When many people think about Meta, their first reaction is still:
Facebook, Instagram, and WhatsApp.
The natural assumption is that the company mainly operates through:
Users consuming content → Platforms displaying advertisements → Meta generating advertising revenue.
This understanding is not wrong.
However, by September 2026, continuing to view META simply as:
“A social media advertising company”
would underestimate the transformation currently taking place.
Because what Meta is truly trying to build now is:
Using the massive cash flow generated by its advertising business to develop AI models, AI products, and AI computing infrastructure.
Therefore, when analyzing META on September 2, the key question is no longer:
“Can Facebook continue growing?”
The more important question is:
Can Meta transform its ecosystem of more than three billion users into one of the largest AI distribution channels in the world?
The Latest Earnings Report Shows Revenue Remains Extremely Strong
On July 29, Meta released its second-quarter 2026 earnings results.
Quarterly revenue reached:
$60.801 billion.
Compared with:
$47.516 billion in the same period last year.
Representing:
28% year-over-year growth.
Within the Family of Apps segment:
Ad impressions increased:
14% year over year.
Average advertising price increased:
12% year over year.
Family Daily Active People, representing the number of daily users across Meta’s applications, reached:
3.6 billion people.
Growing:
3% year over year.
The key point is not only:
Revenue increased 28%.
More importantly, both major drivers of Meta’s advertising business continued operating:
More advertisements being delivered.
And:
Higher value generated from each advertisement.
This indicates that as of Q2, Meta’s core advertising business still maintained very strong monetization capability.
AI Is No Longer Just Meta’s Future Story — It Is Already Improving Today’s Advertising Business
In the past, when people discussed Meta’s AI investments, many viewed them as:
“The company is spending heavily for the future.”
However, the situation has changed.
AI is now improving Meta’s most profitable business:
Advertising.
Through AI, Meta can better understand:
What content users are interested in.
What products they may purchase.
When users are more likely to engage with advertisements.
Which advertisements should be shown to which users.
At the same time, advertisers are increasingly using Meta’s AI tools to automatically generate and optimize advertising materials.
Therefore, Meta does not necessarily need to wait until it launches a paid AI assistant to generate revenue from AI.
Instead, AI can first improve the existing business model through:
Better recommendation efficiency
↓
Longer user engagement time
↓
Higher advertising conversion efficiency
↓
Greater advertising value
This is why Meta stated in its Q2 earnings report that AI is accelerating its core business while also creating the foundation for next-generation products and new business opportunities.
Meta’s Biggest Advantage Is Not Only Its AI Models — It Is Its 3.6 Billion Users
The AI industry is becoming increasingly competitive.
Major players include:
OpenAI.
Google.
Anthropic.
xAI.
And many other AI model companies.
Therefore, simply asking:
“Which company has the top-ranked AI model today?”
may not be the most important question when analyzing Meta.
Meta’s truly difficult-to-replicate advantage is:
Distribution capability.
As of June 2026, Meta’s applications had an average daily user base of:
3.6 billion people.
Including:
Facebook.
Instagram.
WhatsApp.
Messenger.
Threads.
These products already exist on billions of people’s devices.
This means that if Meta introduces new AI capabilities, it does not necessarily need to spend years acquiring users from scratch.
It can directly integrate AI into existing products.
This creates a major difference.
Other AI companies usually need to solve:
“Where do the users come from?”
Meta can instead ask:
“How can AI be integrated into an ecosystem that already serves 3.6 billion users?”
Examples include:
AI-powered content recommendations on Instagram.
AI assistants inside WhatsApp.
Search and recommendation improvements on Facebook.
Generative AI tools for advertisers.
And potentially:
AI-powered glasses in the future.
Meta already owns the user entry points behind these experiences.
Therefore, one of Meta’s most important AI advantages may not simply be:
The model itself.
Instead, it may be:
Model + Users + Data + Distribution.
Together, these four elements represent what makes Meta especially worth studying in the AI era.
However, Building This AI Future Requires Extremely Large Investments
This is one of the most important issues when analyzing META on September 2.
In Q2 2026, Meta’s capital expenditures reached:
$31.08 billion.
The company also raised its full-year 2026 capital expenditure outlook to:
$130 billion–$145 billion.
Compared with its original annual guidance of:
$115 billion–$135 billion,
the new range represented a meaningful increase.
What does investing more than:
$130 billion per year
actually mean?
This is no longer simply:
“Buying more GPUs.”
Meta is building at massive scale:
Data centers.
AI servers.
Networking infrastructure.
Power systems.
Cooling systems.
And complete computing infrastructure.
Meta is gradually adding characteristics of:
An AI infrastructure-intensive company
to what was previously an internet software business.
The July 28 BlackRock Partnership Clearly Demonstrated This Direction
One day before releasing Q2 earnings, on July 28, Meta announced a partnership with BlackRock to develop:
The El Paso AI Data Center Campus.
The planned campus includes:
1GW of computing capacity.
The total development cost is approximately:
$14 billion.
Funds managed by BlackRock will own:
80%
of the joint venture.
Meta will own:
20%.
The campus is expected to begin coming online gradually from:
2028.
