Meta Is "Cheap" for a Reason. The Duel Is Over Whether the Reason Will End.

Meta Is "Cheap" for a Reason. The Duel Is Over Whether the Reason Will End.

Meta grew second-quarter revenue 28% to a record $60.80 billion, beat the Street's sales estimate by about $0.6 billion, and its shares still fell roughly 10% the night of the report. On July 29 the same numbers produced two verdicts.

Bulls read the collapse as a mood, not a math problem: a mega-cap compounding revenue at 28% trades at roughly 20 times forward earnings — a discount to Microsoft, Apple and Nvidia — because the market is terrified of the $130–145 billion Meta will spend on AI infrastructure this year, and that is nearly double last year's bill. Bears read the sell-off as delayed recognition: in the very quarter revenue accelerated, free cash flow collapsed to $784 million, down about 91% from a year earlier, and earnings per share fell 13% even though sales beat.

At this price, only one of those readings has much room for error. Everything below is the fight over a single variable: whether the biggest buildout in the company's history becomes the next leg of earnings, or a permanent tax on profit.

Shared facts

As of the Aug. 24 close, both camps sign the same card:

• Price and scale: ~$559 a share, roughly $1.42 trillion market value on ~2.5 billion shares. The stock is down about 26% over the past year and about 29% below its 52-week high of $790.80; the low is $520.26.
• Business: the last four reported quarters (Q3 2025 through Q2 2026) produced ~$228 billion of revenue, growing about 28% a year. Second-quarter ad revenue was $59.36 billion, up 27%, with impressions up 14% and the average price per ad up 12%.
• Profit: Q2 operating income was $18.78 billion, down 8%; operating margin was 31%, down from 43% a year ago — the biggest single sign of how much the company is now spending to make money.
• Cash:~$90 billion of cash and marketable securities against ~$84 billion of long-term debt at June 30 — roughly net cash. But Q2 capital spending of $31.1 billion consumed 98% of the quarter's $31.9 billion of operating cash flow, leaving that $784 million of free cash flow. Dividends paid ($1.35 billion) exceeded it, and buybacks were zero.
• Guidance: full-year 2026 capex of $130–145 billion, with the floor raised $5 billion in July; total expenses of $165–169 billion; and Q3 revenue guided at $61–64 billion (below the ~$63 billion consensus); and a claim that full-year operating income will still top 2025.
• Users: daily active people across the family of apps averaged 3.60 billion in June, up 3%, barely above Q1 and marginally under the Street's 3.61 billion estimate.
• Consensus: forward earnings sit near $27–30 a share, so the ~20x forward multiple is real rather than an artifact. The most recent consensus I could find (Aug. 22) put the average analyst target near $775 on roughly 70 ratings, with AInvest's aggregate signal also labeling the stock a Buy. That gap — ~$559 versus ~$775 — is the market's disagreement with the crowd, not a fact.

The trailing earnings picture needs one adjustment before it means anything. Reported TTM earnings came to $26.55 a share, for a ~21x trailing multiple. But Q3 2025 carried a $15.93 billion one-time non-cash tax charge (it was signed into law in July 2025), and Q1 2026 benefited from an $8 billion one-time non-cash tax item. Strip both — my calculation, roughly $6 a share out and $3 back in at ~2.5 billion shares — and trailing earnings power is ~$29–30 a share. The pattern underneath is the whole story: in each of the last four reported quarters, tax-adjusted earnings per share landed in a flat $7-to-$9 band while revenue grew 25%-plus. Even the tax-adjusted profit has gone nowhere while sales compounded.

That flatness is one shared fact, and it is the reason the stock de-rated. The camps only disagree on what the flatness means.

Round 1: where the money is going

The bull. This is the Amazon play: operating cash flow of $31.9 billion in a single quarter proves the engine works; the $31.1 billion of capex is the point of the exercise. Meta ran this play twice before. It spent heavily on AI and data centers into 2024–2025, and each time the ad auction got smarter and more profitable. The stock's slide to ~20x forward has priced in far more pessimism than a business growing revenue 28% deserves, and the squeeze is front-loaded: one-time legal and severance charges wrote down Q2 by $3.6 billion, and add those back and operating income actually rose ~9% from a year ago.

The bear. The cost line is the tell. Total expenses rose 55% while revenue rose 28%. Even after removing the $3.6 billion of one-time legal and severance items, costs grew about 42% — far ahead of sales. That is not a rounding problem; that is a compounding cost structure. The depreciation on the $130–145 billion buildout has not yet peaked. The tax rate rose to 16% from 11% a year ago and is guided higher through 2026. And for the first time in years the company stopped buying back stock and paid out more in dividends than it generated in free cash flow. The bear can accept that the ad machine works and still show that, at the margin, every incremental dollar of revenue is currently being handed over to the buildout.

Round to the bear on the near-term P&L, with a visible concession: the one-time items will roll off, and management has explicitly told investors operating income for the full year will still beat 2025. The dispute is only how many quarters of this lie ahead.

Round 2: is a 27% ad business real?

The bull. Pricing power is the prize, and this quarter had it: impressions up 14% and the average price per ad up 12% in the same period. A commodity auction cannot raise prices and volume at once. Meta says its large language models improved ad clicks 8.3% and lifted conversions on Facebook 15.7% — a current, monetized payoff from the AI spending, not a promise. The family of apps still printed $23.4 billion of operating profit in one quarter, and a new slice of business is emerging off the ad engine: "other revenue" crossed $1 billion for the first time, up 73%, from business agent subscriptions and AI tools used by a million businesses a week. That is what diversification looks like before it shows in the headline.

