Meta has set aside up to $145 billion in capital expenditures for 2026 to back its AI and core businesses. Yet second-quarter free cash flow came to only $784 million.
Why has cash gotten this thin at a company whose quarterly revenue has grown to $60.8 billion? The answer is AI.
Meta is using AI to lift ad performance, jumping into the chatbot race with a new model, and putting AI into glasses to reach into people's daily lives. It's pouring money into all of it.
It will also announce third-quarter results after the US market close on October 21. If we first dissect the numbers and products disclosed so far, we can see an outline of where Meta's AI investment is earning money and where it's still spending it.
Has the New AI Model Reached Claude Territory?
Meta unveiled Muse Wonder and Muse Wonder Full on October 12. According to Meta, these models target multi-step tasks beyond chatbots that just keep a conversation going, such as handling an insurance claim, reading a lease, and building an app.
It also put forward Active Memory, which carries over important information from past conversations, and Active Focus, which tries a different approach when a task gets stuck.
Wonder is available through the Meta Model API and Muse Code, and the more powerful Wonder Full has begun rolling out to Muse Code subscribers.
On the independent evaluator Artificial Analysis's intelligence index, Wonder Full scores 56.
By the latest settings on the same evaluation table, Claude Opus 5.5 (Max) shows 58, Claude Sonnet 5.5 (Max) 56, OpenAI's GPT-6.1 Sol (Max) 52, and Google's Gemini 4 Argon (High) 53.
✦ Artificial Analysis Intelligence Index
Scores by model (evaluation setting in parentheses), as of Oct 19, 2026
At least within this evaluation table, Meta has stepped onto the same stage as the top-tier models. Of course, one score doesn't decide the win or loss of an entire product.
ChatGPT and Claude aren't just the names of single models; they're also services that bundle multiple models and features.
What Meta has to prove isn't only benchmark scores, but whether people use it every day, stay for a long time, and pay for paid features or the API.
Meta Is Already Making Money from AI
Meta's most realistic AI monetization comes from its advertising business. Facebook and Instagram have used AI to predict who to show an ad to and which ads will lead to clicks or purchases.
Meta said that in the second quarter, ad impressions rose 14% from a year earlier, and the average price per ad rose 12%. It's as if the effect of showing more ads and the effect of each ad becoming more valuable appeared at the same time.
Management also released results from experiments that added a large language model to ad recommendations. They explained that in early Instagram experiments app event conversions rose 1%, while on Facebook ad clicks rose 8.3% and conversions 15.7%.
They didn't disclose the total revenue contribution of AI taken on its own, but basic evidence has emerged that AI inside the ad system improves actual sales results.
✦ Ad AI Metrics
Results Meta disclosed (including management remarks on the earnings call)
It also said that the annualized revenue run rate of Advantage+, its AI-automated ad tool, topped $75 billion.
Annualized revenue is a figure converted by assuming the recent pace of revenue continues for a year, but in any case, it's clear there's a path where money flows when AI is attached to the ad engine Meta runs.
So Why Has Cash Flow Gotten So Thin?
The 2026 second-quarter results show growth and strain at the same time.
✦ Meta's 2026 Second-Quarter Results
Key financial metrics
The company presented a 2026 capital expenditure outlook of $130 billion to $145 billion. That money isn't spent on AI alone. The second quarter also reflected $2.4 billion in legal costs and $1.18 billion in severance costs from restructuring.
Still, with revenue up 28% while costs and expenses rose 55% and operating profit fell, it's natural for investors to feel uneasy.
In particular, the number investors view uneasily is free cash flow. If the cash left after capital expenditures is $784 million, a rise in model performance alone makes it hard to prove the investment is paying back.
Ad improvements need to turn into profit and cash, new AI products need to create separate revenue, and that pace needs to catch up with the growth in infrastructure spending. Investors doubt whether that's possible.
Are AI Glasses Meta's Next Ad Billboard?
AI glasses seem to be moving past lab demos into real product demand. EssilorLuxottica, which partners with Meta, announced that its AI glasses line sold more than 7 million units in 2025.
It also said AI glasses revenue in the second quarter of 2026 nearly doubled from a year earlier. Both unit sales and revenue growth are signs that the product line is growing fast.
I hear AI glasses are giving schools a headache because of cheating, so it feels like AI glasses are slowly starting to go mainstream.
Meta is also widening its product choices. The Meta Adventurer announced in September starts at $249 and is set to go on sale October 23. Korea is also included among the new launch markets for Meta Glasses.
The appeal of glasses is that they can bring AI into daily life through a camera, microphone, and speaker without constantly staring at a screen the way you do with a phone.
✦ AI Glasses: Key Dates and the Investment Angle
By period and product
But there's an accounting trap here too. Meta's Reality Labs had second-quarter revenue of $431 million but an operating loss of $4.619 billion. This segment includes VR and AR hardware, software, and content together.
So you can't blame this entire loss on AI glasses, but how profitable the glasses are remains uncertain. A partner's strong sales show demand, but they don't tell us how much Meta earns per unit.
What the Stock Asks Is Cash-Generating Power, Not AI Model Scores
✦ Meta Platforms (META, Nasdaq) Closing Prices
Sep 28 to Oct 16, US Eastern Time close (USD)
On October 12, the day of Meta's new model announcement, the stock rose 3.38% from the previous trading day. But from the October 13 close of $762.50, it came down 6.97% to $709.33 on October 16.
Looking at the price movement from September 28 to October 16, it fell about 0.88%, from $715.62 to $709.33. In effect, the new model announcement had no real impact on the stock price.
There are three things to check in this earnings report. First, whether growth in ad impressions and ad prices continues. Second, whether the capital expenditure outlook rises even higher.
Third, whether new AI products like Muse and the glasses connect to actual paid revenue beyond growing the user base.
Meta plans to announce results after the US market close on October 21, so in Korean time you can see the numbers in the early hours of October 22.
Meta's AI investment is already producing some results in its advertising business. That's not empty talk; it's a change confirmed in ad impressions, prices, and ad recommendation experiments.
But getting a good score in the model race and having recovered enough of the AI investment are completely different matters. In fact, the daily average users of the Family of Apps that Meta disclosed was 3.6 billion as of June 2026.
That distribution network is an asset that can spread a new model fast, but the number of users itself doesn't mean paid revenue.
What matters for Meta going forward doesn't end with building a smart model.
It lies in deploying the model inside apps used by billions of people, growing ad efficiency and paid product revenue, and increasing earnings faster than the cash it pours into data centers. (Most AI-related companies are probably like that.)
AI glasses are the test of that strategy expanding beyond the screen, and the third-quarter results are the first chance to check that cost table again.
This isn't investment advice. You know that, right? Let's each protect our own hard-earned money.
That's all for this post. See you next time.
- Samsung 70,000 Won, SK hynix 200,000 Won? The Shock Targets
- Broadcom CEO Likens Some AI Investors to Casino Gamblers
- Bitcoin Firm Strategy (MSTR) at Risk of MSCI Index Removal
🌐 Korean version: 챗GPT, 클로드와 대등한 메타의 AI모델 Muse Wonder 출시! 그런데 메타는 과연 AI로 돈을 벌고 있을까?
No comments:
Post a Comment