The Record-Breaking Expenditure

On Wednesday, Meta reported its first-quarter 2026 earnings, surprising many with a massive increase in capital expenditure forecasts. The company now expects to spend between $135 billion and $145 billion this year, up from $72 billion in 2025. The news sent Meta's stock down over 7% in after-hours trading, even though revenue grew 33% year-over-year, the fastest rate since 2021. Investors are clearly worried about the scale of spending and its potential impact on profitability.

The spending is largely driven by the artificial intelligence boom. Meta is building out massive data centers to train and deploy its next-generation AI models. CEO Mark Zuckerberg attributed the increase to higher component costs, particularly memory pricing. The global memory chip supply has been constrained by a surge in demand from AI companies, leading to a memory crisis that has also driven up prices for consumer electronics like laptops and smartphones.

Why So Much? The Memory Crisis

Meta's capital expenditure is not just about buying servers and GPUs. The company needs vast amounts of high-bandwidth memory (HBM) for its AI accelerators. HBM is a critical component for training large language models, and its supply has been tight since early 2025. Samsung, SK Hynix, and Micron have ramped up production, but demand from hyperscalers like Meta, Google, Microsoft, and Amazon has outstripped supply. This has led to bidding wars and premium pricing, directly impacting Meta's bottom line.

The memory crisis extends beyond AI. Consumer electronics manufacturers have seen their costs rise, and many have passed those costs to consumers. Laptop prices have increased by an average of 15% over the past year, while smartphones have become noticeably more expensive. Analysts expect the situation to persist until at least 2027, when new fabrication plants come online.

A Bet on AI to Regain Momentum

Meta has been perceived as lagging in the AI race. While Google DeepMind, OpenAI, and Microsoft have dominated headlines with advanced models like Gemini and GPT-5, Meta's Llama series, though open-source, has not achieved the same level of adoption for enterprise applications. About 10 months ago, Zuckerberg acknowledged the gap and announced a major catch-up effort, committing billions to R&D and talent acquisition.

A key move was hiring Alexandr Wang, founder of Scale AI, to lead the newly formed Meta Superintelligence Labs. This division is tasked with developing next-generation AI systems that can power everything from recommendation algorithms to autonomous agents. The first product of this lab was Muse Spark, a proprietary model that Meta plans to open-source in the future. Early benchmarks show Muse Spark rivaling the performance of Google's Gemini 3, suggesting the investment is already yielding results.

The Metaverse Hangover

Many industry observers are skeptical of Meta's massive AI bet, given the company's previous experience with the metaverse. Reality Labs, the division responsible for augmented and virtual reality, reported an operating loss of over $4 billion in the first quarter of 2026 on just $402 million in revenue. Since 2020, the division has accumulated losses exceeding $80 billion. While Zuckerberg has doubled down on the metaverse vision, the financial drain has made Wall Street wary of any new grand ambition.

However, experts note that AI has a clearer path to revenue than the metaverse. Meta already uses AI for its core advertising business, and improvements in recommendations and personalization can directly boost advertiser returns. During the earnings call, Zuckerberg emphasized that AI is not a speculative bet but a critical infrastructure upgrade that will enhance user engagement and ad efficiency.

New Products and AI Agents

As part of its AI push, Meta is preparing to launch two new AI agents: one for personal use and one for business use. The personal agent will be integrated into Facebook, Instagram, and WhatsApp, allowing users to have natural conversations, get recommendations, and automate tasks like scheduling and shopping. The business agent will help companies create chatbots, handle customer service, and manage internal workflows.

Zuckerberg revealed that early tests of the business agents have been promising. Weekly conversations with these AIs have grown 10-fold since the start of the year. This suggests strong demand from enterprises looking to automate routine interactions. Meta plans to monetize these agents through subscription fees and advertising, potentially creating a new revenue stream that could offset the massive capital expenditure.

Internal AI Integration and Workforce Changes

AI is also transforming Meta from within. CFO Susan Li noted that over 500 million weekly users across Facebook and Instagram now watch videos that have been automatically translated and dubbed using AI. This feature, powered by Meta's SeamlessM4T model, breaks down language barriers and opens new markets for content creators.

Furthermore, Meta is cutting 10% of its workforce, with voluntary buyouts offered to 7% of US employees. Executives declined to say whether these cuts are directly due to automation, but Li stated that a 'leaner operating model' will help offset the substantial investments in AI infrastructure. This reflects a broader Silicon Valley trend where companies are using AI to reduce headcount while boosting productivity.

Meta's recommendation systems are also getting an AI upgrade. By incorporating Muse Spark into the algorithms that power feeds on Facebook and Instagram, the company aims to hyper-personalize content, showing users exactly what they want to see. Zuckerberg noted that engagement improvements are already visible, and advertiser value is increasing. The system is being phased in gradually to avoid disruption at such large scale.

As Meta navigates this ambitious transformation, the outcome remains uncertain. The $145 billion bet is not just about catching up; it is about defining the next era of computing. If successful, Meta could emerge as a leader in both AI and social platforms. If it fails, the company may face a financial crisis similar to what the metaverse has already caused. For now, all eyes are on the second half of 2026, when Muse Spark is expected to be open-sourced and the new AI agents are rolled out globally.

Source: Gizmodo News