Meta's AI Buildout Pushes Capital Spending Toward $145 Billion

Meta is spending at a scale that is redefining the economics of the AI race

Meta's latest earnings report delivered two sharply different messages at once. Revenue rose 33% in the quarter, the company's fastest pace of growth since 2021. Yet the market reaction was negative, with shares falling more than 7%. The reason was not weak top-line performance. It was the sheer size of Meta's AI bill.

The company said its 2026 capital expenditures will be at least $10 billion higher than previously expected and could exceed $145 billion. Chief executive Mark Zuckerberg said most of the increase is tied to higher component costs, especially memory pricing. That detail matters because it links Meta's spending surge not just to internal ambition, but to a broader supply squeeze created by the data-center expansion underway across the AI industry.

An AI arms race with expensive components

The new capex range illustrates how quickly AI infrastructure has become one of the most capital-intensive bets in technology. Meta recorded $72 billion in capital expenditures last year. A path to more than $145 billion would represent a dramatic escalation in a single year. The company's explanation points to a market where advanced memory has become a choke point, raising costs not only for hyperscalers but across the wider electronics supply chain.

According to the source text, that shortage is feeding into a broader memory crisis that is affecting both AI companies and consumer devices such as laptops and smartphones. For Meta, the direct implication is straightforward: building frontier-scale AI systems now requires not only conviction, but the ability to absorb hardware inflation at extraordinary magnitude.

Zuckerberg is betting on a catch-up strategy

The spending jump also reflects Meta's strategic position. The company has been trying to close ground on rivals that have moved ahead in AI. Roughly ten months ago, Zuckerberg publicly acknowledged the need for a major catch-up effort and began committing large sums to research, development, and talent recruitment. That push included bringing in Scale AI founder Alexandr Wang to lead the new Meta Superintelligence Labs division.

The question for investors is whether this spending wave will produce durable product advantages quickly enough to justify the cost. Meta's recent history makes that a harder sell than it might otherwise be. The company's previous emerging-technology moonshot, the metaverse, remains an expensive cautionary tale. In the same earnings report, Meta said Reality Labs posted an operating loss of more than $4 billion on just $402 million in sales. The source text says the division has lost more than $80 billion over the past six years.

Why the market is drawing a distinction between AI and the metaverse

Even with that backdrop, there are signs that investors and analysts see Meta's AI push differently from its metaverse spending. The company recently introduced Muse Spark, described in the source text as the first release from Meta Superintelligence Labs. Zuckerberg said the model shows the work is on track to build a leading lab, and he argued that a stronger model base should lead to more novel products.

That argument is central to Meta's case. AI is not a side project for the company. It touches advertising, product engagement, software tooling, and future consumer interfaces. If the infrastructure outlay helps Meta improve core products and create new ones, the return profile could be fundamentally different from a hardware-heavy virtual-world strategy that struggled to find mass adoption.

The wider industry signal

Meta's spending forecast is also a signal about where the AI market stands in 2026. The boom is no longer defined only by model announcements or chatbot usage figures. It is increasingly shaped by who can secure components, build capacity, and keep financing the physical backbone of the sector. In that environment, capital expenditure is becoming a competitive weapon.

Meta's report suggests the next phase of AI competition will be won not only in algorithms, but in procurement, infrastructure, and tolerance for enormous balance-sheet strain. The company's revenue momentum shows the business remains strong enough to fund a larger bet. The market selloff shows investors are still deciding whether this particular bet is disciplined scaling or another costly leap into uncertain territory.

For now, Meta is choosing acceleration. The number attached to that choice, potentially $145 billion, is what turned an otherwise strong earnings day into a referendum on how much the AI race is really going to cost.

This article is based on reporting by Gizmodo. Read the original article.