This development is highly important.
Because it reveals a new approach Meta is exploring for AI infrastructure:
Not every data center needs to be entirely funded with Meta’s own capital.
Why Is Meta Bringing in BlackRock?
Because AI infrastructure has become extremely expensive.
If Meta eventually requires:
5GW.
10GW.
Or even larger computing capacity,
building everything independently would consume an enormous amount of capital.
Therefore, Meta is beginning to explore a model where:
Meta provides:
AI infrastructure expertise and operational capabilities
while:
External capital provides part of the funding.
The El Paso project is a clear example.
Meta manages development and will use the future computing capacity.
BlackRock provides significant capital resources.
This allows Meta to secure the computing capacity it needs while reducing some direct capital burden.
The broader implication is that Meta’s AI strategy has entered a new stage.
The question is no longer:
“How many GPUs should we buy?”
The question has become:
“How do we finance, power, and build AI factories worth tens of billions or even hundreds of billions of dollars?”
Meta Is Even Introducing a New Concept: Meta Compute
On August 6, Meta specifically explained why the company is building its own AI data center infrastructure.
Meta stated that its infrastructure supports:
Instagram.
Facebook.
WhatsApp.
Threads.
Meta AI.
And future AI products.
The underlying logic is easy to understand.
If AI is only a small feature within an application, renting cloud computing capacity may be sufficient.
However, if the future scenario becomes:
Billions of users interacting with AI every day,
the scale of computing demand changes completely.
At that level, companies that control:
Data centers.
Power infrastructure.
Networking.
And AI chip deployment capabilities.
may have a significant competitive advantage.
This is why Meta is gradually turning AI infrastructure into one of its core capabilities.
AI Glasses Could Become Another Extremely Important Strategic Opportunity
What makes Meta particularly interesting is that the company does not only want AI to exist inside smartphones.
It is also investing in:
AI Glasses.
On June 23, Meta and EssilorLuxottica announced a new Meta Glasses product series.
The products include:
26 styles
with a starting price of:
$299.
They also include Meta’s own AI capabilities.
Why are AI glasses worth paying attention to?
Because in the smartphone era, interacting with AI still usually requires:
Take out your phone.
Open an application.
Type a question.
But glasses are fundamentally different.
Glasses can:
See what you see.
Hear what you hear.
Understand your surrounding environment.
And communicate with you directly through voice.
If AI truly becomes an:
Always-available personal assistant,
glasses may eventually become a more natural interface than smartphones.
This Is Why Meta Has Continued Investing in Hardware
When many investors look at Reality Labs, they mainly see:
Losses.
That concern is reasonable.
Reality Labs has consumed significant amounts of capital over the years.
The adoption speed of the VR market has also fallen far short of the most optimistic early expectations.
However, Meta’s core objective has remained consistent:
Owning an entry point into the next generation of computing platforms.
During the smartphone era, one of Meta’s biggest disadvantages was:
It did not own the smartphone operating system.
Facebook, Instagram, and WhatsApp all depend on:
Apple iOS
or:
Google Android.
In other words:
Meta owns the users,
but it does not own the operating system running on the devices in users’ hands.
Therefore, in the AI era, the company does not want to repeat the same situation.
AI glasses represent one of the next-generation interfaces Meta is attempting to secure.
If AI Glasses Succeed, Meta Could Create a Unique AI Ecosystem
Meta already owns:
Facebook + Instagram + WhatsApp + Threads
↓
More than:
3.6 billion daily users
↓
Advertising business generates significant cash flow
↓
Cash is invested into:
AI models + GPUs + Data Centers
↓
AI is integrated into:
Social applications + Advertising systems + Smart glasses
↓
AI improves user experience and advertising efficiency
↓
The ecosystem generates more cash flow.
This is the AI cycle Meta is attempting to build.
However, Q2 Earnings Also Revealed a Clear Challenge: Profitability Is Being Pressured by AI Investment
Q2 revenue increased:
28% year over year.
However, operating income declined:
8% year over year.
Operating income reached:
$18.775 billion.
Operating margin declined from:
43% in the same period last year
to:
31%.
Net income decreased from:
$18.337 billion
to:
$15.848 billion,
representing:
14% year-over-year decline.
Why?
Because company spending is increasing even faster.
Q2 total costs and expenses reached:
$42.026 billion.
Increasing:
55% year over year.
The increase included:
AI infrastructure investments,
Employee costs,
Third-party AI token expenses,
And legal expenses.
Meta stated in its 10-Q filing that major contributors to cost growth included:
Data centers and technology infrastructure,
Third-party cloud services,
And third-party AI token costs.
Therefore, Meta’s biggest debate today is becoming increasingly clear:
AI is already helping revenue growth.
But at the same time:
Meta is spending money on AI at an unprecedented pace.
The Change in Free Cash Flow Is Even More Noticeable
In Q2, Meta generated:
$31.86 billion in operating cash flow.
At first glance, this remains extremely strong.
However, because capital expenditures reached:
$31.08 billion,
Q2 free cash flow declined to only:
$784 million.
As of the end of June, Meta held:
$90.26 billion
in cash, cash equivalents, and marketable securities.
