Latest / AI Investment Edge: Stocks, Markets & Money / EP 37 | Microsoft Stock Analysis: AI Bubble? Explain These Earnings.
Transcript
- 0:00SPEAKER_01: Imagine you're running a wildly successful lemonade stand.
- 0:03SPEAKER_00: Okay. I like lemonade.
- 0:05SPEAKER_01: Right. And your sales just shut up 19% across the board.
- 0:08SPEAKER_01: Plus your brand new premium iced tea segment that just skyrocketed by 48%.
- 0:14SPEAKER_00: Oh, wow.
- 0:14SPEAKER_01: Yeah. You've got a line around the block.
- 0:16SPEAKER_01: People are literally throwing money at you.
- 0:19SPEAKER_01: But then, you know, you close up shop, count the till,
- 0:22SPEAKER_01: and you actually have less free cash in your pocket than you did yesterday.
- 0:26SPEAKER_00: Which is incredibly frustrating.
- 0:27SPEAKER_01: Exactly.
- 0:28SPEAKER_01: And that is exactly the tension we're looking at today with a $3 trillion company.
- 0:33SPEAKER_01: Microsoft just posted those exact growth numbers, heavily driven by Azure,
- 0:37SPEAKER_01: and yet their free cash flow fell.
- 0:40SPEAKER_00: Yeah, it creates this fascinating paradox for the market.
- 0:43SPEAKER_00: I mean, you have undeniable paying demand,
- 0:46SPEAKER_00: which completely shatters that whole narrative
- 0:48SPEAKER_00: that the AI boom is just like empty hype or vaporware.
- 0:52SPEAKER_01: Right, to real money.
- 0:53SPEAKER_00: Exactly. Customers are writing massive checks for this compute power, but...
- 0:57SPEAKER_00: That demand is incredibly expensive to service.
- 1:00SPEAKER_01: Which means revenue growth alone doesn't rule out a valuation bubble.
- 1:04SPEAKER_01: And it certainly doesn't rule out wasteful overinvestment.
- 1:07SPEAKER_00: No, it doesn't.
- 1:08SPEAKER_01: So the overarching question for you listening right now is this.
- 1:12SPEAKER_01: If AI is such a goldmine, explain the massive cash bill.
- 1:16SPEAKER_00: That's the billion dollar question.
- 1:17SPEAKER_01: Or, well, multi-billion.
- 1:19SPEAKER_00: Right. Right.
- 1:19SPEAKER_01: Okay, let's unpack this.
- 1:21SPEAKER_01: Welcome to AI Investment Edge.
- 1:23SPEAKER_01: We are jumping straight into a deep dive on Microsoft's colossal infrastructure bet.
- 1:28SPEAKER_00: And to really grasp the scale of that bet,
- 1:32SPEAKER_00: we have to ground ourselves in the sheer breadth of Microsoft's business model.
- 1:35SPEAKER_01: It's massive.
- 1:36SPEAKER_00: Yeah, because yes, businesses and governments are paying a premium
- 1:40SPEAKER_00: for Azure computing and AI infrastructure,
- 1:43SPEAKER_00: but the revenue engine is much wider than that.
- 1:45SPEAKER_01: You've got, what, Microsoft 365, Copilot, Right.
- 1:49SPEAKER_00: Copilot productivity tools, developer services, security, Windows, advertising, and Xbox.
- 1:54SPEAKER_01: It's an absolute behemoth.
- 1:56SPEAKER_01: Subscriptions, metered computing, software licenses, all of it.
- 2:00SPEAKER_01: And for anyone tracking the financials closely,
- 2:02SPEAKER_01: you should note there was a major reporting change back in September.
- 2:05SPEAKER_00: Oh yeah, that consolidation.
- 2:07SPEAKER_01: Exactly. They consolidated the old three segments into just two.
- 2:11SPEAKER_01: So now it's agents and infrastructure plus devices and consumers.
- 2:15SPEAKER_00: Which means any current year-on-year results we look at are recast to fit that new structure.
- 2:21SPEAKER_01: Right. And also, we need to remember that Azure is much broader than just AI.
- 2:26SPEAKER_01: There wasn't any standalone total AI revenue
- 2:29SPEAKER_01: or an updated total paid co-pilots account disclosed on this October call.
- 2:33SPEAKER_00: Yeah, they keep that pretty close to the chest.
