Meta Platforms Inc. has stunned Wall Street with a capital expenditure forecast that could reach a staggering $145 billion in 2026, an increase of more than $10 billion from earlier estimates. The announcement came during a packed earnings day where tech giants Google, Amazon, and Microsoft also reported results, but Meta’s shares fell over 7% despite a 33% revenue jump—its fastest quarterly growth since 2021. Investors clearly focused on the spending surge rather than the top-line strength, highlighting fears that Meta’s AI bet may be repeating the pattern of its troubled Metaverse investment.
The primary driver of the higher costs, according to CEO Mark Zuckerberg, is "higher component costs, particularly memory pricing." The global AI boom has triggered an unprecedented buildout of data centers, creating a shortage of high-bandwidth memory chips used in AI servers. This memory crisis has not only inflated Meta’s bills but also pushed up prices for consumer electronics such as laptops and smartphones. In 2025, Meta spent $72 billion on capital expenditures; the 2026 figure nearly doubles that, signaling an aggressive pivot toward AI infrastructure.
Zuckerberg’s AI Bet and the Memory Crisis
The memory chip shortage is a direct consequence of the AI explosion. Companies like Nvidia depend on advanced memory modules to power their GPUs, and suppliers such as Samsung and SK Hynix are struggling to keep pace with demand. For Meta, which is building one of the world’s largest AI computing clusters, the component cost increase is a painful but necessary step. “We are confident in this investment,” Zuckerberg told analysts. The company’s data center projects span multiple states and countries, requiring massive procurement of servers, cooling systems, and energy infrastructure.
Meta’s rivals are also spending heavily. Microsoft has committed over $80 billion to AI data centers, while Amazon and Google are each planning similar outlays. However, Meta’s spending relative to its revenue is the highest among the four, raising questions about whether the returns will materialize quickly enough. The firm’s Reality Labs division, which houses the Metaverse efforts, posted another quarterly operating loss of $4 billion on just $402 million in sales, bringing total losses to over $80 billion since 2020. Skeptics worry that AI could become another costly detour.
The Catch-Up Effort and Early AI Products
Zuckerberg acknowledged about ten months ago that Meta had fallen behind in the AI race. Since then, the company has embarked on a massive catch-up plan, recruiting top talent and pouring resources into research. A key hire was Alexandr Wang, the founder of Scale AI, who now leads Meta Superintelligence Labs. The division’s first major output was Muse Spark, an advanced AI model that Meta plans to open-source. “This was the first release from Meta Superintelligence Labs, and it shows that our work is on track to build a leading lab,” Zuckerberg said during the earnings call.
Beyond foundational models, Meta is developing two new AI agents—one for personal use and another for business. The company is already testing business AIs internally, and Zuckerberg reported that weekly conversations with these agents have grown tenfold since the start of 2026. The agents aim to automate tasks such as customer service, scheduling, and content creation. On the consumer side, Meta is integrating AI into its core apps. CFO Susan Li noted that over half a billion weekly users on Facebook and Instagram now watch videos translated and dubbed by AI, showcasing a practical application that enhances engagement.
AI Transforming Advertising and Recommendations
Meta is also using AI to overhaul its advertising and recommendation systems. The company’s ad business, which generates the bulk of its revenue, is being infused with the new Muse Spark model to hyper-personalize feeds. “Since our recommendation systems are operating at such large scale, we’ll phase in this new research and technology over time,” Zuckerberg explained. Early tests suggest that AI-driven recommendations can boost user engagement and advertiser return on investment, a critical metric as Meta competes with TikTok and Google for digital ad dollars.
The recommendation system upgrade is part of a broader trend in the industry: using AI to predict what users want before they know it. Meta’s historical data, combined with real-time behavioral signals, allows the model to prioritize content that keeps users on the platform longer. The payoff could be substantial—even a small percentage increase in engagement can translate into billions in additional ad revenue.
Internal Cost-Cutting and Layoffs
While spending billions on AI, Meta is simultaneously cutting costs through layoffs. The company is eliminating 10% of its workforce and offering voluntary buyouts to 7% of its U.S. staff, following a Silicon Valley trend of using AI to automate roles. Executives declined to directly attribute the job cuts to automation, but CFO Li said a “leaner operating model” would help “offset the substantial investments we’re making.” The layoffs signal that Meta sees AI as both a product and a means to reduce operational expenses.
The move has drawn criticism from labor groups, especially as Meta’s AI investments create new roles in data center operations and model training that may not match the skills of displaced workers. Still, the company defends the decision as necessary to remain competitive. The broader tech industry has seen similar patterns, with companies like Google, Microsoft, and Amazon also reorganizing around AI while trimming non-essential teams.
Meta’s journey is a high-stakes gamble. The $145 billion spending target underscores the belief that AI will reshape every facet of its business—from advertising to social media to enterprise software. Whether the bet pays off will depend on the timely deployment of agents, the success of Muse Spark, and the company’s ability to navigate a volatile memory market. For now, Zuckerberg is all in, hoping that this time the investment yields a different outcome than the Metaverse.
Source: Gizmodo News