Latest / Deep Values: Value Investing Deep Dives / Microsoft's Q2 Earnings Beat! 190 Billion Dollar AI Gamble Won?!
Transcript
- 0:00To access more advanced researches, applications, and
- 0:02our latest Deep Values agents. Welcome to explore and join Deep
- 0:06Values APP Track total market temperature.
- 0:09Welcome to search and download iOS app Buffett Indicator from
- 0:13Apple App Store or Bubble Watch from Google Play.
- 0:16Disclaimer All analysis is based on publicly available sources
- 0:20and may be subject to revisions or differing interpretations.
- 0:23All contents from this channel is for research and
- 0:26entertainment purposes only and does not constitute any
- 0:29investment or financial advice or service.
- 0:32Conduct your own due diligence before making investment
- 0:34financial decisions right. So welcome in everyone.
- 0:37We are really glad to have you with us today.
- 0:39Today, today's deep dive is it's a wild.
- 0:41One it really. Is because Microsoft just had
- 0:44the kind of quarter that honestly should make Wall Street
- 0:46break out the champagne. I mean, it completely smashed
- 0:48expectations. Yeah, absolutely crushed.
- 0:49Them right, they demonstrated this hyper growth in their cloud
- 0:52division, but instead of popping bottles, institutional investors
- 0:56completely panicked and the stock actually dropped.
- 1:00Dropped pretty significantly too.
- 1:01Yeah. So today we are figuring out why
- 1:04we're diving into a comprehensive 2026 financial and
- 1:07valuation research report. Answer just one massive question
- 1:11for you. Is Microsoft's aggressive
- 1:13transition into a capital heavy artificial intelligence behemoth
- 1:17A visionary master stroke for the next decade, or are we
- 1:21looking at a dangerous civilizational scale spending
- 1:23bubble that's just going to end in tears?
- 1:25A very expensive bubble. Exactly.
- 1:27OK, let's unpack this. So to really understand the
- 1:30panic we we have to look at the sheer scale of the top line
- 1:33numbers they just reported because they are staggering.
- 1:36Oh yeah, the revenue alone is insane.
- 1:37Right, we are talking about $82.89 billion in revenue for a
- 1:41single quarter. Which is just, I mean, almost
- 1:44$83 billion in 90 days, that is approaching a billion dollars a
- 1:48day in revenue. Yeah, A.
- 1:49Billion a day and their earnings per share came in at $4.27,
- 1:53which you know, comfortably beat Wall Street's expectation of
- 1:55$4.07. And the real engine of all this,
- 1:58the Microsoft Cloud segment, It hit 46.7 billion, with their
- 2:01flagship Azure platform growing at an astounding 39 to 40%.
- 2:05OK, so if the money printer is running that hot and they beat
- 2:09all these expectations, why on earth did the stock drop?
- 2:13Because Wall Street stopped looking at the money coming in
- 2:15and they just hyper focused on the money going out.
- 2:18Yeah, yeah. Despite a 20% overall increase
- 2:21in operating income, the gross margin for the cloud segment
- 2:24actually compressed. It dropped from a historical
- 2:27high of 72% down to 68%. Wait, let me stop you there and
- 2:31just kind of put on my beginner's hat for a second.
- 2:33For a software company, gross margin is essentially the Holy
- 2:36Grail. Oh, completely.
- 2:38It's everything. It's.
- 2:39How much profit they make on their core services before
- 2:43paying for all the corporate overhead.
- 2:45So a 4% drop, you know, it sounds pretty small in everyday
- 2:48life, but in financial terms, what does that actually mean?
- 2:51Well, in institutional finance, they track this in basis points
- 2:54and a 400 basis point drop in your core growth engine that is
- 2:57a fundamental shift in how the business operates.
- 2:59Oh wow. OK.
- 3:00And the culprit behind this margin compression is CapEx
- 3:04capital expenditures. To build out the infrastructure
- 3:08required for this generative AI era, management guided A
- 3:11shocking $190 billion in CapEx for the calendar year 2026
- 3:15alone. Wait, I'm sorry, $190 billion?
