimer, all
- 2:26analysis is based on publicly available sources and may be
- 2:29subject to revisions or differing interpretations.
- 2:32All contents from this channel is for research and
- 2:34entertainment purposes only and does not constitute any
- 2:37investment or financial advice or service.
- 2:40Conduct your own due diligence before making investment or
- 2:43financial decisions. OK, let's unpack this.
- 2:46We have to start with the company that prompted this
- 2:48entire analysis in the 1st place.
- 2:49NVIDIA. NVIDIA because it really sets
- 2:52the Gliss line for what an extraordinary operating business
- 2:54looks like in this current market environment.
- 2:57It really does. And the reconstructed trailing
- 2:5912 month figures which you know we'll just call TTM going
- 3:03forward to keep it simple, they are just staggering.
- 3:05We really are. NVIDIA generated approximately
- 3:08$197.6 billion in operating income and that was on 303
- 3:14billion in revenue. So doing the quick math on that,
- 3:17you arrive at a 65.2% operating margin.
- 3:21Which is insane. It is.
- 3:23And to truly grasp the gravity of that number, you know we have
- 3:27to look at the historical ceiling for corporate
- 3:29profitability, right? Like Microsoft, which is widely
- 3:33considered the greatest, most scalable software business ever
- 3:37created, sits at around a 47% operating margin.
- 3:40OK, 47. Yeah, and Meta is hovering near
- 3:4338%. Apple and Alphabet are around
- 3:4633%. So that implies NVIDIA is like
- 3:50breaking the fundamental laws of industrial physics?
- 3:52Pretty much, yeah. Because Microsoft sells
- 3:55software, right? It's code that costs essentially
- 3:570 to reproduce. Exactly 0 marginal cost.
- 4:00But NVIDIA sells physical, complex semiconductor systems.
- 4:05They require incredibly rare materials.
- 4:08Precision manufacturing, global shipping.
- 4:11So the fact that a hardware company is operating at a 65%
- 4:14margin, vastly outperforming pure software margins, it just
- 4:21feels like it should be physically impossible.
- 4:23I mean, it does feel impossible. Until you dissect their specific
- 4:27supply chain architecture, that is.
- 4:28OK, break that down for me. Well, when we traditionally
- 4:31think of a hardware company, the mental model is like a massive
- 4:35automotive plant or a steel mill.
- 4:37Right vertical integration. Exactly.
- 4:38You think of raw materials entering one side of a factory,
- 4:41massive overhead, thousands of union workers and finished goods
- 4:45rolling out the other side. Yeah.
- 4:46And that requires just constant heavy capital reinvestment.
- 4:49But NVIDIA isn't doing that at all.
- 4:51Right, they are the undeniable engine of the AI boom, right?
- 4:54They're selling the picks and shovels to the gold rush, but
- 4:56they have somehow managed to avoid actually buying the heavy
- 4:59machinery to build the gold mines.
- 5:01What's fascinating here is that they've achieved this state of
- 5:05almost perfect capital offloading.
- 5:07Capital offloading. Yeah, NVIDIA designs the silicon
- 5:10architecture. They write the CUDA software
- 5:12stack that makes the chips useful.
- 5:15But the 10s of billions of dollars required to actually
- 5:19fabricate those microscopic transistors?
- 5:22That capital burden is pushed entirely upstream to TSMC in
- 5:28Taiwan. Oh wow, so TSMC is the one
- 5:31taking the massive financial risk of building the physical
- 5:35foundries? Precisely.
- 5:36Wait, it's like, OK, it's like being a world class architect
- 5:39who designs a billion dollar skyscraper, but instead of
- 5:42taking on the risk of construction, you force the
- 5:44builders to buy all the steel and you force the buyers to
- 5:46secure the land. You just hand over the digital
- 5:49blueprints and collect a massive toll on the final value of the
- 5:52building. That is a perfect analogy.
- 5:55The beauty of that architect analogy is that it really
- 5:57highlights where the leverage sets, but we also have to look
- 5:59downstream because not only is NVIDIA pushing the manufacturing
- 6:03costs upstream to TSMC and memory suppliers like SK Hynix,
- 6:08but they are also pushing the deployment costs downscreen to
- 6:11the hyperscalers. You mean like Microsoft and
- 6:13Amazon? Exactly.
- 6:14Microsoft, Amazon and Alphabet are the ones spending hundreds
- 6:18of billions to build the physical data centers, secure
- 6:21the power grids, run the liquid cooling systems, all just a
- 6:25house Nvidia's chips. So Nvidia's just sitting there
- 6:28completely isolated in the middle of this value chain?
- 6:31Yep. Completely insulated.
- 6:33They're insulated from the physical capital expenditure on
- 6:35both sides while possessing this like absolute pricing power
- 6:39because there is literally no viable alternative to their
- 6:43product right now. Right.
- 6:44And the consequence of that isolation is that 65% margin.
- 6:47Which is just wild to think about.
- 6:49It is, however, any financial analyst looking at a 65% margin
- 6:54in a hardware business has to view it with a pretty healthy
- 6:58degree of skepticism. Regarding its permanence, right?
- 7:00Yes, exactly. Because capitalism kind of
- 7:02abhors a 65% margin. It really does.
- 7:04Like if you were making that much profit, it acts like a
- 7:07giant beacon, just inviting ferocious competition from every
- 7:11corner of the globe. Oh, absolutely.
- 7:13You mentioned pricing power due to constrained supply, but I
- 7:16mean supply chains eventually normalize, don't they?
- 7:19They always do TSMC builds more fabs.
- 7:23Competitors like AMD try to close the gap, right?
- 7:27And the hyperscalers themselves are heavily incentivized to
- 7:30design their own custom silicon just so they don't have to keep
- 7:33paying Nvidia's toll. So that brings up this huge
- 7:37tension when analyzing NVIDIA today, we are looking at
- 7:40absolute peak operating economics.
- 7:43We are, but if you try to extrapolate a 65% operating
- 7:47margin forward for the next 10 years, you're essentially
- 7:49betting against the entire history of hardware cycles.
- 7:52You'd be crazy to make that bet. Supply will catch up.
- 7:56Yeah, the initial infrastructure build out will transition into a
- 7:58maintenance phase. Yeah, the margin will almost
- 8:00certainly face compression. Even with that potential for
- 8:03future compression, though, that 197.6 billion in operating
- 8:09profit makes NVIDIA the most powerful engine of actual
- 8:13business operations right now, undeniably.
- 8:15Which makes this next finding in the source material so
- 8:19incredibly jarring. The alphabet situation.