- 2:36SPEAKER_00: But let's look at the actual earnings quality from those October results
- 2:40SPEAKER_00: because the top line numbers are staggering,
- 2:42SPEAKER_00: but they really require a bit of translation.
- 2:45SPEAKER_01: Hit me with the numbers.
- 2:47SPEAKER_00: So Q3 revenue came in at $92.28 billion.
- 2:50SPEAKER_00: That's up 19%.
- 2:52SPEAKER_00: Operating income was $45.92 billion, which is up 21%.
- 2:56SPEAKER_01: Okay. Huge.
- 2:57SPEAKER_00: Right. But the accounting underneath those numbers tells a very specific story.
- 3:01SPEAKER_01: Yeah. Because if I look at the gap diluted earnings per share...
- 3:05SPEAKER_01: It sits at $4.92, which looks like a massive 32% jump.
- 3:10SPEAKER_01: But that's not the whole story, is it?
- 3:12SPEAKER_00: Far from it.
- 3:13SPEAKER_00: I mean, investment accounting is totally separate from actual cloud sales.
- 3:17SPEAKER_00: If you exclude their OpenAI investment gains and losses, that EPS is actually $4.89,
- 3:24SPEAKER_00: which is an 18% increase.
- 3:25SPEAKER_01: Wait, let me make sure I'm translating this correctly.
- 3:27SPEAKER_01: So that official gap growth number looks artificially fast simply
- 3:31SPEAKER_01: because they recorded a small open AI investment gain this time around?
- 3:36SPEAKER_00: Exactly.
- 3:37SPEAKER_00: And they're comparing it to a period last year where they took a large loss
- 3:40SPEAKER_00: on that same investment. Oh, wow.
- 3:41SPEAKER_01: So it's an accounting quirk.
- 3:42SPEAKER_00: Yeah, it swings the percentage wildly.
- 3:45SPEAKER_00: And as a crucial piece of context, even with that adjustment,
- 3:48SPEAKER_00: these results exceeded management's own July guidance.
- 3:52SPEAKER_00: And we are comparing this performance strictly to what management guided,
- 3:55SPEAKER_00: leaving Wall Street analyst estimates completely out of the picture.
- 3:58SPEAKER_01: Fair enough. So they are generating more money.
- 4:01SPEAKER_01: But if I'm an investor, I care deeply about how much it actually costs them
- 4:04SPEAKER_01: to deliver these services.
- 4:06SPEAKER_00: Absolutely.
- 4:07SPEAKER_01: Are their profit margins getting squeezed by all these shiny new AI servers?
- 4:11SPEAKER_01: Because when I look at the margins, I see...
- 4:15SPEAKER_01: Two conflicting signals.
- 4:16SPEAKER_00: Yeah, the gross versus operating.
- 4:18SPEAKER_01: Right. The group gross margin fell to 67.79%.
- 4:23SPEAKER_01: That's down from 69.05%.
- 4:25SPEAKER_01: Yet the operating margin actually rose to 49.76%.
- 4:31SPEAKER_01: How does the gross margin go down while the overall operating margin goes up?
- 4:35SPEAKER_00: It seems counterintuitive until you look at the physical reality
- 4:38SPEAKER_00: of what they're actually selling.
- 4:40SPEAKER_01: Okay, explain that.
- 4:42SPEAKER_00: Well, gross margin looks at the direct cost of delivering a product.
- 4:46SPEAKER_00: Microsoft has this incredibly fast-growing infrastructure business, right?
- 4:50SPEAKER_00: Building and running data centers is highly capital intensive.
- 4:54SPEAKER_01: Yeah, buying all those servers.
- 4:55SPEAKER_00: Exactly, which inherently drags down that gross margin percentage.
- 4:59SPEAKER_00: But further down the income statement, you have the operating margin.
- 5:03SPEAKER_00: This gets a lift because Microsoft enjoys immense pricing power on its software side.
- 5:09SPEAKER_01: Oh, I see.
- 5:09SPEAKER_00: Yeah, coupled with slower overhead growth.
- 5:11SPEAKER_01: So the software subscriptions are essentially acting as a high margin lever
- 5:15SPEAKER_01: to offset the heavy, expensive physical lifting required on the infrastructure side.
- 5:20SPEAKER_00: That is exactly the dynamic.
- 5:22SPEAKER_00: But, you know, that physical infrastructure requires a lot of capital,
- 5:25SPEAKER_00: which brings us to the balance sheet.