- 3:20Yes, in 12 months. In a single year.
- 3:23I mean, I like to think about this in real world terms.
- 3:25Imagine you own a you know, a a highly profitable delivery
- 3:28business using bicycles, and suddenly you decide to
- 3:32transition your entire fleet to state-of-the-art $1,000,000
- 3:36autonomous delivery trucks. Right, huge upfront investment.
- 3:39Yeah, and you're buying 10s of thousands of these trucks before
- 3:41you even know if your customers actually want to pay premium
- 3:43prices for what you're. Delivering exactly the upfront
- 3:46cost is astronomical, and the payoff is entirely dependent on
- 3:50future customer behavior that just hasn't fully materialized
- 3:53yet. The bulk of that 190 billion is
- 3:55going towards securing advanced GPU's from companies like NVIDIA
- 3:58and AMD and physically constructing the hyperscale data
- 4:02centers to actually house them. But hold on, let me jump in
- 4:04here. 190 billion sounds like they are, you know, boldly
- 4:08building the future. But reading through the source
- 4:10material, management noted that 25 billion of that 2026 spend is
- 4:16purely the result of supply chain component inflation.
- 4:18Yeah, the ships are getting way more expensive.
- 4:21So are they actually expanding their capabilities or are they
- 4:23just paying a massive premium, getting gouged on prices just to
- 4:27stay in the game? What's fascinating here is that
- 4:29Microsoft's executive team is completely aware of this
- 4:33dynamic. They are intentionally absorbing
- 4:36this margin dilution. Really.
- 4:38Intentionally. Yeah.
- 4:39It's not an operational accident.
- 4:41It's not a failure to control costs.
- 4:43It is a deliberate structural shift.
- 4:46They are moving away from being a high margin asset light
- 4:51software company. You know, where you write code
- 4:53once and distribute it globally for practically nothing and
- 4:56turning into a highly capital intensive physical
- 4:58infrastructure company. So they're basically building
- 5:01the digital factories of the 21st century, and you don't
- 5:04spend nearly $200 billion on factories in a single year
- 5:09unless you believe the economic Moat you are building around
- 5:13your business is going to be completely impenetrable.
- 5:15And the current state of that Moat is undeniably massive.
- 5:18Microsoft has 450 million active commercial M365 seats.
- 5:24Which is pretty much everyone. It represents virtually the
- 5:26entire corporate world using Word, Excel, Teams and Outlook
- 5:30every single day. Meaning all the world's
- 5:32corporate data already lives on their servers.
- 5:34Right. And the critical data point from
- 5:36the report is how they are leveraging that data.
- 5:39Their premium AI add on Microsoft 365 Copilot just
- 5:43passed 20 million paid seats. Wow.
- 5:45That translates to an adoption rate of about 4.4% of their
- 5:49total commercial base. So embedding state-of-the-art
- 5:52foundational models directly into the workflow where people
- 5:54already do their jobs, that creates some serious ecosystem
- 5:57lock in. Huge lock in a.
- 5:58Fortune 500 company can't just, you know, unplug its entire
- 6:01global workforce from Microsoft just to try out a different AI
- 6:04tool. The switching costs would be an
- 6:06absolute nightmare. Ecosystem lock in is the
- 6:08ultimate bowl case here, but the source material provides a very
- 6:13sobering short term reality check.
- 6:16What kind of reality check? We are currently seeing a
- 6:18staggering statistic across the industry. 95% of enterprise AI
- 6:23projects fail to generate measurable returns in their
- 6:26early pilot stages. Wait, really?
- 6:2895% failure on ROI. Yes, 95% why?
- 6:32Is it that high? Are the models just not good
- 6:33enough yet or? It's actually less about the
- 6:36models and more about the corporate reality.
- 6:39Integration is incredibly complex.
- 6:40Corporate data is often siloed. It's messy.
