Written by: Frank, Maiton MSX

$135 billion, that's the amount Meta (META.M ) plans to spend in 2026.

The better-than-expected results in Q4 2025 and guidance in Q1 2026 have given many shareholders who were struggling with doubts about "falling behind" a little relief. However, at the same time, the total capital expenditure (CapEx) for 2026 is expected to reach $135 billion, almost double that of last year, which makes it hard not to worry whether this is another aggressive gamble.

Surprisingly, the market seemed to buy into it, with Meta's stock price surging more than 10% in after-hours trading and continuing to rise in overnight trading.

Meta stock price data source: Yahoo Finance

The answer lies hidden in this financial report: at least for now, it shows the market that AI investment is not just a future vision, but has already made tangible improvements to the most core cash cow – the advertising business. Therefore, Wall Street is starting to bet on Meta’s narrative reversal and is willing to pay for this super investment plan.

Ultimately, "daring to spend money and go all in" has always been the defining characteristic of Meta and Zuckerberg. This also means that a win could be a huge narrative reversal; a loss, at least under the current financial structure, is unlikely to escalate into an out-of-control disaster.

I. Quick Financial Report Summary: Both Earnings and Guidance Exceeded Expectations

In terms of results, this is a financial report that is enough to change market sentiment.

In Q4 2025, almost all of the core financial metrics exceeded expectations: revenue of $59.893 billion, up 24% year-over-year, exceeding market expectations of $58.6 billion; net profit of $22.768 billion, up 9% year-over-year; and diluted earnings per share (EPS) of $8.88, up 11% year-over-year, exceeding market expectations of $8.23.

It can be said that Meta delivered a solid and stable Q4 performance, both in terms of the resilience of revenue growth and the pace of profit release.

Looking at the whole year, the growth logic remains the same: full-year revenue in 2025 is US$200.966 billion, up 22% year-on-year; operating profit is US$83.276 billion, up 20% year-on-year, with core indicators still maintaining double-digit growth.

The only exception was the full-year net profit of $60.458 billion, a 3% decrease year-on-year. However, this change was not due to a deterioration in the main business, but mainly due to a one-off tax factor – affected by the Big and Beautiful Act, the company recognized a one-off non-cash income tax expense of approximately $16 billion.

If this factor is excluded, the actual net profit and EPS for the whole year will still achieve considerable growth, which explains the apparent contradiction between the full-year data and the strong quarterly performance.

Source: Meta

At the same time, operational metrics also exhibited a typical "simultaneous increase in both volume and price":

• The number of daily active users (DAP) for the family of apps reached 3.58 billion, a year-on-year increase of 7%, in line with market expectations;
• Ad impressions increased by 18% year-on-year; average price per ad increased by 6% year-on-year.
• Average revenue per user (ARPU) was $16.73, a year-over-year increase of 16%;

All of this data points to the same conclusion: Meta's advertising engine has not only not slowed down, but has continued to evolve in terms of efficiency and monetization capabilities.

Furthermore, what truly stimulated the shift in market sentiment was not only the better-than-expected performance that had already materialized, but also the management's optimistic guidance for the future: According to Meta's forecast, revenue in the first quarter of 2026 will reach $53.5-56.5 billion, corresponding to a year-on-year growth of 26%-34%, significantly higher than the market's previous growth expectation of about 21%. This pricing implies that management believes that the high growth of Reels will continue, while the commercialization progress of Threads is better than the market's previous cautious expectations.

With the core of the advertising market stable, this guidance also directly strengthens market confidence in the sustainability of AI-driven improvements in advertising efficiency.

Details of Reality Labs' losses over the past five years

Of course, it's worth mentioning that the "metaverse" remains Meta's Achilles' heel. Its metaverse division, Reality Labs, recorded an operating loss of $6.02 billion in Q4, a 21% increase year-over-year, with revenue of $955 million, a 13% increase year-over-year. Since the end of 2020, the division's cumulative operating loss has approached $80 billion.

However, unlike in the past, Reality Labs' role in the current financial report is no longer a core variable that influences the company's overall narrative, and it is gradually being marginalized.

II. With a solid social foundation, AI further deepens the competitive advantage.

At least at the core business level, AI has indeed begun to create real value for Meta's (META.M ) commercialization.

In some respects, unlike Google ( GOOGL.M ) or Microsoft (MSFT.M ), Meta is currently the player whose "AI investment directly benefits its main operating cash flow," a fact that has been verified by financial reports.

Firstly, this is reflected in a systemic improvement in advertising efficiency, thanks to the direct embedding of AI into the recommendation system and ad delivery system. This resulted in Meta's average cost per ad increasing by 6% year-on-year in Q4, with impressions surging by 18%. Management has also repeatedly emphasized that the upgrades to the AI ​​recommendation algorithm and delivery system have significantly improved ad conversion rates and delivery efficiency.

Instagram Reels saw a more than 30% year-on-year increase in viewing time in the US market, becoming a core engine for driving ad inventory and monetization capabilities.

Secondly, there is the accelerated commercialization of WhatsApp. Meta plans to fully introduce advertising into WhatsApp Moments this year, which is seen as the company's next potential multi-billion dollar revenue growth point and a key step in expanding AI recommendation and advertising systems to more traffic scenarios.