- 8:22Yeah, because if NVIDIA generates 50 billion more in
- 8:25operating income than Alphabet, cause Alphabet is at what, 147.6
- 8:30billion, correct? So logic dictates NVIDIA should
- 8:33have a higher net income, right? I mean, operating income is just
- 8:37the profit from the core business, right?
- 8:39You subtract interest and taxes and you arrive at net income.
- 8:42It's usually a straight line down the income statement.
- 8:45But the TTM net income figures show the exact opposite.
- 8:48They do. Alphabets net income is a
- 8:50staggering 244.2 billion while NVIDIA sits at 192.9 billion.
- 8:57Alphabet somehow ends up with 50 billion more at the bottom line
- 9:01despite A vastly smaller operating profit.
- 9:04I mean, the math doesn't work unless there is something
- 9:06massive hiding between the operating line and the net
- 9:09income line. And what's hiding there is a
- 9:11relatively recent and highly controversial gap accounting
- 9:15rule. OK, tell me about this rule.
- 9:16It's all about non marketable investments.
- 9:20To really understand the mechanics of this distortion,
- 9:23the sources point us to a specific slice of time.
- 9:26Which is. Alphabets Q22026.
- 9:29Income statement. OK, so let's isolate that
- 9:31quarter. Alphabets, core business search,
- 9:34YouTube, cloud infrastructure, they brought in about 119.8
- 9:38billion in revenue. Right after paying for servers,
- 9:41employee salaries and R&D, they reported an operating income of
- 9:4440.7 billion, which seems like a very healthy, very predictable
- 9:49quarter for a tech monopo. It is a spectacularly strong
- 9:52quarter of actual business operations, yeah.
- 9:54But as you move down the income statement past that operating
- 9:57profit, you hit a line item broadly categorized as other
- 10:01income. Other income, yeah.
- 10:04For that single three month period, Alphabet recorded other
- 10:07income of $97.98 billion. I just want to pause on that
- 10:13number for a second. They made 41 billion running the
- 10:16most utilized digital infrastructure on the planet,
- 10:20and then they just casually tacked on nearly 100 billion in
- 10:23other income in 90 days. It's wild.
- 10:27That brought their net income for the quarter to 112.19
- 10:32billion. And the immediate assumption
- 10:34most people make when they see an anomaly like that is that a
- 10:37massive corporate asset was liquidated.
- 10:40Right. They assume Alphabet must have
- 10:42sold off a division like Waymo or spun out a subsidiary for a
- 10:45mountain of cash. But no cash changed hands at
- 10:49all. Wait, really?
- 10:50None. They didn't sell a single share
- 10:51of anything. So how does that work?
- 10:53This is the mark to market accounting rule in action.
- 10:56OK, here's where it gets really interesting.
- 10:58We can think of it like opening the Zillow app on your phone.
- 11:01OK, the Zillow app I follow. Imagine you own a house and one
- 11:06day you open the app and see the estimated value jumped by 100
- 11:10grand because your neighborhood became suddenly trendy.
- 11:13Nice. Under normal rules, you don't
- 11:14pay taxes on that, and you certainly don't tell your boss
- 11:17you made an extra $100,000 this year.
- 11:19Right, because it's just a number on a screen.
- 11:21Exactly. But this accounting rule forces
- 11:24Alphabet to treat that Zillow app estimate as actual reported
- 11:29income. It's basically a mandated
- 11:31financial hallucination. That's a great way to put it.
- 11:34Because historically, companies only recorded a gain on an
- 11:37investment when they actually sold it and realized the cash.
- 11:41Yes, but the financial accounting standards, GORD or
- 11:44FASB, they changed the rules. Why?
- 11:48Well, they wanted more transparency in the corporate
- 11:50portfolios. So they mandated that equity
- 11:52investments must be marked to their fair value at the end of
- 11:55every single quarter. Oh, wow.
- 11:57And those paper gains or losses must flow directly through the
- 12:01net income line. I mean, I see the intention
- 12:03transparency is generally good, but the execution creates this
- 12:07massive blind spot. It really does.
- 12:09Because where is this specific 98 billion actually coming from?
- 12:14The sources indicate Alphabet holds is roughly 94.1 billion in
- 12:18SpaceX equity and they also have 124.3 billion in non marketable
- 12:24equity securities, which everyone basically understands
- 12:27is dominated by their stake in Anthropic, the AI company.
- 12:32So we have two highly illiquid private companies here.
- 12:36SpaceX launches A revolutionary new rocket, or Anthropic
- 12:40releases a new language model, and they raise a new round of
- 12:43private venture capital at an astronomical valuation right the
- 12:47moment that private funding round closes.
- 12:49Alphabet as accountants are legally required to look at
- 12:52their existing shares, apply the new valuation and write down the
- 12:56difference as net profit. But I really have to push back
- 12:58on this because a private venture round is incredibly
- 13:01finicky. What if only a small handful of
- 13:03investors actually bought into that new and profit valuation?
- 13:07What if the broader market wouldn't actually pay that price
- 13:10if Alphabet tried to sell their entire $124 billion state
- 13:13tomorrow? You've hit on the exact danger
- 13:16of mark to market accounting for illiquid assets.
- 13:19A private valuation is often set by the last dollar in.
- 13:22Right. The marginal buyer, yes, yeah.
- 13:24A venture capital firm might put 500 million into Anthropic at a
- 13:2940 billion valuation. The accountants extrapolate that
- 13:33valuation across Alphabet's entire holding.
- 13:36But if Alphabet actually tried to dump 10s of billions of
- 13:40dollars of anthropic stock on the secondary market, the price
- 13:44would collapse instantly. The liquidity just doesn't exist
- 13:47to support the paper valuation. Which means that 112 billion net
- 13:51income figure is incredibly fragile.
- 13:53Highly fragile. If the tech startup bubble
- 13:55bursts, or if the broader market just has a sudden contraction,
- 13:59the Zillow estimate goes down. And when it goes down, it flows
- 14:02through the income statement as a massive phantom loss.
- 14:05This creates violent non economic volatility in the
- 14:07earnings reports. I can imagine.
- 14:09To illustrate just how detached this is from cash, the sources
- 14:12highlight that in the first half of 2026, Alphabet recorded
- 14:16roughly 136 billion in total equity gains.
- 14:20Out of that staggering sum, only 836 million was actually
- 14:24realized through sales. So roughly 99.4% of that profit
- 14:29was completely theoretical. Completely.
- 14:31Theoretical. This is why any investor or
- 14:34analyst really has to treat the cash flow statement as their
- 14:37primary investigative tool. Because if you look at Alphabets
- 14:41income statement, you see hundreds of billions in profit.