- 6:43Companies really struggle to define clear, measurable
- 6:47business outcomes for these AI tools beyond just, you know,
- 6:50having them write polite emails. Right, which doesn't really move
- 6:52the needle on profits. Exactly.
- 6:54So you have this massive gap. Microsoft is spending capital
- 6:57today at hyper speed, but the revenue realization, the actual
- 7:00return on investment from their corporate clients is severely
- 7:03delayed. OK.
- 7:04So we have a timing mismatch in the short term, but the report
- 7:07also highlights some frankly terrifying long term structural
- 7:10threats to this whole operation. Oh, absolutely.
- 7:12The two massive next seem to be the depreciation treadmill and
- 7:16the power grid. Let's unpack the treadmill
- 7:19first. OK.
- 7:20So the depreciation treadmill is arguably the most severe
- 7:23economic risk highlighted in the analysis.
- 7:26Roughly 2/3 of this $190 billion hyperscale CapEx is going
- 7:31towards short lived assets. Things like the GPU's.
- 7:34Primarily AIGPUS and custom silicon for accounting purposes.
- 7:39Microsoft tells the IRS and its investors that it will
- 7:41depreciate these assets over 5 to 6 years, but their actual
- 7:45technological lifespan is heavily compressed.
- 7:47It's more like 2 to three years. Right, because the hardware gets
- 7:50faster and more efficient so quickly.
- 7:52It's like buying top of the line smartphones for 100,000
- 7:54employees for tax purposes, you say?
- 7:56Oh, they'll last five years, but because the apps and operating
- 7:59systems demand twice as much processing power every couple of
- 8:02years, those phones are essentially bricks.
- 8:05In 24 months you are forced to buy the next generation just to
- 8:08keep the software running. That captures the dynamic
- 8:11perfectly. You are buying highly expensive
- 8:13assets that become obsolete almost immediately.
- 8:16It creates a permanent highly capital intensive treadmill.
- 8:19You have to keep buying new infrastructure just to maintain
- 8:22market parity with Google or AWS.
- 8:24It's. Like a technology subscription
- 8:26you can never cancel. Yes, if you stop paying for the
- 8:28new chips, your cloud falls behind and your mode dries up.
- 8:31Which I guess leads us to the physical limit of all this
- 8:34computing power, the electricity grid.
- 8:37The grid is a massive, massive problem.
- 8:39The research notes that a single one GW data center carries an
- 8:43$8.5 billion annualized total cost of ownership.
- 8:46And beyond the monetary cost, it requires immense local grid
- 8:50support. You cannot just, you know, drop
- 8:52A1 GW data center anywhere. It requires specialized power
- 8:55purchase agreements, upgraded transmission lines, and massive,
- 8:58massive amounts of cooling infrastructure.
- 9:01So I want you, the listener, to really think about with this for
- 9:04a second, Yeah. Are we going to hit a wall with
- 9:05electricity availability long before we hit a limit on what AI
- 9:08can actually do? It is a highly probable
- 9:11scenario. Energy, not silicon supply, is
- 9:14rapidly becoming the ultimate limiting factor to cloud
- 9:17expansion. A company can print money to buy
- 9:20more GPU's, but they cannot print A1 GW power plant.
- 9:24Yeah, and with all these intense capital constraints, the massive
- 9:27margin dilution and the sheer physical limits of electricity,
- 9:30a traditional value investor must be tearing their hair out
- 9:33looking at the stock. Oh, they definitely are.
- 9:35Let's look at how Microsoft actually stacks up against deep
- 9:37value investing standards. Can a strict Benjamin Graham
- 9:41style value investor even compute what Microsoft is doing
- 9:44right now? Well, if you apply rigid old
- 9:47school value metrics, Microsoft utterly fails the test.
- 9:50Completely fails. Right.
