Revenue at the top of its own guidance, a profit miss caused by one-off charges, a higher minimum for spending, and almost no cash left over after that spending. One set of accounts shows the whole sum.
| Line | Expected | Delivered |
|---|---|---|
| Revenue | $59.5B (analysts); own guidance $58–61B | $60.8B (+28% on a year earlier), top of its own range |
| Earnings per share | $7.10–7.23 (analysts) | $6.18; a beat once one-off charges are set aside |
| Q2 costs | — | $42B, +55%; incl. $2.4B legal charge, $1.2B severance pay |
| 2026 capital spending, own guidance | $125–145B | $130–145B; bottom raised |
| 2026 costs, own guidance | $162–169B | $165–169B; bottom raised |
| Q3 revenue, own guidance | $62.7B (analysts) | $61–64B |
| Operating cash flow | — | $31.9B |
| Free cash flow | — | $0.8B |
Adverts shown up 14% on a year earlier, average price per advert up 12%: growth from volume and from price at the same time. Operating cash flow is the cash the business brings in from its normal operations; free cash flow is what is left after capital spending (capex: money spent on things that last for years, such as chips, buildings and power). Shares closed at $585.61 and traded at $529.15 after hours, down 9.6%.
On demand, Meta did everything this site can ask of a company that reports no separate AI revenue. Revenue grew 28%, to the top of Meta's own guidance, and the growth splits cleanly in two: 14% more adverts shown, at prices 12% higher. That is what better AI targeting looks like when it earns its money invisibly, through advertisers bidding more for better-aimed adverts, rather than appearing on any invoice. This is the "tool" outcome working as the front page describes it: AI that makes an existing business better, paid for out of existing budgets, and never appearing anywhere as AI revenue.
The question this page set for these results was whether Meta's commitment to the spending would hold. The answer was a raise: the bottom of its 2026 capital spending range moved from $125 billion to $130 billion, and the bottom of its full-year cost range rose with it. The profit miss, read closely, is not about the spending. Costs rose 55% to $42 billion, but $2.4 billion of that was a charge for legal proceedings and $1.2 billion was severance pay from the job cuts in May. Set those one-offs aside and Meta beat analysts' forecasts. The line that lasts is the cash line. The business brought in $31.9 billion of cash from its normal operations; roughly $31 billion of that went on capital spending, leaving $0.8 billion. The advertising business now pays for the build-out almost to the last dollar. One week, three ways of paying for it: Alphabet borrows and has negative free cash flow; Microsoft pays from its own cash with $19.6 billion to spare; Meta spends almost exactly what it earns, by choice.
Shares fell 9.6% in after-hours trading. Revenue beat forecasts, and the profit miss was largely one-off charges. That is the same reversal Alphabet met a week earlier, and twice as steep. Twice in one round of results, a company delivered and was punished for the spending that came with it. Investors have switched from rewarding AI spending to punishing it, and the patience clock this site tracks moved for the second time in seven days.
Meta remains the cleanest test of whether AI can earn its keep without replacing workers. It reports no AI revenue figure and no AI division, so it has no disclosure it could quietly drop, unlike Microsoft, which had one and stopped updating it. Everything shows up in how well the adverts perform. The revenue floor is the revenue the AI spending must earn every year to pay for itself (the front page's §2). The only way to measure Meta against it is what these results offered: price per advert times adverts shown, set against costs that are rising on a schedule Meta has already committed to. Both are rising. The bet is that the first keeps growing faster than the second.
Apple and Amazon reported on 30 July. Amazon's cloud business, AWS, answered the quarter's last open question on demand with 37% growth, and raised its spending bill at the same time. Apple, the control group, the big tech firm with no AI build-out of its own, was punished by investors without any build-out to blame. The July US jobs report, the other side of the ledger (do the jobs figures move with the results?), lands on 7 August. Nvidia reports on 26 August.