META — Meta Platforms
The ad machine is getting stronger. Management is finding larger things to spend it on.
Meta owns Facebook, Instagram, WhatsApp and one of the best ad machines ever built. AI is already helping the core business. The weird part is what happens after that: management takes this ridiculous cash engine and points it at infrastructure and Reality Labs on a scale that would make most companies faint. I like the business. I am much less automatic about loving every use of the cash.
Meta is one of the weirder positions I own because the operating thesis keeps getting easier while the capital-allocation thesis somehow keeps getting harder.
Why I care about this one
The reason I own Meta is still boring: the Family of Apps is an absurd distribution and monetization machine.
AI can improve recommendations, engagement and ad targeting inside products billions of people already use. Meta does not need consumers to adopt some completely new behavior for that part of the AI thesis to work.
That is a very good setup.
Then management looks at the cash machine and goes, cool, what else can we build?
Q2 revenue reached $60.8B, up 28%. Family daily active people reached 3.60B. Ad impressions rose 14% and average price per ad rose 12%. (Q2 results)
The ad engine is fine.
Operating margin fell to 31% from 43% a year ago. Capex plus finance-lease principal was $31.1B in one quarter. Full-year 2026 capex guidance is now $130B–$145B. (Q2 results)
Free cash flow was $784M.
Some of that is timing. Some of it is literally the point I am worried about.
Meta is taking an incredibly profitable existing business and choosing to turn a very large portion of the cash into AI infrastructure, technical talent and long-duration bets.
Q2 2026
Primary sources: Q2 2026 results ↗
The operating model stays around $28 of normalized EPS with 7% / 15% / 20% growth and 18x / 27x / 31x terminal multiples. The implementation got cleaned up: buy bands now derive from the same anchor at 26x / 24x / 22x, roughly $728 / $672 / $616. Q2 made the ad thesis stronger and the capital-allocation question louder, with 2026 capex now guided to $130B–$145B. Capital Allocation stays 6.5/10.
My position
What I actually bought and sold.
| Date | Type | Account | Dollars | Shares | Price | Return now* |
|---|---|---|---|---|---|---|
| Mar 27, 2026 | Buy | Taxable | $3,500 | 6.496 | $539 | +39.5% |
| Mar 30, 2026 | Buy | Taxable | $5,000 | 9.331 | $536 | +40.3% |
| Apr 10, 2026 | Buy | Taxable | $500 | 0.796 | $628 | +19.7% |
| May 12, 2026 | Buy | Taxable | $6,232 | 10.485 | $594 | +26.5% |
| Jun 25, 2026 | Reinvest | Taxable | $14.23 | 0.025 | $569 | +32.1% |
*Return now uses the latest available quote for buys and reinvestments. Sales and fees show “—”.
Valuation & scenario math
I use roughly $28 of normalized EPS as the operating anchor.
Bear is 7% EPS growth at 18x. Base is 15% at 27x. Bull is 20% at 31x.
The base lands around $1,521 per share. The living page calculates the return from today’s quote to that endpoint.
That is good enough for me to keep liking the stock.
It is not enough to make me forget that the position is already large relative to the 3%–6% target.
The buy framework now derives directly from the same $28 anchor: roughly $728 at 26x, $672 at 24x and $616 at 22x.
No more having the printed multiples and dollar bands quietly disagree with each other.
26x / 24x / 22x normalized earnings
3.2% above $728 starter
| Band | Buy price | Base return to Sep 2031 |
|---|---|---|
| Starter | $728 | +16.0% |
| Normal | $672 | +17.9% |
| Aggressive | $616 | +20.0% |
Even if the stock enters a buy zone, my position size still matters. Meta is already above the top of the dated 3%–6% target range. A good price is not permission to ignore concentration.
I model the core ad engine and leave the spending problem inside the assumptions. The premium outcome only works if AI-driven monetization keeps outrunning the infrastructure and Reality Labs bill. 'Meta has great ads' is not enough by itself.
| Normalized EPS today | $28 |
| Annual EPS growth | 15% |
| EPS by Sep 25, 2031 | ~$56 |
| Terminal P/E | 27x |
| Implied base price | ~$1,521/share |
The base case is basically ~$56 of Sep 25, 2031 EPS at 27x, or about $1,521/share. Same math, just without making the reader decode equation soup.
| Case | What has to be true | Target range | Annualized return from today |
|---|---|---|---|
| Bear | 7% annual EPS growth · 18x terminal P/E. The ad engine grows, but AI and Reality Labs spending absorb much of the incremental economics and the market de-rates the stock. | $657–$756 center ~$707 | -1.2% |
| Base | 15% annual EPS growth · 27x terminal P/E. AI keeps improving engagement and monetization, per-share earnings compound around the mid-teens and the infrastructure bill earns a good enough return. | $1,414–$1,627 center ~$1,521 | +15.3% |
| Bull | 20% annual EPS growth · 31x terminal P/E. AI materially lifts monetization, new products contribute real revenue and infrastructure intensity eventually normalizes enough to create operating leverage. | $2,009–$2,311 center ~$2,160 | +23.8% |
Returns use the current quote and the time remaining to Sep 25, 2031. The scenario targets stay fixed until the underwriting changes.
Capital allocation
Great cash generation. Management also has some extremely expensive hobbies.
The core AI spend is already showing up in engagement and monetization, so I give that real credit. Reality Labs, giant infrastructure commitments and very broad management discretion are why the score is nowhere near perfect even though the balance sheet can handle it.
Stage-relative, not some objective rating-agency grade. The number is there to force the judgment into the open. See the rubric →
What management is actually doing with the cash
Meta is why I separate business quality from capital allocation. AI infrastructure has a pretty direct link to the ad moat and there is already operating evidence behind it. Reality Labs is a much longer-duration bet. With 2026 capex guided to $130B–$145B, management has to keep proving this is reinvestment rather than just a very profitable company discovering it can spend almost anything.
What I’m doing with the next dollar
Probably not Meta right now.
The position is already above the top of my target range, so the sizing answer is pretty easy for now.
The live buy framework can move into a better zone without changing the sizing answer. Even if the stock gets cheaper, the next dollar still has to justify making an already-large exposure even larger.
Conviction is allowed to stay high while position size says no.
Honestly, that is exactly when the sizing rules are useful. Conviction is easiest to manage when it is low.
What can break this
Facebook and Instagram do not have to break for this to go wrong.
Meta can simply spend faster than the core business can economically justify.
The AI infrastructure has a clear connection to engagement, ads and future products. Reality Labs has a much longer and less proven path. Legal and regulatory costs are not theoretical either.
Meta can afford almost all of this.
Cool. “Can afford it” and “best use of the money” are still two different sentences.
The spending gets easier to defend if engagement, ad economics and new monetization keep improving while infrastructure intensity eventually starts to normalize.
The patience runs out if capex becomes a permanently rising percentage of the cash engine, Reality Labs keeps absorbing enormous capital without stronger evidence, or the ad machine itself starts weakening underneath it.
I still think Meta is one of the best businesses I own.
And yeah, I also think I own enough of it.
The evidence I’m watching
AI investment keeps improving engagement and ad monetization, new products start becoming real revenue streams and infrastructure intensity falls relative to the cash coming out of the core business.
Capex stays structurally huge without enough incremental monetization, Reality Labs keeps eating capital with weak evidence, or the Family of Apps ad engine itself starts losing momentum.