- 9:51Benjamin Graham's philosophy demands strict valuation
- 9:53discipline, focus heavily on capital preservation and
- 9:55tangible physical assets. The benchmark deep value
- 9:58standard requires a price to earnings or PE ratio of under
- 10:0215. Microsoft's PE currently sits at
- 10:0523.41. And just to clarify for everyone
- 10:07listening, APE of 23 essentially means investors are willing to
- 10:11pay $23 today for every $1.00 of earnings the company makes over
- 10:16a year. So Grant would say you're
- 10:18overpaying for that dollar. Correct, you are drastically
- 10:21overpaying in his eyes and it looks even worse on the balance
- 10:24sheet. How so?
- 10:25Well the deep value standard for a priced book or PB ratio
- 10:29requires a multiple of under 1.5.
- 10:32Microsoft is trading at APB of 8.15.
- 10:35Wow, let's break that one down too.
- 10:36Price to book is basically comparing the market value of
- 10:40the stock to the actual physical assets the company owns if they
- 10:43sold out everything tomorrow. Right.
- 10:44Yes, exactly. And trading at over 8 times its
- 10:47book value means it fails Graham's test spectacularly.
- 10:51But, and this is crucial, we have to recognize the limitation
- 10:54of those old rules. Modern software companies
- 10:57generate immense intangible assets, millions of lines of
- 11:00proprietary code, oceans of user data, global network effect.
- 11:03Stuff you can't just put in a warehouse.
- 11:04Exactly. Those simply do not show up on a
- 11:06traditional balance sheet like a factory or a piece of machinery
- 11:09would. So the industrial error rules
- 11:10don't quite fit the digital era. But when the report applies
- 11:14modern quality and cash generation metrics, the story
- 11:17completely flips, doesn't it? It absolutely does.
- 11:19Because the numbers here are just wild.
- 11:21They have a 24% free cash flow margin, and they generated $74.5
- 11:26billion in free cash flow over the trailing 12 months.
- 11:29It's a money machine. Furthermore, their return on
- 11:31invested capital or ROIC is 31.5%.
- 11:36Wait, let's break that acronym down for a second.
- 11:37Return on invested capital. What does a 31.5% ROIC actually
- 11:42look like in practice for Microsoft?
- 11:44So it measures how efficiently a company uses the money invested
- 11:47in it to generate profit. A 31.5% ROIC means that for
- 11:51every dollar Microsoft puts into its business, whether that's
- 11:54paying developers or building those expensive data centers,
- 11:56they are generating nearly $0.32 in economic profit above their
- 12:00cost of capital. That is massive.
- 12:01That is a phenomenal rate of return, especially while
- 12:04simultaneously funding $190 billion CapEx cycle and they
- 12:08also easily pass the Peter Lynch PEG ratio test.
- 12:11And the PEG ratio that compares their PE valuation to their
- 12:15actual growth rate. Yes, a PEG ratio under one point
- 12:17O is generally considered excellent, indicating that a
- 12:20stock might be undervalued relative to its earnings growth.
- 12:23Microsoft is currently sitting at .76.
- 12:25OK, so that means their underlying earnings growth is
- 12:28vastly outpacing the premium valuation multiple the market is
- 12:32assigned them precisely. OK, so they generate a mountain
- 12:34of cash, but they are spending a mountain of cash.
- 12:37Let's get into the hard math to figure out what the company's
- 12:39actually worth. The report runs a discounted
- 12:43cash flow, or DCF, analysis to find Microsoft's true intrinsic
- 12:48value. Right, the DCF models.
- 12:49Before we hit the numbers, give us the simplest possible
- 12:51explanation of how ADCF model actually works.
- 12:54OK. ADCF model is essentially
- 12:56financial time travel. I like that it tries to figure
- 12:59out what all the cash Microsoft will make in the future is worth
- 13:01into those dollars because money today is worth more than money
- 13:04tomorrow due to inflation and opportunity cost.
- 13:07Makes sense? So you project out the cash they
- 13:10will generate over the next 10 years, and then you discount it
- 13:12back to today's value based on a specific interest rate.
- 13:15The report ran three different projections.
- 13:18Walk us through the scenarios. Let's start with the worst case.