- 5:26SPEAKER_01: Let's hear it.
- 5:28SPEAKER_00: On September 30th, cash and short term investments sat
- 5:30SPEAKER_00: at sixty four point five one billion dollars,
- 5:33SPEAKER_00: which includes four billion in restricted investments.
- 5:35SPEAKER_01: OK, pretty healthy.
- 5:37SPEAKER_00: On the liability side, stated borrowings are $35.48 billion.
- 5:42SPEAKER_00: Then you add in recognized finance leases of $70.97 billion and operating leases
- 5:48SPEAKER_00: of $27.48 billion.
- 5:50SPEAKER_01: Okay, I have to stop you there because here is where it gets really intense.
- 5:54SPEAKER_01: Those recognized numbers are big,
- 5:56SPEAKER_01: but there is a massive distinction we have
- 5:59SPEAKER_01: to make regarding their future commitments.
- 6:01SPEAKER_00: Yes, the uncommenced leases.
- 6:03SPEAKER_01: Right. Aside from everything you just listed,
- 6:06SPEAKER_01: there are an additional $364.60 billion in uncommenced leases.
- 6:13SPEAKER_01: We are talking about hundreds of billions of dollars.
- 6:16SPEAKER_00: It's a staggering figure, yeah.
- 6:18SPEAKER_00: And it requires careful context so we don't misinterpret the balance sheet.
- 6:22SPEAKER_01: Go ahead.
- 6:23SPEAKER_00: So these are future commitments,
- 6:25SPEAKER_00: primarily for data centers and some conditional agreements
- 6:29SPEAKER_00: with terms ranging anywhere from one to 20 years.
- 6:31SPEAKER_01: 20 years. Wow.
- 6:33SPEAKER_00: Yeah. And crucially, these terms start rolling across fiscal 2027 to 2034
- 6:40SPEAKER_00: because they haven't commenced yet.
- 6:42SPEAKER_00: They are not liabilities already recognized on the September balance sheet.
- 6:46SPEAKER_01: Let me wrap my head around that.
- 6:48SPEAKER_01: They are signing 20-year leases for data centers that won't even kick off until the early 2030s?
- 6:54SPEAKER_00: Yeah, they are laying the groundwork now.
- 6:56SPEAKER_01: That is a massive lock-in of future capital.
- 7:00SPEAKER_01: And speaking of capital, we also saw them price a $25 billion notes offering
- 7:05SPEAKER_01: on October 29th, which is expected to settle on November 4th.
- 7:09SPEAKER_00: Yes, and it's really important not to jump to conclusions about that offering.
- 7:12SPEAKER_00: That is real financing arranged for general corporate uses.
- 7:16SPEAKER_01: Like what?
- 7:17SPEAKER_00: Like working capital, debt repayment, general operations.
- 7:21SPEAKER_00: We can't just assume outside funding magically erases their investment risk
- 7:25SPEAKER_00: or that all of those proceeds are earmarked exclusively for AI.
- 7:28SPEAKER_01: Right. It's not a magical AI slush fund.
- 7:31SPEAKER_01: This all loops back to our lemonade stand.
- 7:34SPEAKER_01: I want to look at the physical cash moving in and out of the door.
- 7:37SPEAKER_00: Let's do it.
- 7:38SPEAKER_01: Because reported CapEx was $45.20 billion,
- 7:42SPEAKER_01: but that actually differs from the cash spent on property and equipment,
- 7:46SPEAKER_01: which was $39.37 billion.
- 7:49SPEAKER_01: Operating cash was a massive $58 billion, which is up 29%.
- 7:54SPEAKER_01: Yet free cash flow fell 27% to $18.63 billion.
- 7:59SPEAKER_01: Walk us through the physical reality of that cash bridge.
- 8:02SPEAKER_00: Okay, let's trace the actual dollars.
- 8:04SPEAKER_00: You start with $20.94 billion in beginning cash.
- 8:09SPEAKER_00: The business operations perform beautifully, adding $58.00 billion from operations.
- 8:14SPEAKER_01: Okay, great so far.
- 8:16SPEAKER_00: But then you have to subtract $41.97 billion for investing.
- 8:21SPEAKER_00: You also subtract $16.64 billion for financing. Yeah.
- 8:26SPEAKER_00: Then factor in a miner minus $0.05 billion for currency effects,
- 8:31SPEAKER_00: and you end near $20.28 billion in ending cash.
