Meta gambles on AI: Spending $135 billion, is Zuckerberg in 2026 worth believing?

Strong cash flow backing, hundreds of billions more invested in AI, and the market seems to be applauding the “money spending” this time. Will this be different from the metaverse?

Article by: Frank, Mai Tong MSX

$135 billion, that’s the amount Meta (META.M) plans to invest by 2026.

The double beat on Q4 2025 performance and Q1 2026 guidance has eased many shareholders who were worried about falling behind, but at the same time, the full-year capital expenditure (CapEx) for 2026 is projected to reach $135 billion, nearly double last year’s, which naturally raises concerns about whether this is another aggressive gamble.

Surprisingly, the market seems to have chosen to buy in, with Meta’s stock price surging over 10% after hours and continuing to rise in the night session.

Meta stock data source: Yahoo Finance

The answer is hidden in this earnings report: at least at this stage, it shows the market that AI investments are no longer just a future vision but are already effectively improving the most core cash cow—advertising business. Wall Street is betting on a narrative reversal for Meta and is willing to pay for this massive investment plan.

Ultimately, “dare to spend, dare to go all-in” has always been Meta and Zuckerberg’s core approach. This also means that winning could be a major narrative reversal; losing, at least under the current financial structure, is unlikely to turn into a catastrophic out-of-control disaster.

1. Quick Earnings Review: “Double Beat” in Performance & Guidance

From the results, this is a report capable of changing market sentiment.

The key financial indicators for Q4 2025 nearly all exceeded expectations: revenue of $59.893 billion, up 24% YoY, surpassing the market expectation of $58.6 billion; net profit of $22.768 billion, up 9%; diluted EPS of $8.88, up 11%, higher than the expected $8.23.

It can be said that whether in revenue resilience or profit release pace, Meta delivered a solid and stable Q4 performance.

Looking at the full year, the growth logic remains valid: total revenue for 2025 is projected at $200.966 billion, up 22%; operating profit of $83.276 billion, up 20%, with key metrics still maintaining double-digit expansion.

The only “countertrend” figure is the full-year net profit of $60.458 billion, down 3%. However, this change is not due to deteriorating core business but mainly caused by one-time tax factors—affected by the “Big and Beautiful Act,” the company recognized approximately $16 billion in one-time non-cash income tax expenses.

Excluding this factor, the full-year net profit and EPS would still show considerable growth, explaining the apparent contradiction between the annual data and strong quarterly performance.

Source: Meta

Meanwhile, operational metrics also show typical “volume and price increase” features:

• Daily active users (DAU) of family apps reached 3.58 billion, up 7% YoY, in line with market expectations;
• Ad impressions increased by 18% YoY; the average price per ad increased by 6% YoY;
• Average revenue per user (ARPU) was $16.73, up 16% YoY;

These data collectively point to a conclusion: Meta’s advertising engine not only has not slowed down but continues to evolve in efficiency and monetization capability.

Additionally, what further boosts market sentiment is not just the already realized strong performance but also the management’s optimistic guidance for the future: Meta expects Q1 2026 revenue to reach between $53.5 billion and $56.5 billion, representing a YoY growth of 26%–34%, significantly higher than the previous market expectation of about 21%. This pricing implies that management believes Reels’ high prosperity will continue, and the commercialization of Threads is progressing better than market’s cautious expectations.

With the advertising foundation stable, this guidance also directly reinforces market confidence that AI-driven improvements in ad efficiency will be sustainable.

Details of Reality Labs’ losses over the past five years

Of course, it’s worth noting that “metaverse” remains Meta’s bleeding point. Its metaverse division, Reality Labs, recorded an operating loss of $6.02 billion in Q4, an increase of 21% YoY, with revenue of $955 million, up 13% YoY. Since the end of 2020, this division’s cumulative operating loss has approached $80 billion.

But unlike before, Reality Labs’ role in this earnings report is no longer a core variable influencing the overall narrative; it is gradually being marginalized.

2. Solid Social Foundation, Deepening AI “Moat”

At least at the core business level, AI has indeed started to create tangible value for Meta (META.M).

To some extent, unlike Google (GOOGL.M) or Microsoft (MSFT.M), Meta is currently the most direct and report-verified “AI investment directly feeding main cash flow.”

This is first reflected in the systematic improvement in advertising efficiency, thanks to AI directly embedded into recommendation and ad delivery systems, leading to a 6% YoY increase in the average price per ad and an 18% increase in impressions in Q4. Management has repeatedly emphasized that upgrades to AI recommendation algorithms and delivery systems significantly enhance ad conversion rates and efficiency.

For example, Instagram Reels in the US saw over 30% YoY growth in watch time, becoming a core engine driving ad inventory and monetization.

Second is the accelerated commercialization of WhatsApp, Meta plans to fully introduce ads into WhatsApp Stories this year, which is seen as a potential new billion-dollar revenue stream, and a key step in expanding AI recommendation and ad systems into more traffic scenarios.