- 13:20The first is the no growth case. This assumes total stagnation.
- 13:25It models a world where the AI monetization thesis completely
- 13:28fails, Corporate clients cancel their Copilot subscriptions, and
- 13:31Azure just stops growing entirely.
- 13:33It's a total stall. Yeah, under that highly
- 13:36pessimistic scenario, the intrinsic value of Microsoft is
- 13:39calculated at $149.99 per share. Which is drastically below where
- 13:44it's currently trading, but I guess it's helpful to know where
- 13:46the absolute floor is. What's the second scenario?
- 13:48The second is the conservative case.
- 13:50This acts as an aggressive stress test.
- 13:52It simulates a severe macroeconomic depression.
- 13:55It assumes Microsoft's cash flow doesn't just stagnate, but
- 13:57actually shrinks by 8.7% every year for a decade.
- 14:01Ouch. Yeah, that apocalyptic scenario
- 14:04values the stock at just $81.05 per share.
- 14:08OK, well, let's hope that doesn't happen.
- 14:10And the third model, the one that expects them to actually
- 14:12succeed. The third is the nominal case.
- 14:14This assumes Microsoft successfully navigates this
- 14:17highly expensive CapEx cycle and maintains the exact historical
- 14:21average cash flow growth rate they have demonstrated over the
- 14:24past 10 years. Which is what exactly?
- 14:26About 13.1% annual growth. That scenario values Microsoft
- 14:31at $402.34 per share. Here's.
- 14:34Where it gets really interesting, the current stock
- 14:36price as of this research report is $395.50.
- 14:41That is almost exactly the nominal case value of $402.34.
- 14:46So I had to push back here and ask, doesn't this mean there is
- 14:48absolutely 0 margin of safety for investors right now it?
- 14:51Absolutely means that if the. Copilot AI revenue doesn't
- 14:54perfectly offset this $190 billion data center build out.
- 14:58Won't those massive depreciation schedules we talked about just
- 15:01permanently crush future earnings?
- 15:03If we connect us to the bigger picture, you are seeing the
- 15:05exact tension that caused the institutional panic in the 1st
- 15:09place. At 395 a share, investors are
- 15:11paying for flawless execution. There is no structural margin of
- 15:15safety. It reveals A profound truth
- 15:18about Microsoft's management strategy today.
- 15:20They are not trying to maximize short term free cash flow yield.
- 15:24They don't care about looking pretty for value investors this
- 15:27quarter. Not at all.
- 15:28Their objective is much larger. They are attempting to capture
- 15:31the entire global AI workload layer before Amazon Web Services
- 15:35or Google can scale their infrastructure to meet the
- 15:38demand. Wow.
- 15:39It is a high wire act being performed with a $3 trillion
- 15:43company. And executing a high wire act
- 15:45with 0 margin of safety requires immense trust from the people
- 15:49backing you. Satya Nadella is essentially
- 15:51asking Wall Street for a decade of blind trust to pull off this
- 15:55190 billion gamble. How's the leadership team
- 15:58handling that pressure? Well, CEO Satya Nadella and CFO
- 16:01Amy Hood, they have actively tried to maintain that trust
- 16:04through aggressive transparency. They haven't tried to hide the
- 16:06margin dilution from wall. Street, they got out ahead of
- 16:09it. Exactly.
- 16:09They proactively forecasted it months in advance.
- 16:12They are framing it as a deliberate 10 year lock in
- 16:16strategy. Furthermore, their compensation
- 16:18is heavily tied to performance metrics and they are still
- 16:21maintaining a massive $60 billion share buyback program to
- 16:25return capital to shareholders. Though it's worth noting the
- 16:28pace of those buybacks has slightly moderated recently,
- 16:31simply because they need the liquid cash to feed that $190
- 16:34billion CapEx monster. Naturally.
- 16:37And structurally, they do have the backing of Wall Street's
- 16:40heavyweights. 73% of the stock is held by institutional giants
- 16:44like Vanguard, Black Rock and State Street.
