What is Meta’s China exposure?

Following the announcement yesterday that China’s National Development and Reform Commission (NDRC) has ordered Meta to reverse its acquisition of Manus, the WSJ reports that Meta may be preparing to unwind the transaction:

Beijing has handed the two companies a preliminary deadline of several weeks to unwind the deal and fully restore Manus’s Chinese assets to their original state, some of the people said. This includes stripping any previously transferred data or technology from Meta. Beijing has also considered imposing penalties on Manus and Meta if the deal couldn’t be fully rescinded, these people said …

Chinese authorities believe they have the authority to demand the deal be unwound because Beijing Butterfly Effect Technology remains a Chinese company, The Wall Street Journal previously reported. Chinese law dictates that any foreign investments that may carry a national-security risk could be subject to review by authorities.

It remains to be seen what fully abandoning this acquisition might entail; as I noted in my coverage, Manus is already meaningfully integrated into the Meta Ads Manager, and China is reportedly demanding that any data transferred from Manus to Meta be expunged from Meta’s systems. My belief is that this acquisition is being used to gain leverage in a potential trade agreement, given President Trump’s planned visit to China next month as part of a trade delegation. Should the Beijing summit proceed, it will be the first time in nearly a decade that a sitting US President has visited the country (the last visit was by Trump during his first term, in November 2017).

Motivation aside, it’s interesting to consider the potential ramifications if Meta doesn’t or can’t comply with China’s demands. Meta has material exposure to China: substantially all of its VR and mixed-reality hardware is manufactured in the country, and Chinese advertisers represent a meaningful portion of its revenue. While Meta doesn’t break out its advertising revenue by country of origin, it does provide disaggregated regional revenue by its customers’ addresses. While these revenue figures span all revenue sources (advertising, Reality Labs, and “Other,” which was larger than the Reality Labs segment in 2025), since advertising represented 97.6% of Meta’s revenue in 2025, I’ll treat these figures as advertising revenue for the purpose of this piece.

In 2025, the Asia-Pacific reporting region, as measured by customer address, contributed 26.8% of Meta’s total revenue. Recent reporting from Reuters states that Meta generated roughly $18BN in revenue from Chinese advertisers in 2024, representing 10.9% of the company’s total revenue that year and 40% of its revenue in APAC. Because the distribution of revenue by customer geography didn’t change meaningfully from 2024 to 2025, I used the 2024 percentages as a proxy to estimate absolute China revenue of $21.8BN in 2025.

It’s tempting to view this revenue estimate as being “at risk” for Meta if Chinese regulators find it in violation of their order to reverse the acquisition of Manus. But as I explain in Digital advertising and tariffs: what can COVID explain?, given the second-price nature of many social media advertising auctions, if one advertiser leaves the ecosystem altogether, the platform loses revenue roughly equivalent to the incremental pricing pressure that advertiser exerted, not its ad spend. From that piece:

But it’s important to consider the auction dynamics at play when large advertisers pull back in spend. Because social media platforms tend to operate modified Vickrey (second-price) auctions, when one advertiser drops out of the auction completely, the platform doesn’t lose an amount of revenue equivalent to their ad spend — they lose roughly the difference between the old second- and third-place bids. This is because, in a second-price auction, the advertiser with the highest bid pays slightly more than the second-highest bid …

In cases where auctions are dense, the deltas between the top three bids may not be very large — meaning the second-highest bidder may win more bids when the highest bidder departs but may not pay much less than they were previously paying. This is good for the platform — it won’t lose much money, since a ready bidder is willing to pay almost as much as the bidder the platform lost. But this doesn’t leave room for material expansion of bids (and budget) by the new highest bidder.

This concept is explored in depth in Hal Varian’s seminal 2009 paper, Online Ad Auctions: in a Vickrey-Clarke-Groves (VCG) auction mechanism, an advertiser pays the cost imposed on other advertisers by its presence in an ad impression. Given that, by definition, Chinese advertisers are only reaching non-Chinese audiences, as I unpack in detail in Temu’s impact on social media advertising prices, then whatever revenue is generated by them is by its nature sourced globally and available to other advertisers.

But there is a wrinkle in this assessment: a large proportion of advertising revenue from Chinese advertisers may be for scams, as alleged in the aforementioned Reuters reporting. Per that reporting, an internal Meta memo estimated that a significant portion of Chinese advertising revenue in 2024 ($3BN) was generated from “ads for scams, illegal gambling, pornography and other banned content.” That same memo states that the percentage of these scam ads declined from 19% of overall Chinese advertising revenue to 9% over the course of 2024, but rose to 16% by mid-2025; that same 16% proportion would amount to roughly $3.5BN if applied to my 2025 revenue estimate.

This portion of revenue, should Chinese advertisers be excluded from Meta’s properties, likely would be at risk because it represents lower-demand, largely untargeted inventory: the impressions that aren’t being competed for by other advertisers. This reflects a form of adverse selection: scam and policy-violating ads tend to concentrate in segments of supply where mainstream advertiser demand is weaker.

This could, conceivably, be eliminated from Meta’s bottom line if Chinese advertisers are barred from the service. Good riddance. The revenue is immaterial on a relative basis; these ads degrade the product experience for users that are exposed to them; and the reputational damage likely destroys far more enterprise value than the revenue contributes. If Meta is prevented from serving Chinese advertisers, the revenue at risk is merely the incremental bid pressure they applied, not the total value of their spend. But Meta may also gain the opportunity to rid its ecosystem of a large volume of predatory ads that create logistical and reputational headaches. What appears to be $20BN+ of revenue exposure is, in reality, mostly pricing pressure interspersed with a liability that is challenging to quantify.