- 8:36SPEAKER_01: So that investing line is just devouring the operational success.
- 8:41SPEAKER_00: It absolutely is.
- 8:42SPEAKER_01: And keep in mind, investing is a broader category than just CapEx.
- 8:46SPEAKER_01: It includes equipment purchases and complex supplier flows.
- 8:50SPEAKER_00: Right.
- 8:51SPEAKER_01: And what about the financing piece?
- 8:53SPEAKER_00: Financing includes debt repayments, stock repurchases, and dividends.
- 8:58SPEAKER_00: Now, inside that $58 billion in operating cash, there's a nuance.
- 9:02SPEAKER_00: You have $11.10 billion of depreciation.
- 9:06SPEAKER_01: Let's clarify depreciation for a second because that's an accounting term,
- 9:09SPEAKER_01: not a physical check they are writing today.
- 9:12SPEAKER_00: Exactly. Depreciation is spreading the cost of a past investment over time.
- 9:16SPEAKER_00: It is not new cash leaving the building today.
- 9:19SPEAKER_01: Okay, got it.
- 9:20SPEAKER_00: Interestingly, Microsoft made an accounting change in July,
- 9:23SPEAKER_00: extending the estimated useful lives of their server network equipment and buildings
- 9:27SPEAKER_00: to 15 to 25 years.
- 9:30SPEAKER_00: But that extension applies to the buildings, not the GPUs inside them,
- 9:33SPEAKER_00: which become obsolete much faster.
- 9:36SPEAKER_01: Good point. The GPUs are definitely not lasting 25 years.
- 9:40SPEAKER_01: So what does all this mean for someone looking to actually buy a piece of this business?
- 9:44SPEAKER_01: Let's frame the valuation using the November 2nd provider daily close
- 9:49SPEAKER_01: of $507.12 USD, which we checked on November 3rd before the U.S. Open.
- 9:56SPEAKER_01: Using that specific price and a trailing gap EPS of $19.15,
- 10:01SPEAKER_01: we're looking at an implied multiple of about 26.48 times earnings.
- 10:06SPEAKER_00: And based on that November 2nd $507.12 close and assuming net yield reinvestment,
- 10:13SPEAKER_00: The public weighted ratings across the advisor models are neutral
- 10:16SPEAKER_00: at the first horizon and buy at the longer one, plus a no action keep monitoring.
- 10:22SPEAKER_00: I want to be absolutely clear here, this is not a strong buy.
- 10:26SPEAKER_01: Let's look at the dates for those horizons.
- 10:28SPEAKER_01: The base model dates are fixed at 18 September 2027 and 18 September 2031.
- 10:34SPEAKER_01: Based on our reference price,
- 10:35SPEAKER_01: those models imply 5.41% and 58.38% cumulative price appreciation, respectively.
- 10:42SPEAKER_00: Yeah, but that's just the stock price.
- 10:44SPEAKER_00: Microsoft also pays you to wait.
- 10:46SPEAKER_01: They do.
- 10:47SPEAKER_01: The quarterly dividend is 98 cents per share, which equates to $3.92 annualized.
- 10:54SPEAKER_01: Though it's worth noting that only the December 10th,
- 10:56SPEAKER_01: 2026 payment is actually officially declared right now.
- 11:00SPEAKER_00: Correct.
- 11:01SPEAKER_01: But if we project that out, assuming flat future payments,
- 11:05SPEAKER_01: that means $3.92 per share before that 2027 date, representing four payments.
- 11:11SPEAKER_01: And it means $19.60 before the 2031 date, which represents 20 payments.
- 11:18SPEAKER_00: When you combine the price appreciation and those dividends,
- 11:21SPEAKER_00: you get a 6.18%
- 11:23SPEAKER_00: and 62.24% gross cash plus price return before taxes and fees
- 11:29SPEAKER_00: and without any reinvestment.
- 11:31SPEAKER_01: Pretty solid.
- 11:32SPEAKER_00: Yeah. And those numbers exclude September cash that was already paid while assuming
- 11:37SPEAKER_00: the later payments are made.
- 11:38SPEAKER_01: But those are base projections.
- 11:40SPEAKER_00: Yeah.
- 11:40SPEAKER_01: In a landscape moving this fast, we have to stress test this.
- 11:44SPEAKER_00: We do, which means looking at the qualitative stressed scenarios.
