Meta kept just 20.1% of its third-quarter revenue as operating profit

Sales rose by a quarter, operating profit went the other way

Meta Platforms, the US company behind Facebook and Instagram, earns most of its money by selling advertising. In the third quarter of 2026 it took USD 64.6bn in revenue, up 26.1% from USD 51.2bn in the same quarter of 2025. The figures are consolidated, covering the whole group, in US dollars and under US GAAP, the American accounting rules. For a company of this size, that is a fast quarter.

Operating profit, what is left from revenue after the costs of running the business, did not follow. It fell from USD 20.5bn to USD 13.0bn, a drop of 36.8%. The operating margin, the share of each dollar of revenue that survives as operating profit, narrowed from 40.1% to 20.1%. That is a fall of 20.0 pp in a single year.

The bottom line points the opposite way, which is what makes the quarter easy to misread. Net profit rose 264.0%, to USD 9.9bn from USD 2.7bn, and the net margin widened from 5.3% to 15.3%. Everything that separates operating profit from net profit sits below the operating line: tax, interest, gains and losses on investments. A year earlier something on those lines held net profit unusually low, and the quarterly summary does not say which item it was. I would not read the jump in net profit as a sign that the business improved. The operating line is where the business itself shows up, and that line went down.

The extra revenue was spent before it arrived

The arithmetic is short. Revenue grew by USD 13.4bn against the same quarter of 2025. Operating profit fell by USD 7.5bn. So operating costs grew by more than the entire increase in revenue. Every additional dollar Meta earned was spent, and then some.

Margins usually shrink during growth for one of three reasons. The new business was won with lower prices. Input costs, such as wages or energy, rose faster than prices. Or the company is building capacity, people and overhead before that capacity pays for itself.

The first explanation fits poorly here. A company discounting its way to growth rarely beats expectations on revenue, and Kurzy.cz reported that Meta's third-quarter revenue came in above what analysts had forecast. The third explanation fits best. Alongside the results, according to the same publisher, Meta raised its outlook for capital spending and made clear it would keep investing heavily in artificial intelligence. Data centres, chips and specialist engineers are costs that arrive first. Revenue from them, if it comes, arrives later.

There is a detail in how such spending reaches the income statement. Money spent on servers and buildings is not charged as a cost all at once. It is spread over their useful life as depreciation. That means today's building programme will keep weighing on operating profit for years, whatever happens to revenue. The quarterly summary does not break operating costs down by line, so it does not show how much of the squeeze is depreciation and how much is people or other spending. The figure that would settle it is depreciation and research and development costs set against the same quarter a year earlier.

The balance sheet is carrying the build

The spending also shows on the other side of the accounts. Total assets grew from USD 304bn to USD 479bn within a year. Equity, the part of those assets that belongs to shareholders, rose 40.4% to USD 272bn. Yet the equity ratio, equity as a share of total assets, fell from 63.9% to 56.9%, down 7.0 pp. Assets grew faster than equity, so a larger share of the group is now financed by what it owes rather than by what its owners put in or left in.

Cash at the end of the quarter stood at USD 11.2bn, against USD 10.2bn a year earlier. For a group of this size that is a modest balance relative to what it is building.

The financing story continued after the quarter closed. Kurzy.cz reported that Meta announced a new class of shares without voting rights together with its results, and a few days later a plan to raise a large sum by selling shares to fund further AI development. For existing owners that has a plain consequence if the plan goes ahead: the build would be paid for partly with new shares, so each existing share would own a smaller slice of the company. A company with a margin of 40.1% rarely needs to sell shares. A company that has decided to build at this pace apparently does.

What to check in your own business

Most businesses are not building data centres, but many grow the same way: a new team hired ahead of demand, a second site, a new product line, a sales push into a new market. The pattern in the accounts is identical. Revenue rises, the margin slips, and the year looks better than it is because the top line is the first thing people read.

Start with the operating margin for each of your last eight quarters, not just the annual figure. Operating profit divided by revenue. If revenue is rising and the margin is falling, the growth is costing more than it brings in, at least for now.

Then split it. Work out the margin on the customers and products you added in the last year separately from the rest of the business. Allocate to them the costs that exist because of them: the new hires, the extra rent, the depreciation of equipment bought for them, any discounts given to win them. If the new business runs at a much thinner margin than the old, the squeeze lives there.

Finally, ask your finance team one direct question: which costs in this quarter's income statement belong to revenue we have not earned yet, and in which quarter do we expect that revenue to arrive? Write the answer down with a date. Spending ahead of revenue is a reasonable strategy only while someone is checking whether the revenue turns up on time.

Meta is a larger business than it was a year ago and a less profitable one per dollar, by choice rather than by accident. The net profit line flatters the quarter. The operating margin is the number that will say whether the build was worth it, and for now it says the bill came first.