ByteDance generated about $120 billion of revenue in the first half of 2026, up roughly 30% from a year earlier, while net profit fell by a single-digit percentage to about $20 billion, according to The Information, citing people with knowledge of the private company's results. On those rounded figures, first-half net margin was about 16.7%.
The margin movement is more revealing than the Meta comparison. If $120 billion represents roughly 30% year-on-year growth, the comparable first-half revenue base was about $92 billion. Because current profit is about $20 billion and was reported to have declined, prior-year profit must have been above $20 billion on the same headline scope. That puts the prior reported net margin above roughly 21.7% and implies margin compression of about five percentage points or more. The inputs are rounded, so the calculation is directional rather than an exact restatement of ByteDance accounts.
Meta reported $117.111 billion of revenue for the six months to June 30. ByteDance was therefore about $2.9 billion ahead on the two reported top-line figures. But Meta's number comes from a filed public-company statement with cash flow, capex and segment disclosures; ByteDance's comes from privately sourced reporting. Scale is comparable. Disclosure quality is not.
AI spending is now visible in the margin before it is visible in the accounts
The Information attributed ByteDance's profit decline to heavier spending on artificial intelligence. That is consistent with the strategic transition across the largest internet platforms: recommendation, advertising and commerce systems are becoming more dependent on model development, accelerators and data-centre capacity.
For ByteDance, the problem is not whether the spending can support growth. Revenue is still expanding rapidly. The harder question is how much of the incremental revenue is being consumed by compute, depreciation and model-development expense before the company offers public-market investors a full cash-flow statement.
The private accounts do not disclose enough to separate TikTok, Douyin, e-commerce and other businesses, or to reconcile AI capital expenditure against operating expense and free cash flow. That prevents a clean comparison with Meta's investment cycle even when the revenue totals sit in the same range.
The first-half numbers settle one argument: ByteDance operates at Western-megacap revenue scale. They also expose the next one. A reported net margin near 17%, down by at least roughly five points on the rounded year-on-year figures, means the AI build-out is already showing up in earnings. The valuation case now depends less on proving scale than on showing how much of that margin can return after the current infrastructure cycle.