Learning to read a company through 3 lenses (value, growth, quality). Tried it on Microsoft, is 23x fair or am I missing something?

I have been teaching myself to look at a company through three separate lenses, Value, Growth and Quality, and writing down what each one says on its own. MSFT was the boring one to start with. Numbers are from the FY25 10-K (filed 2025-07-30), price around $389, market cap about $2.89T. Posting my read to get corrected where I am off.

The tension I keep coming back to: You are paying a premium (P/E \~23, P/S \~9) for a business that has to keep growing around 15% at 39% net margins, while it spends about $97B a year on capex for the AI buildout, which is squeezing free cash flow right now. Clearly a great business. Not an obviously cheap price.

**Value:** P/E 23, P/S 9, P/B 7. Not cheap on any absolute basis. I think the multiple is defensible given the fundamentals, but it assumes execution holds and doesn't leave much room for a growth slowdown or a margin slip. If it went back to a market-average 18x, that's roughly a 20% drop from here just on the re-rate. So you are not really getting paid to be wrong.

**Growth:** Revenue up about 15% year over year, in line with the \~13% growth rate over the last nine years, and net income has compounded closer to 19% a year over that stretch as margins expanded. The part I keep chewing on is capex. That $97B is about 31% of revenue, and it's pulled free cash flow margin down from \~27% to \~25%. Honestly this is the piece I'm least sure how to read: is $97B of AI datacenter buildout a strength (funding the next decade) or a warning (huge money on a bet that isn't proven to pay yet)?

**Quality:** This is where it's clearly excellent and I didn't have much to argue with. 39% net margin, 30% ROE, 27% ROCE, all trending up over the decade. Balance sheet is about as clean as it gets: \~$72B net cash, debt/equity 0.10, interest coverage around 51x. $170B of operating cash flow turns into \~$73B free cash flow, and the gap is the capex reinvestment, not a cash generation problem. Stock comp is \~$12B (about 4% of revenue), so steady dilution. The one real weak spot is gaming (they announced 3,200 Xbox layoffs on July 6, hardware's been slumping), but that's small against a $282B revenue base.

**Bull case:** Microsoft is basically the enterprise AI default. Copilot in M365, GitHub, and Azure plugs into customer relationships and switching costs that are hard to unseat. It's compounded revenue \~13% and net income \~19% a year for nine years while lifting net margin from 23% to 39%. The $97B capex is building the infrastructure for the next decade, and with $72B net cash and $170B operating cash flow they can afford it and still weather a bad year. If AI monetization shows up, 23x isn't expensive for that.

**Bear case:** 23x and 9x sales assume \~15% growth and \~40% margins hold basically forever, with no room for error. AI monetization at scale still isn't proven, and free cash flow margin has already slipped from 27% to 25% as capex surged. If Copilot attach rates stall or open models commoditize the AI layer, that $97B a year turns into a drag instead of a growth engine.

**Where I land:** Honestly somewhere in the middle, which is why I'm posting. The quality isn't in question. The whole thing rests on whether the AI spend converts, and I can't tell that from the filing yet.

So, is 23x fair for a business this good with a balance sheet like that, or does the $97B capex mean you're funding an unproven AI bet at a price with no room for error? And if you'd score any of the three lenses differently than I did, I'd genuinely like to hear why.