At the same time, long-term debt reached:
$83.66 billion.
These figures are important.
Meta is not a company without financial resources.
On the contrary, its core business continues generating substantial cash.
The challenge is:
AI is consuming an increasing amount of that cash.
Therefore, the key question investors need to verify in the future is:
How much additional profit can today’s more than $100 billion AI investment ultimately generate?
This Is the Most Important Financial Question Facing META
Assume Meta spends an additional:
$100 billion per year
building AI infrastructure.
If the result is only:
A slightly improved Instagram recommendation system,
then the investment would clearly be difficult to justify.
However, if these AI investments eventually create:
Higher advertising conversion rates.
Longer user engagement time.
New AI monetization opportunities.
Enterprise AI tools.
AI glasses.
New personal AI interfaces.
then today’s extremely large capital expenditures could become the foundation for a new growth cycle.
Therefore, analyzing META should not simply conclude:
“Capital expenditures are too high, so the investment is negative.”
Nor should it conclude:
“AI is powerful, so any amount of spending is justified.”
The key factor is:
Return on investment.
Another Risk That Cannot Be Ignored: Regulation and Legal Challenges
On August 26, Meta announced an agreement with U.S. state attorneys general from both political parties.
The agreement involved approximately:
$18 billion in payments.
The payments will be made over:
10 years,
with part of the amount depending on specific conditions.
Meta expects this agreement to result in approximately:
$10 billion in legal expenses
recognized in Q3 2026.
Importantly, this expense was not included in the full-year expense guidance provided in the July 29 earnings report.
This development was highly relevant when analyzing META on September 2.
Because it highlighted another long-term challenge beyond AI investment:
Regulatory risk.
Why Regulatory Pressure Cannot Be Ignored
Meta serves billions of users worldwide.
Its platforms are connected with issues including:
Privacy.
Advertising practices.
Youth protection.
Content recommendation.
AI data usage.
Competition policy.
Any of these areas can create regulatory pressure.
As companies become larger, regulatory costs often become increasingly significant.
The August 26 agreement was only one specific example.
It demonstrated that legal risks can eventually become:
Tens of billions or even hundreds of billions of dollars in real financial costs.
Another Risk Is That AI Competition Is Far From Over
Meta has users and distribution.
But its competitors are also extremely strong.
Google has:
Search + Android + YouTube + Gemini.
OpenAI has:
A strong AI brand and growing user base.
Other companies are also rapidly advancing:
AI Agents,
Enterprise AI,
And consumer AI products.
Therefore, Meta cannot rely only on:
“We have 3.6 billion users.”
The company must prove:
Those users will continue actively using Meta’s AI products.
Having distribution does not automatically guarantee user adoption.
This is one of the key factors that still needs to be validated.
On September 2, META Was at a Particularly Interesting Stage
On one side:
The core advertising business remained strong.
Q2 revenue increased:
28% year over year.
Advertising impressions increased:
14%.
Average advertising price increased:
12%.
Meta’s applications reached:
3.6 billion daily users.
On the other side:
AI investment was becoming extremely large.
Full-year capital expenditure was expected to reach:
$130 billion–$145 billion.
At the same time:
Operating profit declined.
Free cash flow was significantly pressured by capital spending.
Regulatory costs continued to exist.
This means META was no longer simply a:
“Advertising growth story.”
It was becoming a:
“AI investment return story.”
However, Meta Has an Advantage That Many AI Companies Cannot Easily Replicate
Meta is not starting from zero to find a business model.
The company already has:
3.6 billion daily users.
It already generates:
More than $60 billion in quarterly revenue.
It already operates:
A mature global advertising system.
It already owns major distribution platforms including:
Instagram, Facebook, WhatsApp, and Threads.
This creates a major difference between Meta and many AI startups.
Many AI companies follow a path like:
Build AI → Find users → Develop monetization.
Meta’s path is different:
Already have users → Already have cash flow → Integrate AI into the ecosystem.
This sequence is extremely important.
Therefore, Studying META on September 2 Was Not Simply About Whether Facebook Could Continue Growing
The more important question was:
Can Meta transform one of the world’s largest social networks into one of the world’s largest consumer AI distribution networks?
If AI only improves advertising recommendations slightly, then the value created may simply be:
Improving the efficiency of the existing business.
However, if AI eventually becomes deeply integrated into:
WhatsApp,
Instagram,
Facebook,
Enterprise services,
and AI glasses,
then Meta could create an entirely new and much larger monetization curve.
At the same time, the challenge is equally clear:
The company is investing:
$130 billion–$145 billion in annual capital expenditures
to build this future.
Therefore, the most important question on September 2 was not:
How large does the AI opportunity sound?
The real question was:
When will this unprecedented AI investment begin translating into unprecedented profits?
If the answer gradually becomes clear, Meta would no longer simply be:
A company that sells advertisements through Facebook and Instagram.
Instead, it could become:
A platform company that combines user distribution, advertising infrastructure, AI models, computing infrastructure, and next-generation AI devices.
This was the core transformation worth monitoring for META as of:
September 2, 2026.