Overall, despite ongoing external competition from TikTok and others, Meta’s social foundation remains firm. By deeply embedding AI into recommendation and ad systems, it further deepens its moat.

Source: Meta

Looking back over the past year, Meta’s actions in AI have been quite aggressive—from spending hundreds of millions to acquire Scale AI shares, hiring Alexandr Wang to lead the “Superintelligence Laboratory (MSL),” to continuously recruiting top talent, restructuring AI teams, acquiring Manus for billions, launching Meta Compute, and planning to build tens of GW-scale computing and power infrastructure within this decade…

These series of actions evoke a familiar script: aggressive investment, grand narrative, long return cycle. In other words, we are once again seeing “Zuckerberg of the metaverse era.”

But unlike the metaverse period, this time management has provided clear bottom-line expectations: even with significantly increased infrastructure investment, operating profit in 2026 will still be higher than in 2025, and the cost growth path for the huge investments in 2026 is highly transparent, mainly focused on computing power, depreciation, third-party cloud services, and high-end technical talent.

In short, within Meta’s strategic framework, AI is not just a future betting narrative but a practical tool that continuously improves core cash flow. Its logic is simple: once AI is deeply embedded in recommendation and ad systems, even marginal improvements—such as getting 3.6 billion users to stay a few more seconds daily or increasing ad conversion rates by 1%— can be rapidly amplified into significant, repeatable cash flow increases given Meta’s current traffic scale and ad base.

Under this high-leverage structure, the efficiency gains from AI are real and tangible, effectively offsetting or even covering the $135 billion annual CapEx. In other words, Wall Street no longer fears Meta’s spending because it has already seen the tangible benefits AI can bring.

Interestingly, from a macro perspective, in this AI arms race in Silicon Valley, besides the mainstream path of exporting computing power, models, and tools globally—selling shovels and tools—another approach is Meta’s model—internalizing AI into its own business system, directly amplifying existing traffic and monetization engines.

It is this mode—focusing on improving internal monetization efficiency rather than selling new products externally—that makes Meta’s AI investment path distinctly different from other tech giants centered on large models or cloud services. This also leads the market to reevaluate Meta’s valuation foundation:

AI here is not a distant story waiting for realization but a real variable that can continuously and quantifiably feed back into main cash flow through advertising systems.

Perhaps this is the fundamental reason why the market is willing to reprice Meta.

3. Going All-In, an Unwinnable War?

“Superintelligence” has become one of the most frequently mentioned keywords by Zuckerberg and Meta’s management.

During this earnings call, Zuckerberg did not hide his ambitions: “I look forward to advancing personal superintelligence for global users,” which also marks a long-term strategic focus involving talent, computing power, and infrastructure.

From the capital expenditure perspective, as mentioned above, Meta has begun a full-blown all-in gamble: in 2026, total operating expenses will reach $162–169 billion, up 37%–44% YoY, significantly above the previous market expectation of about $150–160 billion.

At the same time, Meta is signaling “trade-offs” through actions: this month, media reports disclosed plans to cut about 10% of Reality Labs staff, involving around 1,500 people, further shrinking metaverse-related operations and reallocating resources to AI and core businesses.

More strategically, Meta is re-evaluating its computing power and infrastructure investments. On January 12, Zuckerberg personally posted about launching a new top-tier 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 US before 2028.

However, Meta’s CFO Susan Li later clarified that this figure does not solely cover AI server procurement but also includes data center construction, computing and power infrastructure, and additional costs for new staff and supporting expenses needed for US operations.

Objectively speaking, in terms of talent density, computing scale, and infrastructure strength, Meta’s investments in AI already rival or even surpass some major competitors in certain dimensions.

Of course, this path is inherently a double-edged sword. If revenue growth, ad efficiency, or new model progress cannot sustain cost expansion, market tolerance will quickly decline, and valuation and profit expectations could face backlash.

In other words, this is not a trial-and-error experiment but a strategic war that, once launched, is hard to turn back.

Final Words

As early as September 2025, in a blog episode, Zuckerberg openly said that if billions of dollars are ultimately wasted, it would be very unfortunate. But on the other hand, falling behind in the AI wave could pose even greater risks to Meta.

“For Meta, the real risk is not whether the investments are too aggressive but whether it hesitates at critical moments,” this statement, in today’s context, can almost be seen as a footnote to all of Meta’s strategic moves over the past year.

Of course, history is not easily forgotten. In the last metaverse narrative, Zuckerberg also chose to bet early and push forward with full force, but the final outcome did not meet the market’s initial expectations.

The difference this time is that Meta now controls the densest, most monetizable user traffic entry globally; and AI is reshaping the connection efficiency between people, content, and commerce in unprecedented ways.

Whether the $135 billion is a historic strategic sprint or another costly lesson, the answer still awaits time to reveal.