- 11:49SPEAKER_00: In a stressed bull scenario,
- 11:51SPEAKER_00: you see all these paid agents and Azure consumption actively converting
- 11:55SPEAKER_00: to higher customer revenue.
- 11:57SPEAKER_01: So they actually make the money back.
- 11:58SPEAKER_00: Right. Custom silicon and better utilization drive unit costs down
- 12:03SPEAKER_00: and cash conversion recovers
- 12:04SPEAKER_00: because their investment pace eventually slows down relative
- 12:07SPEAKER_00: to their cash generation.
- 12:09SPEAKER_01: Okay. And the bear scenario?
- 12:10SPEAKER_00: The bare scenario is where the hardware trap snaps shut.
- 12:13SPEAKER_00: You face fierce competitive token pricing that ruins margins
- 12:17SPEAKER_00: or disappointing returns on customer AI projects.
- 12:20SPEAKER_00: Or worse, all those massive hardware and power commitments we talked about continue
- 12:24SPEAKER_00: to drag on cash while the actual top line revenue weakens.
- 12:28SPEAKER_01: You can really see that tension in the four quadrant dynamics.
- 12:31SPEAKER_01: You have your ordinary drivers, right?
- 12:33SPEAKER_01: The immense Azure demand, the incredibly high switching costs
- 12:37SPEAKER_01: for enterprise workflows, the push toward higher value usage.
- 12:41SPEAKER_00: Absolutely.
- 12:42SPEAKER_01: But pushing against that, you have ordinary frictions,
- 12:45SPEAKER_01: the relentless infrastructure spend, the heavy depreciation hitting the books,
- 12:50SPEAKER_01: and inevitable pricing pressure.
- 12:52SPEAKER_00: And beyond the ordinary, you have both tails.
- 12:54SPEAKER_01: Let's talk about this.
- 12:55SPEAKER_00: Well, the opportunity tail is a future
- 12:56SPEAKER_00: where Reliable autonomous agents start executing valuable multi-step work
- 13:01SPEAKER_00: without human intervention or where custom silicon vastly improves unit costs
- 13:06SPEAKER_00: beyond current assumptions.
- 13:07SPEAKER_01: What about the risk tail?
- 13:09SPEAKER_00: Because with $364 billion in uncomminsed leases, the downside seems steep.
- 13:14SPEAKER_01: The risk tail is severe.
- 13:15SPEAKER_00: Imagine customers canceling uneconomic AI projects just
- 13:19SPEAKER_00: as all this new data center capacity finally comes online.
- 13:22SPEAKER_01: Yikes.
- 13:22SPEAKER_00: Right? Or consider physical constraints.
- 13:25SPEAKER_00: Grid restrictions could literally strand expensive equipment,
- 13:29SPEAKER_00: meaning Microsoft paid for servers they can't even power up.
- 13:32SPEAKER_01: Oh, wow. That would be brutal.
- 13:34SPEAKER_01: This brings us to a concept in the research that I think is the biggest aha moment
- 13:37SPEAKER_01: of this entire deep dive, the alpha gap.
- 13:40SPEAKER_00: Yes, a really important concept here.
- 13:42SPEAKER_01: Because when you see Azure growing at 48%, you just assume the stock is a screaming buy.
- 13:48SPEAKER_01: But the research shows there is no near-term bargain here just because demand is impressive.
- 13:53SPEAKER_00: Exactly. The market is incredibly efficient, but it has blind spots.
- 13:57SPEAKER_00: Wall Street might be underestimating Microsoft's massive enterprise distribution advantage.
- 14:01SPEAKER_00: You know, their ability to push AI into software that Fortune 500 companies already use.
- 14:06SPEAKER_01: Sure.
- 14:06SPEAKER_00: But simultaneously,
- 14:08SPEAKER_00: the market is overestimating how quickly this capital intensive AI revenue actually
- 14:12SPEAKER_00: turns into free cash flow.
- 14:14SPEAKER_01: I was thinking about this like a city building a massive,
- 14:17SPEAKER_01: sprawling new public transport network.
- 14:20SPEAKER_01: Microsoft is out there laying down thousands of miles of new train tracks,
- 14:24SPEAKER_01: the data centers, the GPUs.
- 14:25SPEAKER_00: I like this analogy.
- 14:26SPEAKER_01: Right. Now, the demand is clearly there. The platforms are packed.
- 14:29SPEAKER_01: People want to ride the train.
