Is Meta Becoming "Myanmar's Northern Region-Like"?

Author | Song Wanxin

Editor | Zhang Fan

On November 6th, technology giant Meta was exposed in a "fraudulent advertising" incident. An internal Meta document shows that about 10% of the company's revenue in 2024 came from fraudulent ads and ads for prohibited goods, which amounts to approximately $16.9 billion.

According to internal company estimates, its social platforms display 15 billion fraudulent ads to users every day.

However, a Meta spokesperson responded that the company's internal estimate that 10.1% of its revenue came from fraudulent and other prohibited ads was "rough and overly inclusive," and the actual figure was lower. But he refused to provide updated data. The spokesperson said that as of 2025, Meta had removed more than 134 million pieces of fraudulent ad content.

It is reported that the U.S. Securities and Exchange Commission (SEC) has launched an investigation into Meta's placement of financial fraud ads.

Just the previous week, Meta unexpectedly raised the lower limit of its annual total expense range by $2 billion, naturally aiming to continue expanding the construction of its AI data centers. Meanwhile, Meta's cash flow in the third quarter plummeted, leaving only 72.1 billion yuan.

(Source: Wind)

The dismal earnings data caused Meta's stock price to plunge by more than 11% on that day, and it fell by nearly 20% in the following five trading days. Its current market value is approximately $1.57 trillion.

01 Instagram Sells the Most Expensive Ads

In fact, fraudulent ads are hard to avoid on any internet platform. What matters is how the platform establishes a regulatory mechanism. But according to this internal company document, Meta's handling method is very lenient.

First, Meta's control system stipulates that only when the system determines that the probability of an ad placer being a fraudster exceeds 95% will it be banned.

That means most of the fraudulent ads users see on the platform have already been marked as "suspicious" by Meta but have not been removed.

What has drawn more criticism is that for these ad placers with a relatively high risk but not meeting the ban criteria, as a punishment, Meta charges these suspicious advertisers a higher advertising rate. This is equivalent to allowing potential fraud institutions to pay a "protection fee" in exchange for exposure on the Meta platform.

In addition, the document also shows that Meta's review policy clearly favors large customers. Small advertising customers will be banned if they are marked by the system at least eight times, while large customers can continue to place ads even if they have accumulated more than 500 violation records.

From the documents intercepted by the media, it can be seen that Meta has long been aware of the prevalence of fraud information on its platform and the proportion of fraud cases.

In May this year, Meta's security team estimated that about one-third of successful fraud cases in the United States involved Meta's platforms. The company's internal review also found that fraudsters generally believe that "it is easier to place fraudulent ads on Meta's platforms than on Google."

Looking deeper, behind the platform's regulatory method is the platform's will. The out-of-control explosion of fraudulent ads is hard to separate from the platform's interests and strategic orientation.

(Chart: 36Kr)

Meta changed its name to Meta in the third quarter of 2021 and announced its full commitment to the metaverse. After that, the company's stock price tumbled, dropping from around $350 to nearly $100 by the end of 2022. To appease investors, at the earnings conference in the fourth quarter of 2022, Meta announced that 2023 would be the "Year of Efficiency."

In addition to layoffs, Meta also began to increase the advertising openness of Reels, an advertising market that has not been fully explored under its umbrella. The ad loading rate of Reels was about 13% in 2023, and it has reached 20% this year.

Meanwhile, according to the earnings report, the average price of Meta's ads has been increasing quarter after quarter since 2023. According to the calculation of Guosen Securities, the average CPM across Meta's platforms increased by 9% year-on-year in 2024, benefiting from elections, sports events, and the application of AI tools.

This year, the CPM on the Meta platform has skyrocketed. eMarketer predicts that Instagram will become the first social platform with a CPM exceeding $10 in the fourth quarter. In the fourth quarter of 2022, Instagram had the lowest CPM among major social platforms, and its increase so far has reached 91%.

(Source: Emarketer)

02 The AI Money-Burning Continues

The rise in ad CPM is partly due to the application of AI.

Since the popularity of generative AI in 2023, the application of large models in ad placement has completely changed the previous way of ad buying. Using automated tools like Meta's Advantage+, advertisers no longer need to manually set targeting parameters but rely on the system to automatically match audiences and materials.

AI has significantly improved the ad conversion rate, and advertisers are therefore willing to pay a higher CPM.

However, AI algorithms only optimize "precise audience reach" and "efficient conversion," not "content authenticity," which has also led to a higher "conversion rate" for fraudulent ads.

While the fraudulent ads were exposed, Meta's latest capital expenditure data also shocked the market and caused its stock price to fluctuate.

Meta's management said in the last quarter's earnings conference call that the capital expenditure increase in 2026 will "significantly exceed" that in 2025. The latest guidance for the company's capital expenditure in 2025 is about $70 billion to $72 billion, which is also an upward adjustment from the previous range of $66 billion to $72 billion.

In 2023 and 2024, Meta's capital expenditures were $28.1 billion and $39.2 billion respectively. The year-on-year growth rate of capital expenditure this year reached a new high of 84%. These increasingly large capital expenditures are mainly concentrated on AI computing infrastructure.

Meta's aggressive approach to AI is also causing a talent crisis. The latest news is that Meta's chief AI scientist, Yann LeCun, plans to leave. Just last month, more than 600 people in the FAIR lab he is in charge of were laid off.

This veteran technologist, who joined Facebook in 2013, established the FAIR lab, which represents long-termism. But now, as Zuckerberg adjusts the AI strategy and loses patience, turning to short-term returns, the FAIR lab has been discarded.

Zuckerberg's AI anxiety stems from Meta's lack of money.

Data shows that Meta's free cash flow is no longer sufficient to cover AI investments. The latest third-quarter report shows that Meta's cash and cash equivalents are only 72.1 billion yuan, compared with more than 300 billion yuan last year.

Looking at the earnings reports of the past two years, thanks to AI, Meta's ad revenue recovered in 2023, reaching $133 billion, a 16% increase, and further increased by 22% to $162.3 billion in 2024. But obviously, the current growth rate of capital expenditure far exceeds the growth rate of ad revenue empowered by AI, which has raised market concerns.

Since Zuckerberg has no better way to generate revenue, he can only increase the ad loading volume. Meta is exploring advertising on WhatsApp and Threads.

This seems a bit ironic. First Apple, then Meta. After three years of the booming development of large models, Silicon Valley's tech giants still think that tapping into the advertising business is the most practical.

Follow for more information