- 14:45But if you Scroll down to the cash flow statement, the very
- 14:48first section, cash from operating activities, it
- 14:51includes a massive reconciliation line.
- 14:54The accountants literally have to subtract 135.8 billion back
- 14:58out of the net income because they know no actual cash entered
- 15:02the bank accounts. The cashless statement strips
- 15:04away the accounting fiction and returns us to reality.
- 15:08And you know this is not an isolated phenomenon either.
- 15:11The entire big tech cohort is cross pollinated with venture
- 15:14capital investments, meaning their net income figures are all
- 15:17polluted by the exact same distortion.
- 15:20Yeah, the sources show Amazon reported 53.4 billion in non
- 15:24operating income, heavily driven by their own massive investment
- 15:28in Anthropic. Microsoft saw a 4.9 billion bump
- 15:32from open AI valuations. Even NVIDIA, which already has
- 15:36exclusive operating income, recorded 23.7 billion in equity
- 15:41gains to further inflate their net income.
- 15:44So if you are trying to rank these companies by
- 15:46profitability, net income is temporarily useless.
- 15:50Yeah, it's just noise. It is too heavily skewed by
- 15:53venture capital markups and as we established earlier,
- 15:56operating income doesn't account for the massive physical
- 15:59investments required to keep the lights on.
- 16:01Which means we need a metric that bypasses the accounting
- 16:04noise entirely. We need to track the raw
- 16:06velocity of cash exactly. We need to look at free cash
- 16:09flow. Free cash flow is the great
- 16:10equalizer here. It measures the actual cash
- 16:13generated by the business operations minus the capital
- 16:16expenditures, the property, plant and equipment required to
- 16:19maintain those operations. The PP and.
- 16:21Yes, it is the money that can actually be distributed to
- 16:25shareholders, used for buybacks or just stored in the treasury.
- 16:29And if we strip away the Zillow app Paper gains right and look
- 16:33purely at TTM free cash flow, the hierarchy of big tech
- 16:37reshuffles completely. It really does.
- 16:39Because it isn't NVIDIA despite their 65% operating margin.
- 16:43According to the reconstructed data, the absolute king of free
- 16:47cash flow generation is Apple. Apple.
- 16:49Yes, Apple. Apple is pulling in
- 16:51approximately 136.7 billion in TTM free cash flow.
- 16:56They are beating NVIDIA by nearly 10 billion in pure cash
- 17:00generation. Which is incredible.
- 17:02It is, and the mechanics of how Apple pulls this off require us
- 17:05to look at their cash conversion cycle, which is frankly, a
- 17:08masterpiece of supply chain leverage.
- 17:10It really is a masterpiece. Let's look at the first nine
- 17:13months of Apple's fiscal year 2026.
- 17:15OK. The operating cash flow, the
- 17:17sheer volume of cash entering their accounts from selling
- 17:20iPhones, Macbooks, iPads, App Store subscriptions was 117.0
- 17:25billion. That's the gross cash coming in
- 17:27the front door. But the metric that truly
- 17:29defines Apple's dominance is what goes out the back door to
- 17:33actually build those devices. To generate that 117 billion in
- 17:37operating cash, Apple spent only 6.8 billion on property, plant
- 17:42and equipment. 6.8 billion in capital expenditures for a
- 17:47company moving hundreds of millions of physical, highly
- 17:50sophisticated electronic devices across the globe every single
- 17:55year. It's just a microscopic capital
- 17:57intensity ratio. It is.
- 17:59I really have to put that in perspective for you, the
- 18:01listener, because it sounds absurd.
- 18:03A traditional physical manufacturer would have to spend
- 18:0510s of billions of dollars constantly upgrading factory
- 18:09tooling, maintaining robotic assembly lines, building new
- 18:12warehouses just to maintain current production levels.
- 18:15Right, Let alone trying to grow. So how does Apple sell a
- 18:18physical product but completely bypass the physical cost?
- 18:21They utilize the same architectural offloading as
- 18:23NVIDIA, but they've been perfecting it for decades.
- 18:27If we connect this to the bigger picture, Apple doesn't own the
- 18:30massive, sprawling factory cities in Shenzhen or India.
- 18:35Foxconn and Pegatron take on the brutal low margin, high CapEx
- 18:39burden of physical assembly. And Apple doesn't own the
- 18:42semiconductor fabs either. TSMC spends the 20 billion
- 18:46required to build a single foundry for Apple's M series
- 18:49chips. So Apple's balance sheet is
- 18:52incredibly late. Incredibly.
- 18:53Late They own the intangible alchemy that makes the glass and
- 18:57metal valuable. They own the iOS ecosystem,
- 19:00which creates massive switching costs for consumers.
- 19:03They own the brand identity, the industrial design, the custom
- 19:06chip architecture and the direct retail relationship with the
- 19:09customer. This dynamic is the very
- 19:11definition of a Buffett quality business, right?
- 19:14Warren Buffett has historically searched for compounders,
- 19:17businesses that can increase their profits without requiring
- 19:20massive, continuous infusions of capital.
- 19:22Right, because if a business needs a dollar of capital to
- 19:25generate a dollar of growth, it is basically on a treadmill.
- 19:28It's exactly a treadmill. Apple has figured out how to
- 19:31extract massive growth while keeping their capital
- 19:34expenditures relatively flat. They just let their suppliers
- 19:38run on the treadmill while they collect the tickets at the door.
- 19:40That's a great way to say it. And there's another fascinating
- 19:42layer to this, which is how Apple handles working capital.
- 19:46Because of their massive leverage over suppliers like
- 19:49Foxconn, Apple can demand incredibly favorable payment
- 19:53terms. They might sell an iPhone to a
- 19:56customer today, collect the cash instantly, but not actually have
- 20:00to pay Foxconn for the assembly of that phone for 90 or 120
- 20:04days. And that creates a negative cash
- 20:06conversion cycle. Apple is effectively using its
- 20:10own supply chain as a free short term financing mechanism.
- 20:14That's brilliant. They collect the cash, they earn
- 20:16interest on it, and they pay their bills months later.
- 20:19When you combine that working capital dominance with
- 20:21incredibly low PP and E spending, you get a business
- 20:26that converts almost all of its operating profit directly into
- 20:29distributable free cash flow. It's the ultimate cash
- 20:31harvesting machine. It really is.
- 20:33But the flip side of that equation is equally important.
- 20:36If Apple's lack of fiscal spending makes their free cash
- 20:39flow look like a masterpiece, what happens to the companies
- 20:43who are currently trapped in the most expensive arms race in
- 20:46human history? What you are referring to the AI
- 20:50infrastructure build out? I am.