Overall, despite continued external competition from TikTok and other giants, Meta's core social foundation has not weakened. Instead, it has further strengthened its competitive advantage by deeply embedding AI into its recommendation and advertising systems.

Source: Meta

Looking back over the past year, Meta's moves in the AI ​​field have been undeniably aggressive—from spending billions of dollars to acquire a stake in Scale AI and hiring Alexandr Wang to lead the "Super Intelligence Lab (MSL)," to continuously poaching talent with high salaries and restructuring its AI organizational structure, to spending billions to acquire Manus, and launching Meta Compute with plans to build tens of gigabytes of computing power and power infrastructure within the next decade...

This series of actions has reminded many of a familiar script: aggressive investment, grand narrative, and long payback period. In other words, we seem to be seeing "Zuckerberg in the metaverse era" again.

However, unlike the metaverse era, this time the management has given a clear bottom-line expectation, stating that even with a significant increase in infrastructure investment, the operating profit in 2026 will still be higher than in 2025, and the cost growth path of the huge investment in 2026 is highly transparent, mainly concentrated in computing power, depreciation, third-party cloud services and high-end technical talent.

In short, within Meta's strategic framework, AI is not merely a technological narrative betting on the future, but a real tool that is continuously improving its core cash flow. The logic is not complicated: when AI is deeply embedded in recommendation and advertising systems, even the smallest marginal improvements, such as getting 3.6 billion users to stay on the platform for a few dozen seconds longer each day, or increasing ad conversion rates by 1%, will be rapidly amplified into a considerable and repeatable increase in cash flow on top of Meta's current traffic scale and advertising base.

It is precisely under this high-leverage structure that the efficiency improvements brought by AI are effectively offsetting and even covering the annual capital expenditure of up to $135 billion. In other words, Wall Street is no longer afraid of Meta burning money, partly because it has seen the real money brought by AI.

Interestingly, from a broader perspective, in this AI arms race in Silicon Valley, besides the mainstream approach of busy exporting computing power, models, and tools to the world as "selling shovels and tools," another approach is the Meta model —internalizing AI into the heart of one's own business system, directly amplifying existing traffic and monetization engines.

It is precisely this model—which doesn't rely on selling new products but instead achieves returns by improving its own monetization efficiency—that clearly distinguishes Meta's AI investment path from the monetization logic of other large technology companies, which centers on large models or cloud services. Because of this, the market is beginning to re-examine Meta's pricing strategy.

AI here is not a distant story waiting to be realized, but a real variable that can be continuously and quantitatively fed back into the main cash flow through the advertising system.

This may be the fundamental reason why the market is willing to reprice Meta.

III. All-in betting with violence: a war that cannot be lost?

"Superintelligence" has become one of the most frequently used keywords by Zuckerberg and the Meta management team.

During the earnings call, Zuckerberg made no secret of his ambition: "I look forward to advancing personal superintelligence for users around the world," which has become a long-term Meta strategy encompassing talent, computing power, and infrastructure.

First, looking at the capital expenditure figures, as mentioned above, Meta has embarked on an all-in bet, with total operating expenses for 2026 expected to reach $162-169 billion, representing a year-on-year increase of 37%-44%, significantly higher than the market's previous expectation range of approximately $150-160 billion.

At the same time, Meta is also sending a “trade-off signal” to the market through its actions. Just this month, the media revealed that it plans to lay off about 10% of the employees at Reality Labs, involving about 1,500 people. This means that Metaverse-related businesses are being further compressed to make room for AI and core businesses.

More strategically significant is Meta's reclassification of computing power and infrastructure. On January 12, Zuckerberg personally posted that he had "launched a new top-level strategic project called Meta Compute." According to disclosed information, Meta plans to invest at least $600 billion in data centers and related infrastructure in the United States by 2028.

However, Meta's CFO, Susan Li, later clarified the figure, stating that the investment was not solely for the purchase of AI servers, but rather covered the construction of data centers, computing power and power infrastructure in the United States, as well as the additional staff and related costs required to support US business operations.

Objectively speaking, in terms of talent density, computing power scale, and infrastructure strength, Meta's investment in AI is no less than, and in some dimensions even surpasses, its main competitors.

Of course, this path is inherently a double-edged sword. Once revenue growth, advertising efficiency, or new model developments fail to keep pace with cost expansion, market tolerance will drop rapidly, and valuations and earnings expectations may face a backlash.

In other words, this is not an experiment that can be repeatedly tested and failed, but a strategic war that, once started, is difficult to turn back from.

In conclusion

Back in a blog post in September 2025, Zuckerberg stated that it would be very unfortunate to waste hundreds of billions of dollars, but on the other hand, the risk to Meta might be even greater if it fell behind in the AI ​​wave.

"For Meta, the real risk is not whether the investment is too aggressive, but whether it will hesitate at critical moments." In today's context, this statement can almost be regarded as a footnote to all of Meta's strategic moves over the past year.

Of course, history will not be easily forgotten. In the previous one-dimensional universe narrative, Zuckerberg also chose to bet early and push forward with all his might, but the final result did not meet the market's initial expectations.

The difference is that this time Meta has the world's densest and most commercially viable user traffic entry point; and AI is reshaping the efficiency of the connection between people and content, and between people and business in an unprecedented way.

As for the $135 billion, whether it is a historic strategic move or another costly lesson, only time will tell.