- 14:31SPEAKER_01: But traffic can surge on new train lines long
- 14:34SPEAKER_01: before those lines actually repay their initial construction
- 14:37SPEAKER_01: and ongoing maintenance costs.
- 14:38SPEAKER_00: So true.
- 14:39SPEAKER_01: A ticket booking isn't cash in the bank yet.
- 14:42SPEAKER_01: And top line revenue is an owner profit.
- 14:44SPEAKER_00: That analogy hits the nail on the head.
- 14:47SPEAKER_00: You are watching them build the tracks,
- 14:49SPEAKER_00: but the market is pricing the stock as if the construction debt is already paid off.
- 14:53SPEAKER_01: Exactly.
- 14:54SPEAKER_00: And this is why there is such substantive disagreement among the AI advisory models we analyze.
- 15:00SPEAKER_00: We're looking at 14 correlated AI contributions here, spanning nine frameworks,
- 15:05SPEAKER_00: three versions, and 12 Gemini outputs.
- 15:07SPEAKER_01: How do they split on a giant like Microsoft?
- 15:10SPEAKER_00: Well, the high weight Claude polymath view is highly focused on the timing of it all.
- 15:15SPEAKER_00: It argues that while demand survives,
- 15:17SPEAKER_00: heavy depreciation hits the income statement long
- 15:20SPEAKER_00: before the new server capacity actually pays itself back.
- 15:23SPEAKER_01: Okay. And the other side?
- 15:25SPEAKER_00: Conversely, the high weight GPT polymath focuses on the long term recurring earnings
- 15:30SPEAKER_00: after the equipment and leases are finally paid off.
- 15:33SPEAKER_00: But it warns that a phenomenally fine business can still be just
- 15:36SPEAKER_00: an ordinary investment if you pay too high a price for it today.
- 15:40SPEAKER_01: It's the classic difference between a great company and a great stock.
- 15:43SPEAKER_01: And then you have the more extreme perspectives on both sides.
- 15:46SPEAKER_00: Yes. The Burry-style cautious view asks a structural question.
- 15:51SPEAKER_00: Does the constant relentless need for hardware replacement,
- 15:53SPEAKER_00: like refreshing GPUs every few years,
- 15:56SPEAKER_00: mean this low-cash conversion is a structural permanent reality rather
- 16:00SPEAKER_00: than a temporary build-out phase?
- 16:02SPEAKER_01: Right. And the optimistic view?
- 16:05SPEAKER_00: On the flip side, the titan
- 16:07SPEAKER_00: or value optimistic view expects Microsoft's deeply entrenched distribution
- 16:11SPEAKER_00: to monetize this capacity seamlessly, eventually releasing a flood of cash.
- 16:16SPEAKER_01: So if you're listening to this and trying to figure out which side is right,
- 16:19SPEAKER_01: you can actually test both theories in real time.
- 16:21SPEAKER_00: Yes, you can.
- 16:22SPEAKER_01: You test the optimists by watching the least adjusted cash and you test the skeptics
- 16:27SPEAKER_01: by tracking billable utilization.
- 16:30SPEAKER_01: Are enterprise customers actually deploying and paying
- 16:33SPEAKER_01: for this massive infrastructure long term?
- 16:35SPEAKER_00: And when we look at the modeled quarterly path,
- 16:38SPEAKER_00: all 20 numerical weighted quarter to quarter forecast steps are positive.
- 16:42SPEAKER_01: Really?
- 16:43SPEAKER_00: All 20.
- 16:43SPEAKER_01: All 20.
- 16:44SPEAKER_00: The very first December 2026 target sits slightly below today's reference price,
- 16:49SPEAKER_00: but the path rises steadily from there.
- 16:52SPEAKER_00: The models suggest that these early windows will test cash conversion
- 16:55SPEAKER_00: and depreciation, while the fiscal 2027
- 16:57SPEAKER_00: and 2028 windows will really test utilization and monetization.
- 17:02SPEAKER_01: We have some concrete events coming up that will give us clues, too.
- 17:05SPEAKER_01: Ignite runs November 17th through the 20th.
- 17:08SPEAKER_01: The ex-dividend date is November 19th.
- 17:10SPEAKER_01: The annual meeting is December 8th.
- 17:12SPEAKER_01: And that declared dividend payment is December 10th.
- 17:15SPEAKER_01: And as we look at those future clues,
- 17:17SPEAKER_01: I want to bring up a metric that can be easily misunderstood. Which one?