- 16:46The biggest players in the game. Yeah, who view Microsoft as a
- 16:49core pillar of the global economy.
- 16:51But, and this is a big but, looking at the recent December
- 16:542025 annual shareholders meeting, that trust is starting
- 16:58to crack. It really.
- 16:59Is and surprisingly, it is not just because of the financial
- 17:02metrics. No2 shareholder proposals gained
- 17:05massive, unexpected traction regarding geopolitics and
- 17:09ethics. Proposal 8 demanded A
- 17:11comprehensive report on censorship and human rights
- 17:13risks associated with Microsoft's data centers in
- 17:16Saudi Arabia. That proposal won over 27% of
- 17:20the vote. Over 27%, that's huge for a
- 17:22shareholder vote. And what was the second one?
- 17:25Proposal 9 sought a formal assessment of human rights due
- 17:28diligence regarding the potential misuse of Microsoft's
- 17:31AI tools by military or state actors that secured over 26% of
- 17:36the shareholder vote. I want you to pause and consider
- 17:39that over 1/4 of the shareholders voting against the
- 17:42board on human rights issues for a massive tech stock that is
- 17:45highly unusual. Extremely unusual.
- 17:47That shows that the risks here aren't just about microchips and
- 17:50accounting depreciation and profit margins.
- 17:52We are talking about permanent geopolitical and reputational
- 17:56blowback. Definitely.
- 17:57When a company builds civilizational scale
- 17:59infrastructure across the globe, it ceases to be just a software
- 18:02vendor, becomes a geopolitical actor.
- 18:04The report explicitly warns that in emerging global markets,
- 18:08regulatory blowback, shareholder activism, or forced market exits
- 18:12could permanently impair their revenue growth trajectories.
- 18:15Yeah, that makes sense. If a foreign government decides
- 18:17to seize a localized data center or heavily regulate what a
- 18:20foundational AI model is allowed to output, that directly and
- 18:24permanently impacts the bottom line.
- 18:26It's the ultimate paradox of scale.
- 18:28The bigger the Moat gets, the more physical infrastructure it
- 18:31requires and the more vulnerable it becomes to the physical and
- 18:35political realities of the world.
- 18:36Precisely. Let's pull all this together.
- 18:38We have covered a tremendous amount of ground in this deep
- 18:40dive. We really have.
- 18:41We are basically looking at a $3 trillion compounding machine
- 18:46that is currently acting with the capital Hungary.
- 18:49Hyper aggressive mentality of an early stage startup.
- 18:52Yeah, they're betting $190 billion in 2026 alone.
- 18:56They are risking immense short term margin compression, all to
- 19:00secure a monopoly on the AI future.
- 19:02It's a huge swing. They pass modern cash flow
- 19:04metrics with flying colors, but they offer zero margin of safety
- 19:08at current prices. If they are wrong about
- 19:10enterprise AI adoption or if the costs continue to spiral, the
- 19:14financial consequences will be historic.
- 19:17It leaves us with a fascinating final thought to consider
- 19:20returning to that physical power grid constraint we talked about.
- 19:23Oh yeah. If energy availability rather
- 19:26than silicon, is rapidly becoming the ultimate limiting
- 19:28factor to cloud expansion, will software giants like Microsoft
- 19:32eventually be forced to become massive energy and utility
- 19:34companies themselves? That is wild to think about,
- 19:37right? Will they have to build their
- 19:38own nuclear reaction sectors or solar farms just to keep their
- 19:41AI ambitions alive? A software company pivoting to
- 19:44become a global power utility. It sounds like science fiction,
- 19:48but given the massive numbers we just walked through, it might
- 19:51just be the inevitable next step.
- 19:53So what does this all mean? It means the future of
- 19:56technology isn't just about code anymore.
- 19:58It's about copper wire, concrete, and watts.
- 20:01Keep digging, keep questioning the numbers, and we'll see you
- 20:04on the next deep dive.