The bear. The growth is increasingly monetization of a saturated audience. Daily active people grew only 3%, and slightly below the analysts' number — the expansion story is over. From here, growth must come from raising prices on advertisers, and eventually advertisers push back. The legal-regulatory web is thickening: live youth-safety litigation in the U.S., Canada moving to bar under-16s from social media, and a $2.4 billion legal charge already sitting in the quarter's results. And Reality Labs still burns ~$4.6 billion a quarter on $431 million of revenue. The bull keeps calling the AI spend an investment; the bear notes the whole thing is forced — rivals like Google, OpenAI and Amazon set the pace, and Meta either matches the capex or loses the ad-tech arms race.

Round to the bull on the evidence of the quarter (pricing power plus conversion), with this honest concession: the patent pool of new users is nearly tapped. The durability of this growth rests on how long Meta can keep extracting more revenue per existing user — and on the courts.

Round 3: who pays for the next $100 billion?

This is the variable that could move either case most. Meta says "hundreds of billions" of spending lies ahead before the end of the decade. Its annual operating cash flow is ~$130 billion; after 2026's capex it is roughly zero free cash. So where does the rest come from?

The bull. The balance sheet is net cash, the cash flow engine is the strongest in the industry, and debt capacity exists. When the Financial Times reported in June that Meta was weighing an equity raise of tens of billions, and Meta called the story "pure speculation," no banks had been hired, and no offering has come. A dilutive raise at a stock that is down 26% would be a self-inflicted wound management has so far declined to take.

The bear. The precedent is already set: Alphabet just completed roughly $85 billion of equity offerings to fund its AI buildout, and the June report knocked 6% off Meta in a single day — the market already prices dilution as likely. The honest math is that "hundreds of billions" of capex cannot be funded by $130 billion of annual operating cash flow plus a stock that no longer buys itself back. One of the three doors — equity, heavy leverage, or a slowdown of the buildout — opens eventually. A raise at these levels transfers value from existing holders to new ones at the possible worst timing.

Round to the bull on the current facts (no raise done, net cash), with the admission that this is the single binary that could flip the entire outcome. It converts the bull's share-count and EPS math into the bear's overnight.

What the price already demands

Put both stories through the same frame, not different ones. Call normalized earnings power ~$29 a share and forward consensus ~$28 with the squeeze embedded. The stock at $559 and ~20x forward, ~13x enterprise value-to-EBITDA (the lowest of the six largest U.S. tech companies, against roughly 17–24x for Amazon, Microsoft and Alphabet), is not requiring earnings growth. It is requiring that earnings merely stop falling, and is charging a P/E discount of roughly 7 to 15 points below Microsoft, Apple and Nvidia on top of that.

Reverse-engineer the numbers. For the stock to reach even the ~$775 average analyst target, EPS needs to hold roughly flat while the multiple normalizes ~4–5 points — a modest ask on a business compounding revenue 25%+. To reach it, the bear needs earnings to keep sliding on rising depreciation, a higher tax rate and possible dilution, and the discount to persist. One side is betting on stabilization and re-rating; the other on a two-year earnings decline on top of an already-depressed multiple. The market has effectively already paid the bear's facts. The remaining wager is the direction of the next four delivered quarters.

Variable	Shared fact	Bull's assumption	Bear's assumption	What decides it
Where capex lands	2026 capex $130–145B vs $72B in 2025	Front-loaded; FCF inflects and depreciation flattens in 2027	Multi-year treadmill to "hundreds of billions"; EPS keeps falling	FY2027 capex guidance and quarterly FCF
Growth durability	Q2 ad revenue +27%, impressions +14%, price +12%	AI raises conversion and pricing power; new revenue lines compound	Monetizing a 3%-growing, saturated base; legal caps reach	Ad price growth and daily active people over two quarters
Funding	Net cash; equity raise denied so far	No raise needed; debt capacity unused	Alphabet's ~$85B precedent and the 2030 spend target force dilution	An actual equity offering announcement

Ruling

Both the business case and the stock case go narrowly to the bull at today's price — but the near-term evidence belongs to the bear, and a buyer should expect the next two or three prints to keep looking like the bear's: EPS under pressure, free cash flow near zero, buybacks paused. The ruling rests on the gap between that weather and the price. The market has already taken the stock down 26% and de-rated it to ~20x forward, paying for the squeeze; what remains is whether the biggest capex program in the company's history produces an earnings and cash inflection, or a permanent depression of margins. The visible evidence — revenue accelerating to 28%, prices and conversions rising on AI it is spending to build, a new revenue line up 73%, a net-cash balance sheet — supports the inflection more than it supports the opposite. The bear's case is not absurd; it needs the cost-growth gap to persist for a year and dilution to arrive, which the price currently does not require. The bear carries the burden of proving the buildout is unrecoupable, and that is the stronger claim.

The ruling flips on named observations, on roughly a 12-month clock:

1. FY2027 capex guidance at the Q4 2026 report in early 2027. Above ~$150 billion, the buildout is still accelerating and the bear's treadmill thesis gains real ground.
2. Free cash flow. If quarterly FCF is still under a few billion dollars (or short of dividends) by mid-2027, "temporary squeeze" stops being sustainable.
3. An actual equity offering. A dilutive raise at these levels converts the EPS math to the bear's version instantly; today it is only a rumor Meta denied.
4. EPS direction. If normalized earnings are still falling year over year through Q4 2026 while revenue holds ~25% growth, depreciation is winning the race.

The bullish version of the same clock is a single quarter where free cash flow inflects while the capex number stops climbing. Until then, the market is paying you ~20 times flat earnings to find out whether the biggest AI bet in the industry ends in a dividend check or a dilution filing.