- 17:22SPEAKER_01: The remaining performance obligations, or RPO.
- 17:25SPEAKER_01: It sits at a massive $677 billion.
- 17:28SPEAKER_00: It's a huge number, but we have to be precise about what it represents.
- 17:32SPEAKER_01: Exactly, because $677 billion sounds like a Scrooge McDuck vault of cash sitting in Redmond.
- 17:38SPEAKER_01: I want to clarify for the listener, RPO is Contracted Future Revenue Recognition.
- 17:43SPEAKER_00: Right.
- 17:43SPEAKER_01: It includes unearned amounts that might have already been paid and unbilled amounts.
- 17:49SPEAKER_01: It is a pipeline of corporate IOUs.
- 17:51SPEAKER_01: It is absolutely not a pile of all future cash available today.
- 17:55SPEAKER_00: A vital distinction. Right.
- 17:57SPEAKER_00: Another critical measure that requires clarity is the relationship with OpenAI.
- 18:01SPEAKER_01: Oh, this is a big one.
- 18:02SPEAKER_00: Microsoft earned $10 billion of commercial revenue from its customer, OpenAI, in this quarter.
- 18:09SPEAKER_00: That is Microsoft's revenue generated by providing services to OpenAI.
- 18:14SPEAKER_00: It is not OpenAI's own total consumer sales.
- 18:17SPEAKER_01: And we can see OpenAI's footprint on the balance sheet, too.
- 18:21SPEAKER_01: OpenAI accounts receivable was $7.20 billion in September, up from $6 billion in June.
- 18:27SPEAKER_00: Which are just snapshots.
- 18:28SPEAKER_01: Exactly. Those are just point in time balance sheet snapshots.
- 18:32SPEAKER_01: You absolutely cannot infer actual cash collections just by subtracting those two numbers.
- 18:36SPEAKER_00: No, you can't.
- 18:37SPEAKER_00: Now, to truly understand the momentum here,
- 18:40SPEAKER_00: we have to look at the genuine business changes that occurred over the last batch
- 18:44SPEAKER_00: of reporting.
- 18:45SPEAKER_01: Let's map it out.
- 18:45SPEAKER_00: If we map out the chronology,
- 18:47SPEAKER_00: our previous July 5th research came before the actual July 29th financial results.
- 18:52SPEAKER_01: And those results, which were fully available for the September assessment, were massive.
- 18:57SPEAKER_01: Cloud hit $59.30 billion of 27%,
- 19:02SPEAKER_01: and their AI Foundry had 100,000 customers.
- 19:06SPEAKER_01: Furthermore, Foundry revenue grew more than two times year on year in July.
- 19:11SPEAKER_00: Which brings us to the October update we are dissecting today.
- 19:15SPEAKER_00: Cloud grew to $63.90 billion, and Foundry now has over 120,000 customers.
- 19:22SPEAKER_01: Which I have to explicitly note for the listener,
- 19:25SPEAKER_01: going from 100,000 to over 120,000 is roughly 20% more customers.
- 19:32SPEAKER_01: We are not saying customers tripled or doubled,
- 19:34SPEAKER_01: and we are certainly not saying Foundry tripled to 120,000 customers.
- 19:39SPEAKER_01: Customer counts are not dollar amounts.
- 19:42SPEAKER_01: Separately though, Foundry revenue more than tripled year on year in October.
- 19:46SPEAKER_00: These are real tangible operating changes on the ground,
- 19:49SPEAKER_00: not just shifts in research sentiment or analyst panel counts.
- 19:53SPEAKER_00: And this entire narrative, the painful transition
- 19:55SPEAKER_00: from heavy physical infrastructure investment to actual owner cash,
- 20:00SPEAKER_00: is a theme we've explored before.
- 20:01SPEAKER_01: Oh, heavily.
- 20:02SPEAKER_01: Listeners of the Deep Dive will immediately recognize the parallels to Alphabet,
- 20:07SPEAKER_01: EP5, where we directly compared massive infrastructure investment
- 20:11SPEAKER_01: to actual owner cash.
- 20:12SPEAKER_00: Yes.
- 20:13SPEAKER_01: Or Amazon, EP28, where we looked at cloud profits versus
- 20:17SPEAKER_01: the sheer physical investment required to sustain them.