- 20:52And when you apply this exact same free cash flow metric to
- 20:55companies like Alphabet, Amazon and Meta, the optical illusion
- 21:00reverses. They suddenly look like
- 21:02deteriorating businesses. OK, Life dissect this AI CapEx
- 21:06mask because the headline numbers for these companies look
- 21:09terrifying if you don't understand the mechanics they
- 21:12do. We established earlier that
- 21:13Alphabet has 147.6 billion in operating profit.
- 21:17A spectacular number. But their TTM free cash flow is
- 21:20only about 53.3 billion. That is a massive Cliff.
- 21:25Their free cash flow margin plummets to just 11.9% and.
- 21:28The cash isn't vanishing, right? It is being buried in the ground
- 21:32in the form of concrete, fiber optic cables and highly
- 21:35specialized liquid cooled data centers.
- 21:38In the first half of 2026 alone, Alphabet spent 80.6 billion on
- 21:42PP and E. They doubled their capital
- 21:44expenditures compared to the exact same period the previous
- 21:48year. 80 billion in physical infrastructure in six months?
- 21:51Yeah, that is like the GDP of a small country being poured
- 21:55directly into servers. It is.
- 21:57And somehow Amazon's numbers make Alphaget look conservative.
- 22:01Oh, Amazon's numbers. Amazon generated 3/4 of a
- 22:04trillion dollars, 775.7 billion in trailing 12 month revenue,
- 22:10but their free cash flow it is -7.6 billion.
- 22:15A company doing 775 billion in top line revenue is literally
- 22:19bleeding fast. Why?
- 22:21Because their net PP and E purchases exploded to 169
- 22:25billion, that is a level of capital expenditure that is
- 22:29difficult to even contextualize. So if a novice investor is
- 22:32looking at a stock screening tool, right, and they filter for
- 22:35free cash flow and operating margins, they see Apple
- 22:39generating 136 billion in pure cash and they see Amazon with
- 22:43negative cash flow and an operating margin sitting around
- 22:4612%, right. The natural deduction is that
- 22:49Amazon is a fundamentally flawed, low quality business.
- 22:53And that deduction would be a catastrophic analytical error.
- 22:55Because it treats Amazon as a monolith.
- 22:58Exactly. Amazon is not a single business
- 23:00entity. It is a chimera.
- 23:02It is 2 completely different economic models forcibly fused
- 23:06together on one consolidated income statement.
- 23:09You have the e-commerce retail logistics network on one hand,
- 23:12and you have Amazon Web Services, the cloud computing
- 23:14division, on the other. Precisely.
- 23:16And when you analyze the consolidated numbers, the brutal
- 23:19capital intensity of the retail delivery network, the thousands
- 23:22of fulfillment centers, the fleets of airplanes, the
- 23:25hundreds of thousands of delivery vans, it all drags down
- 23:28the overall margins, right. But if you perform a structural
- 23:30carve out and isolate AWS, the narrative changes instantly.
- 23:35Yeah. The sources show that AWS alone
- 23:38generated roughly 148 billion in TTM revenue.
- 23:42And on that cloud revenue, they pulled in about 55 billion in
- 23:46operating income. Which yields an operating margin
- 23:48of approximately 37%. That's huge.
- 23:52AWS is a deeply entrenched elite software and infrastructure
- 23:56business operating at margins that rival Meta and Eclipse
- 23:59Apple. But because it is housed inside
- 24:02a company that is also trying to deliver paper towels to your
- 24:04doorstep in two hours, the true quality of the AWS engine is
- 24:08heavily masked. That is fascinating.
- 24:11And we can apply the same dual nature analysis to Meta, can't
- 24:14we? We can.
- 24:15Their consolidated operating margin is around 38%, but that
- 24:18high is a massive internal wealth transfer.
- 24:21Meta is essentially running a highly profitable monopoly to
- 24:24fund an enormously expensive startup.
- 24:27OK, break that down. In the first half of 2026,
- 24:29Meta's family of apps, which is the digital real estate of
- 24:32Facebook, Instagram and WhatsApp, generated A staggering
- 24:3650.3 billion in operating profit.
- 24:39Which is just a massive cash cow.
- 24:41It is a digital advertising cash cow with incredibly low marginal
- 24:45costs. Every time someone scrolls
- 24:47Instagram, it costs Meta fractions of a penny, but they
- 24:50charge advertisers a premium. Exactly.
- 24:53But they aren't dropping that 50 billion to the bottom line for
- 24:55shareholders. No, they are diverting that
- 24:58river of cash to heavily finance Reality Labs.
- 25:02They're augmented and virtual reality division alongside their
- 25:06massive AI supercomputer build outs.
- 25:09Wow. In that exact same 6 month
- 25:11period, Reality Labs posted an operating loss of 8.65 billion.
- 25:16So Mark Zuckerberg is basically taking the cash generated by
- 25:19your Instagram feed and incinerating it to build a
- 25:22hypothetical future computing platform.
- 25:24Pretty much. Which brings us back to the core
- 25:26flaw of using standard free cash flow to judge these companies.
- 25:29Right now, traditional free cash flow formulas penalize all
- 25:33spending equally. The standard accounting formula
- 25:35for free cash flow simply takes operating cash flow and
- 25:39subtracts total capital expenditures, right?
- 25:41It does not different differentiate between the intent
- 25:43behind the spending and this is where elite financial analysis
- 25:48requires distinguishing between two critical concepts,
- 25:52maintenance CapEx and growth CapEx.
- 25:54This is a vital distinction. Let's break down the mechanics
- 25:56here. Maintenance CapEx is basically
- 25:58the mandatory toll you pay just to stop your business from
- 26:02decaying. It is the cost of standing still
- 26:05for Amazon's retail business. Maintenance CapEx is replacing a
- 26:09broken conveyor belt, repaving a warehouse parking lot, or buying
- 26:12new tires for the delivery vans. For a cloud provider.
- 26:16It is replacing servers that have reached the end of their
- 26:19useful life. If you do not spend maintenance
- 26:22CapEx, your revenue will physically shrink as your
- 26:25infrastructure collapses. It's defensive spending.
- 26:27Yes, purely defensive. Growth CapEx, on the other hand,
- 26:30is entirely offensive growth. CapEx is discretionary.
- 26:34It is Alphabet deciding to build a brand new multibillion dollar
- 26:38nuclear powered data center to train the next generation of
- 26:42generative AI models. They don't have to build it to
- 26:44maintain their current search revenue.
