In one breath
Meta’s ads business is growing faster than almost any company its size, with Q2 2026 revenue up 28% year-over-year. But the company is now spending so heavily on AI infrastructure that free cash flow, the actual cash left after building things, collapsed to near zero for the quarter. The stock is priced below its own decade-median multiple, meaning the market is paying less per dollar of profit than it historically has. The open question is whether the spending is a bridge to a bigger business or a bet that doesn’t pay off on any near-term timeline.
What arrived, and the sentence that matters
On July 29, 2026, Meta filed its Q2 2026 earnings release, and the sentence that stops you cold is this one: capital expenditures for the quarter more than doubled year-over-year. That single line explains almost everything else in the filing. Revenue grew. Profit shrank. Cash nearly vanished. All of it traces back to one decision: Meta is building AI infrastructure at a pace the business has never attempted before.
To understand why that matters, consider the backstory. In 2022, Meta was in genuine trouble. Revenue fell for the first time in the company’s public history, the 10-K for fiscal year 2024 shows net margin had collapsed well below where it would eventually recover, and the stock lost roughly two-thirds of its value. Zuckerberg declared a “year of efficiency,” cut headcount, and rebuilt margins to nearly 38% by fiscal year 2024. That recovery is what makes the current moment so legible: the company earned back the market’s trust, and is now spending it.
What the filing changed
Three things shifted materially between Q2 2025 and Q2 2026. Operating margin fell sharply, meaning for every dollar of revenue, significantly fewer cents reached operating profit. Net income dropped double digits year-over-year. And free cash flow, which was measured in the tens of billions annually not long ago, landed at $784 million for this quarter, roughly what a mid-sized regional bank earns in a year, not what a company of Meta’s scale usually generates in ninety days.
Two items inside the expense line deserve their own sentence. The Q2 results release shows substantial legal charges and severance costs sitting inside that quarter’s expenses. These are one-time in character, meaning they inflated the damage for this specific quarter. Strip them out mentally and the margin picture is still compressed, but less dramatically so. The underlying AI spending pressure is real and recurring; the legal and severance hits are not.
Headcount fell slightly year-over-year, the result of a May 2026 reduction affecting roughly thousands of people. Most of those departures won’t clear the books until Q3, so some severance cost is still coming.
The bet being placed
Meta’s Q2 prepared remarks raised the full-year 2026 capital expenditure guidance above the range disclosed in the fiscal year 2025 10-K. To put that in human terms: Meta is planning to spend more on building things this year than the entire company earned in revenue just a few years ago. The majority goes to AI infrastructure, data centers, servers, and chips.
The strategic logic, as Zuckerberg stated on July 29, is that AI is already accelerating the core ads business. Ad impressions and advertising revenue both grew meaningfully in Q2, and MediaPost, citing eMarketer projections, reported that Meta is on track to surpass Google in global digital ad revenue for the first time in 2026. That’s the return the spending is supposed to buy.
Wedbush analyst Ygal Arounian, as reported by Investing.com, said the market needs clearer signs of monetization before the AI investment thesis closes. That’s a fair characterization of where the conversation sits: revenue is responding, but the path from compute spending to incremental profit is not yet drawn in filed numbers.
What the filing cannot tell us yet
Reality Labs, the division building augmented and virtual reality hardware, posted a large operating loss against modest revenue in Q2. The fiscal year 2025 10-K shows Reality Labs lost money at a rate that exceeded its revenue by nearly nine times across the full year. That division remains a long-duration wager with no filed timeline for profitability.
The AI infrastructure buildout carries the same uncertainty. Meta holds more cash than long-term debt, so the balance sheet can absorb the spending. But the return on a year of capital expenditure at this scale is not a number any filing can supply today. Management guided Q3 2026 revenue to a range that would represent continued strong growth, but the margin trajectory depends entirely on whether legal charges and severance costs fade as expected while AI spending stays elevated.
Reading the numbers
Operating margin, 31% in Q2 2026 vs. 43% in Q2 2025. An operating margin is what’s left of each dollar of revenue after paying all the bills to run the business, before taxes and interest. At 43%, Meta was keeping 43 cents of every dollar. At 31%, it keeps 31 cents. A household analogy: if a family earning $10,000 a month used to save $4,300 after expenses and now saves $3,100, the income grew but the spending grew faster.
Free cash flow, $784 million for Q2 2026. Free cash flow is the cash actually left after building and maintaining the business, the number that can fund buybacks, dividends, or debt repayment. One quarter ago this figure was measured in the tens of billions annually. At $784 million for a single quarter, Meta is essentially reinvesting every dollar it earns right now. That’s not distress, it’s a choice, but it’s a choice that leaves little margin for error.
P/E of 25.3 vs. a decade median of 29.2 (our data). A price-to-earnings multiple, or P/E, is the price tag the market puts on each dollar of annual profit. Think of it as how many years of today’s earnings a buyer is paying upfront. At 25.3, Meta’s multiple sits in the 29th percentile of its own decade-long range, meaning the market has valued it more richly about 71% of the time. If earnings recover toward the filed trend and the multiple returned to its decade median, the math would look different from today’s price. That’s a conditional scenario, not a forecast, and it depends entirely on whether the spending produces the profit growth management expects.
Sources
- Meta Q2 2026 earnings release, July 29, 2026
- Meta Q2 2026 prepared remarks
- Meta 10-K, fiscal year ended December 31, 2025
- Meta 10-K, fiscal year ended December 31, 2024
- Investing.com, Q2 2026 earnings call transcript
- MediaPost, citing eMarketer 2026 ad revenue projections
- Daily Maverick, Meta cash flow analysis, July 30, 2026