- 20:20SPEAKER_01: And even NVIDIA, EP13, which gave us the computing supplier's perspective
- 20:25SPEAKER_01: on where all this hardware spending is going.
- 20:27SPEAKER_00: It also draws a surprisingly clean parallel to the very last stock we analyzed, GSK36.
- 20:33SPEAKER_01: Yes. Now, they operate in completely different universes.
- 20:36SPEAKER_00: Very different.
- 20:37SPEAKER_01: With GSK, the pharma giant, they faced the massive capital burden
- 20:41SPEAKER_01: of funding new medicines and replacing expiring patents.
- 20:45SPEAKER_01: Microsoft faces the burden of funding server replacement,
- 20:48SPEAKER_01: navigating complex 20-year leases,
- 20:51SPEAKER_01: and the relentless need for software monetization.
- 20:54SPEAKER_00: Different risks, same question.
- 20:56SPEAKER_01: Exactly. They are entirely different risks,
- 20:59SPEAKER_01: but fundamentally it boils down to the exact same core question for the investor.
- 21:03SPEAKER_01: What actual cash remains in the till after sustaining the business?
- 21:08SPEAKER_00: Which naturally leads us to invalidation.
- 21:11SPEAKER_00: If you are holding this stock or monitoring this thesis,
- 21:14SPEAKER_00: What should you watch to see if the models break?
- 21:17SPEAKER_00: First, you need to watch if Azure Delivery actually meets management's optimistic plans.
- 21:23SPEAKER_00: Second, watch if their premium pricing survives
- 21:25SPEAKER_00: as token competition inevitably heats up across the industry.
- 21:29SPEAKER_00: And crucially, watch if gross profit can consistently outgrow
- 21:33SPEAKER_00: that heavy depreciation we discussed.
- 21:35SPEAKER_01: And don't forget to watch the financing piece.
- 21:38SPEAKER_01: We mentioned that $25 billion notes offering that priced on October 29th
- 21:42SPEAKER_01: and is expected to settle November 4th.
- 21:44SPEAKER_01: Yeah, that's key.
- 21:45SPEAKER_01: You need to watch the actual settlement and how they allocate that capital.
- 21:49SPEAKER_01: Do not just assume that outside funding erases the underlying investment risk.
- 21:54SPEAKER_00: Exactly. The bearish case weakens significantly
- 21:57SPEAKER_00: if profitable utilization recovers despite all this spending.
- 22:00SPEAKER_00: But the bullish case weakens if demand grows but without adequate cash returns to the owner.
- 22:05SPEAKER_01: So synthesizing all of this, it brings us right back to the tension we opened with.
- 22:10SPEAKER_01: Strong earnings absolutely challenge a simplistic bubble story.
- 22:14SPEAKER_01: Microsoft is not selling vaporware.
- 22:16SPEAKER_01: There is real, tangible paying demand.
- 22:18SPEAKER_00: Absolutely.
- 22:19SPEAKER_01: But the cash bill and that starting valuation multiple dictate the actual investment case.
- 22:25SPEAKER_01: A neutral to buy rating is highly nuanced.
- 22:28SPEAKER_01: Paying demand does not automatically mean the stock is a near-term bargain.
- 22:32SPEAKER_01: And it doesn't guarantee that the returns
- 22:34SPEAKER_01: on all this expensive depreciating equipment will be adequate.
- 22:37SPEAKER_00: Well said.
- 22:38SPEAKER_01: For the bigger picture, visit iPulse AI to explore our complete rankings of covered assets.
- 22:43SPEAKER_01: Discover the strongest buys and see the research behind the ratings all in one place.
- 22:49SPEAKER_01: Primary research conducted through the iPulse AI platform by Future Edge Group.
- 22:53SPEAKER_01: If you found this deep dive valuable, please like, comment, and subscribe.
- 22:57SPEAKER_01: We are publishing a new podcast every day of the week except Thursday.
- 23:01SPEAKER_01: Stay tuned for the next podcast on RxSight.
- 23:04SPEAKER_00: RxSight lets doctors adjust the implanted lens after cataract surgery.
- 23:08SPEAKER_00: So its business economics are the next question to explore there.
- 23:11SPEAKER_01: Should be a fascinating one.
- 23:12SPEAKER_00: Yeah.
- 23:13SPEAKER_01: Until then, keep questioning the cash flow.
- 23:15SPEAKER_01: And thanks for joining us on this deep dive.
- 23:17SPEAKER_00: Take care, everyone.