- 26:47They were choosing to sacrifice current cash flow to capture
- 26:50future market share. So by indiscriminately deducting
- 26:53all of that growth CapEx, the traditional free cash flow
- 26:56formula makes Meta, Alphabet and Amazon look artificially sick
- 27:01today, when in reality they might be building an
- 27:04unassailable fortress for the next decade.
- 27:06Which forces you to make a philosophical judgement call.
- 27:10When you look at Amazon bleeding cash, you have to ask, is this
- 27:14visible profitability being suppressed by brilliant
- 27:18forward-looking reinvestment right?
- 27:20Or are they getting swept up in a technological hype cycle,
- 27:23burning billions on AI infrastructure that will never
- 27:27generate A sufficient return on invested capital?
- 27:29That's the billion dollar question.
- 27:31Exactly, if you believe it's brilliant reinvestment then the
- 27:34current free cash flow numbers are completely useless for
- 27:37valuing the long term equity. That is a perfect transition
- 27:39because we've been talking about how high capital expenditure can
- 27:43mask a truly phenomenal business.
- 27:45But what about the inverse? What happens when a business
- 27:47currently displays explosive margins and massive profits, but
- 27:52underneath the surface the structural foundation is highly
- 27:55unstable? This introduces the danger of
- 27:58the cyclical trap, and the prime case study provided in our
- 28:01sources is Micron Technology. Let's look at Micron, because if
- 28:05you just glance at their current trailing numbers, they look like
- 28:08they should be the most valuable company on the planet.
- 28:11They really do. Their raw operating margin is
- 28:1365.6%. Which slightly edges out
- 28:17Nvidia's margin actually. Yeah.
- 28:19And their top line growth is practically vertical.
- 28:22Micron's Q3 revenue exploded to 41.5 billion.
- 28:27Wow. To put that in context, just one
- 28:29year earlier, in the exact same quarter, they generated only 9.3
- 28:34billion. They more than quadrupled their
- 28:37revenue in 12 months while maintaining a 65% margin.
- 28:41Unbelievable. If you put those numbers in a
- 28:43vacuum, you would crown Micron the undisputed king of the tech
- 28:46sector. But you cannot analyze A
- 28:48semiconductor company in a vacuum.
- 28:51Memory chips, specifically the DRAM and NAND flash memory that
- 28:55Micron produces, behave fundamentally like a commodity,
- 28:58right? And commodity markets operating
- 29:00within an oligopoly are notorious for vicious boom and
- 29:03bust cycles. I really want to dig into the
- 29:04mechanics of that cycle because it explains why Wall Street
- 29:07refuses to give Micron the same valuation multiple as an Apple
- 29:11or a Microsoft. The cycle is essentially the
- 29:14bullwhip effect in global supply chains, right?
- 29:16Exactly. It always begins with a demand
- 29:18shock. Right now, the AI revolution
- 29:21requires an unprecedented amount of high bandwidth memory to feed
- 29:24the NVIDIA GPU's. Suddenly, the hyperscalers are
- 29:27ordering millions of memory chips.
- 29:29But Micron can't just flip a switch and double production by
- 29:32Friday. Fabricating memory requires
- 29:34multibillion dollar clean rooms, ASML lithography machines, and a
- 29:38production cycle that takes months from silicon wafer to
- 29:40finish chip. So it's slow.
- 29:42Very slow supply is highly inelastic in the short term, so
- 29:46massive demand meets constrained supply and the inevitable result
- 29:50is that prices skyrocket. And because the fixed cost of
- 29:54running a semiconductor fab are already paid for, every extra
- 29:58dollar of price increase falls straight to the bottom line.
- 30:01That's how you get a 65% operating margin.
- 30:05They are in the absolute peak of the boom phase.
- 30:08But the boom plants the scenes of its own destruction.
- 30:11Micron, Samsung and SK Hynix see these massive 65% margins and
- 30:17they rationally decide to expand capacity.
- 30:20Of course, they take their billions in profit and break
- 30:22ground on new fabrication plants.
- 30:24And it takes two or three years to build those plants, so the
- 30:27supply response is delayed. But eventually all those new
- 30:30factories come online simultaneously.
- 30:32Suddenly, the market is flooded with new memory chips.
- 30:35Supply catches up with and eventually overtakes demand.
- 30:39And in a commodity market, the moment there is oversupply,
- 30:42pricing power evaporates. It just vanishes.
- 30:45Prices crash, inventory piles up, and those 65% margins can
- 30:49violently compress down to single digits or even turn into
- 30:52operating losses. So what does this all mean?
- 30:54It's like a localized economic weather event.
- 30:57Imagine a surf shop on a beach during a freak month long heat
- 31:01wave where the temperature hits 105° every single day.
- 31:05The shop is selling out of surfboards and sunscreen at a
- 31:09400% markup. They're raking it in.
- 31:12They are making money hand over fist.
- 31:14Their margins look like apples. A very apartment analogy for the
- 31:17cyclical peak. But if you are a private equity
- 31:19firm looking to buy that surf shop, you wouldn't value the
- 31:22business by taking in that single month of heat wave
- 31:25profits and multiplying it by 12.
- 31:26Definitely not. You know winter is coming.
- 31:29The heat wave is an anomaly. The baseline economics of the
- 31:32surf shop are much lower. And that is the cyclical trap.
- 31:36The current AFL environment might structurally raise the
- 31:39floor for Micron's baseline over the next decade, but an
- 31:42intelligent analyst must attach a significantly lower multiple
- 31:46to $1.00 of peak cyclical operating earnings from Micron
- 31:50than they would to $1.00 of recurring durable earnings from
- 31:54a structurally insulated business.
- 31:56Which sets up the perfect contrast?
- 31:58If Micron represents the volatility of the cyclical heat
- 32:01wave, who represents the durable baseline?
- 32:04Well, if we threw out current valuation and just look at the
- 32:07underlying structural economics of the business model, who has
- 32:10the widest Moat? Yeah, who is it?
- 32:13The sources and most historical financial analysis point to
- 32:17Microsoft as the gold standard for durability.
- 32:20Over Apple? Over NVIDIA?
- 32:22Why does Microsoft win the durability test?
- 32:25Because of the nature of enterprise switching costs,
- 32:29Microsoft does not rely on consumer hardware upgrade cycles
- 32:32like Apple. They do not rely on volatile
- 32:35commodity pricing like Micron. Their profits are generated by
- 32:38an incredibly diversified, deeply entrenched collection of
- 32:41recurring software subscriptions, like what we are
- 32:43talking about Azure Cloud computing, Microsoft 365,
- 32:47Windows, GitHub, Active Directory, and cybersecurity
- 32:50solutions. They supply the invisible
- 32:52plumbing that allows the modern corporate world to function, and
- 32:55the key difference is how these products behave during
- 32:57macroeconomic stress. Let's imagine a severe global
- 33:00recession hits tomorrow. Interest rate spike.
- 33:03Consumer spending plummets. What happens to these tech
- 33:06giants? If a recession hits, Alphabet
- 33:08might see digital advertising budget slashed overnight.
- 33:11Meadow will see similar ad revenue contraction.
- 33:14Amazon might see a massive drop in e-commerce volume, right?
- 33:18Even the hyperscalers might decide they can delay purchasing
- 33:21the next $5 billion cluster of NVIDIA GPU's by a few quarters
- 33:25to preserve cash growth. CapEx is the first thing cut in
- 33:29a recession. But what happens to Microsoft?
- 33:32Can a Fortune 500 company just pause their Microsoft 365
- 33:35subscription to save money? It is practically impossible.
- 33:39Imagine ACEO walking into a board meeting and announcing to
- 33:42whether this recession we are uninstalling Word, Excel,
- 33:45Outlook teams, and our entire Azure cloud infrastructure
- 33:49effective immediately. Everyone would go back to
- 33:51physical ledgers and typewriters.
- 33:52The company would cease to function by noon.
- 33:55The switching costs are insurroundable.
- 33:57You can't migrate 10s of thousands of employees to a new
- 34:00operating system or e-mail client without grinding
- 34:02productivity to a halt. Microsoft is deeply embedded in
- 34:05the muscle memory of the global workforce.
- 34:08Entire custom enterprise applications are built
- 34:11specifically depending on Microsoft's back end
- 34:14infrastructure. That integration creates an
- 34:16extraordinary economic Moat. It provides pricing power that
- 34:21persists even through recessions.
- 34:23It does. So while NVIDIA might possess
- 34:25superior raw economics today due to supply constraints, Microsoft
- 34:30possesses vastly more predictable economics over a
- 34:33full 10 year cycle. Exactly.
- 34:35This reinforces A crucial principle for you, the listener.
- 34:38Profit size does not equal profit quality.
- 34:41A dollar of highly cyclical hardware profit simply is not
- 34:44worth the same as a dollar of sticky, highly integrated
- 34:48enterprise software subscription profit.
- 34:50OK. Having established this
- 34:52incredibly nuanced framework for analyzing tech companies, I
- 34:55mean, we've stripped up paper gains, distinguished between
- 34:57growth and maintenance CapEx and factors in cyclicality and
- 35:00switching costs. We've covered a lot.
- 35:02We have, but now we have to address a glaring error that
- 35:05frequently corrupts financial media.
- 35:07We must address the danger of bad comparisons.
- 35:11The sources specifically call out a viral financial graphic
- 35:15that attempted to compare the big tech companies we've been
- 35:17discussing with major banks and holding companies using the
- 35:21exact same metrics on a single bar chart.
- 35:24Right. The graphic confidently labeled
- 35:26JP Morgan as having 84 billion in operating income and Bank of
- 35:31America as having 42 billion, putting their bars right next to
- 35:34NVIDIA and Apple. From a layperson's perspective,
- 35:38this seems entirely reasonable. A bank makes money.
- 35:41A tech company makes money. Why is it analytical malpractice
- 35:45to put them on the same chart? First, there's a fundamental
- 35:48data error. The figures used in the graphic
- 35:51for the banks were essentially pre tax income, not strictly
- 35:54operating income in the industrial sense.
- 35:56OK, that's a big error. But more importantly, the
- 35:58mathematical frameworks used to evaluate an industrial tech
- 36:01company and a complex financial institution are fundamentally
- 36:06incompatible. You were trying to measure
- 36:08distance with a thermometer. Let's breakdown why the
- 36:10frameworks are incompatible. Let's look at the raw materials.
- 36:13When we calculate Nvidia's operating income, we take the
- 36:16revenue from selling chips, subtract the cost of goods sold,
- 36:20like paying TSMC, and subtract R&D and administrative overhead,
- 36:24right? If NVIDIA happens to earn a
- 36:26billion dollars in interest because they have a pile of cash
- 36:29sitting in treasury bills, that interest income is peripheral.
- 36:33It's a nice bonus, but it is not the core operation of designing
- 36:37chips. Exactly.
- 36:38But for JP Morgan or Bank of America, interest is the core
- 36:41operation. It is the raw material of the
- 36:43factory. A bank is primary mechanism for
- 36:48generating revenue. Is the net interest margin or in
- 36:51M They take in deposits, pay the depositors a small amount of
- 36:55interest, and then lend that money out for mortgages or
- 36:58corporate loans at a higher interest rate.
- 37:00The spread between those two rates is the lifeblood of the
- 37:02business. So you cannot strip out interest
- 37:05and taxes to calculate an EBIT earnings before interest and
- 37:08taxes margin for a bank. Because if you strip out
- 37:10interest, you've stripped out the entire business model.
- 37:12It would be identical to evaluating a bakery's
- 37:15profitability, but deciding you aren't going to count the cost
- 37:18of flower in your calculations because it's a commodity
- 37:20expense. The math becomes meaningless.
- 37:23So if we can't use operating income or free cash flow to
- 37:26measure JP Morgan, what are the actual metrics analysts use to
- 37:30evaluate the health and profitability of a bank?
- 37:33The scorecard for banks is completely different.
- 37:35Analysts focus heavily on return on tangible common equity, or
- 37:40ROTCE. ROTCE.
- 37:41Yes. This measures how efficiently
- 37:44the bank is generating profit from the actual tangible capital
- 37:47invested by shareholders, stripping out intangible assets
- 37:50like goodwill from past acquisitions.
- 37:53And the sources point out that when you use the correct
- 37:55metrics, JP Morgan is vastly superior to its peers, boasting
- 37:59A roughly 29% reported ROTCE compared to Bank of America's
- 38:0317%. Right.
- 38:04But again, you cannot compare that 29% ROTCE to Nvidia's 65%
- 38:10operating margin. You also have to evaluate a
- 38:12bank's balance sheet risk, which is something you rarely have to
- 38:14do with a software company. Analysts look at CET One or
- 38:18common equity tier one capital. What's that?
- 38:20This is a regulatory metric that compares a bank score equity
- 38:24capital against its total risk weighted assets.
- 38:28It essentially measures a bank's ability to withstand severe
- 38:31financial stress and loan defaults without going bankrupt.
- 38:35You don't have to stress test Microsoft's balance sheet to see
- 38:38if they survive or run on the bank because they don't hold
- 38:41customer deposits. The risk profiles are entirely
- 38:45alien to each other. Totally alien.
- 38:46And this bad comparison problem isn't just limited to banks.
- 38:50The viral graphic also attempted to chart Berkshire Hathaway.
- 38:54Yes, the graphic claimed Berkshire had 117 billion in
- 38:58operating income, and categorizing Berkshire on a
- 39:00simple bar chart is perhaps the most egregious error of all.
- 39:04Because Berkshire Hathaway is an impossible beast to categorize.
- 39:07It isn't a single operating company.
- 39:10It is a massive conglomerate containing wholly owned
- 39:13industrial businesses like the BNSF railroad, utility companies
- 39:17and GEICO Insurance. And that insurance wing
- 39:19generates massive amounts of float cash collected from
- 39:22premiums that hasn't been paid out in claims yet, right?
- 39:25Warren Buffett takes that float and invests it into a gigantic
- 39:29portfolio of publicly traded equities, notably including a
- 39:33massive stake in Apple. Which means Berkshire suffers
- 39:36from the exact same mark to market accounting distortion
- 39:39that we saw with Alphabets venture capital investments, but
- 39:42on a much larger, more volatile public scale.
- 39:45Berkshire Hathaway explicitly warns its own investors in the
- 39:49very first pages of its annual report to ignore their bottom
- 39:52line net income figure. Literally tells them to ignore
- 39:55it, Yes. The wild swings in the stock
- 39:57market cause enormous artificial earnings volatility on their
- 40:01income statement. If the stock market drops 10%
- 40:04and a quarter, Berkshire might report a massive multi billion
- 40:08dollar Jap net loss even if the railroad and insurance companies
- 40:11operated Florida State and generated record cash.
- 40:14So if you see a chart attempting to compare Nvidia's operating
- 40:18income to Berkshire's income or JP Morgan's income, you should
- 40:21immediately recognize that the author fundamentally
- 40:24misunderstands accounting methodology.
- 40:26It is analytical malpractice. You are comparing apples,
- 40:29oranges and banks. OK.
- 40:30We have systematically torn down the bad metrics.
- 40:34We stripped away Alphabets Paper Venture Games, We adjusted for
- 40:37Amazon and Meta's massive growth CapEx.
- 40:40We warned about Micron's cyclical heat wave and we threw
- 40:43out the bad bank comparisons. We covered a lot of ground.
- 40:46Now it's time to build something truly useful.
- 40:50If you listening right now want to sit down, open an SEC filing
- 40:54and truly evaluate the economic power of a business, what is the
- 40:58ultimate formula? The Holy Grail metric
- 41:01popularized by Warren Buffett and utilized by serious
- 41:03fundamental analysts is called normalized owner earnings.
- 41:07Normalized owner earnings. Let's unpack this equation step
- 41:10by step, explaining the mechanics of why we add or
- 41:12subtract each component. Where do we start?
- 41:14We begin with normalized after tax operating earnings.
- 41:17Use operating earnings to isolate the core business, but
- 41:20the keyword is normalized. Normalized meaning what?
- 41:23We have to manually strip out all the one time weird
- 41:27accounting anomalies. We remove the legal settlements,
- 41:30the restructuring charges and crucially those phantom mark to
- 41:34market gains from private venture investments.
- 41:37We want the clean baseline profit of the company actually
- 41:41selling its products. OK.
- 41:43So we have a clean earnings number.
- 41:44What's the next step? Next, we add back depreciation
- 41:48and amortization, commonly known as DNA.
- 41:51Why are we adding an expense back into our profit?
- 41:54DNA is an accounting fiction. It is a non cash charge.
- 41:57When Amazon buys a $10 million server rack, the accountants
- 42:01don't deduct 10 million from the income statement on day one.
- 42:04They spread that cost out depreciating the asset by say 2
- 42:07million a year over five years. But that 2 million expense
- 42:10hitting the income statement this year doesn't represent
- 42:13actual cash leaving the company's bank account today.
- 42:16The cash left five years ago, so it's fine.
- 42:19The real cash generated today, we must add that non cash DNA
- 42:22expense back into our earnings. That makes perfect sense.
- 42:25We start with clean earnings. We add back the phantom DNA
- 42:28expenses. Now we have a massive pool of
- 42:31theoretical cash, but we have to keep the business running.
- 42:34Which leads to the third step, we subtract maintenance CapEx.
- 42:39And as we discussed heavily during the AI CapEx segment, we
- 42:43are only subtracting the physical spending required to
- 42:46keep the lights on the servers running and the existing
- 42:48warehouses patched exactly. We do not subtract the billions
- 42:52being spent on brand new discretionary growth
- 42:55initiatives. Correct.
- 42:56We want to ascertain what the current existing business
- 42:59generates in cash, completely independent of how management
- 43:02decides to aggressively expand in the future.
- 43:05OK. So we've subtracted maintenance
- 43:07CapEx. Is there one final piece?
- 43:09The final subtraction is required working capital.
- 43:12Working capital. Yes, this is the day-to-day cash
- 43:15tied up in the operational friction of the business.
- 43:18If a company has to hold massive amounts of physical inventory in
- 43:22warehouses, or if it takes 90 days to collect accounts
- 43:25receivable from customers that ties up cash, We must subtract
- 43:29the incremental working capital needed to run the day-to-day
- 43:32operations. So the master formula is
- 43:35normalized after tax operating earnings plus DNA minus
- 43:39maintenance CapEx minus required working capital that yields our
- 43:42normalized owner earnings. It is the most accurate
- 43:45reflection of the distributable cash generated by the core
- 43:48business. But you know, calculating that
- 43:50number is only the first part of the analysis.
- 43:52There's more. The true test of a companies
- 43:55greatness requires asking one final, incredibly powerful
- 44:00question regarding return on invested capital.
- 44:03This raises an important question.
- 44:04What is it? How much incremental capital
- 44:07must be reinvested back into the business to grow those owner
- 44:10earnings by one single dollar? Oh, this is the treadmill
- 44:13concept we touched on earlier. Yes, exactly.
- 44:15Let's build a hypothetical case study to explain the mechanics
- 44:18of this, because it's the secret to compounding wealth.
- 44:21Let's imagine two companies, Business A and Business B.
- 44:25Both businesses are currently growing their owner earnings at
- 44:27a fantastic rate of 15% a year. On the simple stock Screener,
- 44:32looking only at growth rates, both companies look like massive
- 44:35winners. But let's look at capital
- 44:37intensity. Business A is a heavy industrial
- 44:41manufacturer. In order to achieve that 15%
- 44:44growth, they have to constantly build new factories, buy new
- 44:48heavy machinery, and stockpile vast amounts of steel inventory
- 44:51right. To get that growth, Business A
- 44:54is forced to reinvest 100% of its current earnings right back
- 44:59into the company. Which means there is absolutely
- 45:010 distributable cash leftover for the owners.
- 45:04The business is growing, but it is consuming all of its own cash
- 45:07just to sustain that growth trajectory.
- 45:09It's just a capital hungry monster.
- 45:11Exactly. Now look at Business B.
- 45:13Let's say Business B is a dominant software platform.
- 45:16To achieve that exact same 15% growth rate, they just need to
- 45:19deploy some new code and perhaps lease a slight amount of extra
- 45:22server space. Yeah, Business B only requires
- 45:26reinvesting 20% of its earnings back into the company to achieve
- 45:29the growth. That leaves 80% of the owner
- 45:32earnings available to be distributed to you, the
- 45:34shareholder, either through pure cash dividends or stock buybacks
- 45:38that increase your ownership percentage.
- 45:39So even though both businesses are growing at the exact same
- 45:4315% clip, Business B is an infinitely superior economic
- 45:47asset because it actually yields cash you can take home.
- 45:50Growth is only valuable if it doesn't cost you all your cash
- 45:53to achieve it. And this framework allows us to
- 45:56synthesize everything we have uncovered in our deep dive
- 45:58today. It really does.
- 46:00When we analyze our big tech Titans through this
- 46:02sophisticated lens of normalized owner earnings, capital
- 46:06intensity and structural durability, who actually wins?
- 46:10Based on the deep reconstruction of the sources, it isn't a
- 46:12single winner. It is a highly nuanced race with
- 46:15three very different champions depending on the time horizon
- 46:18you're optimizing for. If you're only metric is present
- 46:22day economic power, who has the most explosive current margins
- 46:25and raw operational leverage right now?
- 46:29The second NVIDIA is the undisputed champion.
- 46:32A 65% hardware margin is a historical anomaly that demands
- 46:36respect. However, if the ultimate test is
- 46:39capital efficiency, who is the best representation of business
- 46:42B from our example, capable of extracting colossal amounts of
- 46:45cash without requiring massive physical reinvestment?
- 46:49Apple claims the Crown by. Far.
- 46:51By offloading the capital intensive manufacturing to
- 46:54Foxconn and TSMC, Apple extracts over 136 billion in pure cash
- 47:00while spending only a fraction of that on physical
- 47:03infrastructure. They are the ultimate asset
- 47:05light compounder. But if we change the time
- 47:07horizon, if we are projecting 10 years into the future and we are
- 47:11asking whose cash flows we feel most mathematically confident
- 47:15forecasting A decade from now, taking into account inevitable
- 47:18recessions, consumer hardware fatigue and vicious
- 47:21semiconductor cycles. Under the strict test of long
- 47:23term durability, Microsoft likely wins.
- 47:26The enterprise switching costs are simply too high for their
- 47:28corporate customers to leave. They possess an unparalleled
- 47:32economic mode that insulates their software margins from
- 47:34macroeconomic shocks. It is genuinely fascinating to
- 47:38see how the definition of the best company completely morphs
- 47:41depending on how you define and measure profit.
- 47:44We started this journey looking at a simple deceptive bar chart
- 47:47of net income and we ended up disassembling the entire
- 47:51architectural concept of corporate earnings.
- 47:53We learned why Alphabets seemingly unstoppable $98
- 47:57billion profit surge in a single quarter might just be a fragile
- 48:01paper game heavily dependent on illiquid venture capital
- 48:04valuations holding steady. We learned why Apple's distinct
- 48:08lack of physical factories isn't a weakness, but rather their
- 48:11absolute superpower, allowing them to utilize negative working
- 48:15capital and harvest cash with ruthless efficiency.
- 48:18And perhaps most importantly, we learned why a massive,
- 48:21frightening drop in free cash flow might not mean Amazon and
- 48:24Meta are deteriorating businesses.
- 48:26It might just mean they are brilliantly and aggressively
- 48:28buying the foundational infrastructure for the next
- 48:31decade of AI computing. It fundamentally changes how you
- 48:34consume financial news. You can't just read an earnings
- 48:37press release, look at the net income headline, and think you
- 48:39understand the economic reality of a company.
- 48:42You can't. You have to do the forensic
- 48:44detective work. You have to understand the
- 48:46accounting rules, separate the growth CapEx from the
- 48:49maintenance CapEx and find the real cash.
- 48:52It requires rigor. And as we conclude this deep
- 48:55dive into the true profitability of Big Tech, I want to leave you
- 48:59with a final, slightly provocative thought to Mull
- 49:01over, building on our analysis of the AI infrastructure race.
- 49:06I love good thought experiment. Let's hear it.
- 49:07We've established that Alphabet, Meta and Amazon are currently
- 49:12sinking hundreds of billions of dollars into data centers and
- 49:15GPU's. It's a tidal wave of growth
- 49:17CapEx that is severely suppressing their visible free
- 49:20cash flow today, right? But consider the timeline of
- 49:23infrastructure. What happens in three or four
- 49:26years if that infrastructure is fully built, the initial AI arms
- 49:29race stabilizes and that massive growth CapEx requirement
- 49:33suddenly drops back down towards 0?
- 49:35If their core revenues from search, cloud and advertising
- 49:38hold steady, we might be on the precipice of the largest, most
- 49:41violent explosion of raw free cash flow in the history of
- 49:44modern capitalism, as all that currently suppressed cash is
- 49:48suddenly released back to the owners.
- 49:49That is an incredible bull case. The cash spigot just turns wide
- 49:53open. But, and this is the flip side,
- 49:56the inverse scenario carries devastating consequences.
- 50:00What's that? If they build all of this
- 50:02physical infrastructure and the end user enterprise and consumer
- 50:06demand for AI doesn't materialize at the massive scale
- 50:10they are projecting in their models, well then they didn't
- 50:12just temporarily suppress their cash flow, right?
- 50:15They just spent hundreds of billions of dollars building the
- 50:17world's most sophisticated, most expensive, empty warehouses.
- 50:21That is a terrifying, A thrilling thought to end on.
- 50:24It all comes down to whether those billions are a
- 50:26masterstroke of visionary reinvestment or the largest
- 50:29cyclical trap ever laid. Thank you so much for joining us
- 50:32on this deep dive. Stay curious, keep digging into
- 50:36the footnotes of those cash flow statements, and always look
- 50:39beyond the headline numbers.