Against the backdrop of steadily rising investment in artificial intelligence, Meta Platforms (META) saw its share price decline sharply in early trading on Thursday, falling as much as 4% and still down approximately 3.7% as of 10:49 a.m. Eastern Time. The market sentiment fluctuation primarily stemmed from CEO Mark Zuckerberg’s candid admission to employees that the company’s AI business progress had fallen short of expectations, as well as Wall Street institutions’ prudent assessments of capital expenditure pressures and the cloud business outlook.
According to media reports, Zuckerberg acknowledged at Thursday’s internal all-hands meeting that Meta’s development in the AI agent space over the past four months “has not accelerated as expected,” while the recent organizational restructuring and layoffs have also failed to achieve desired results. In May of this year, Meta launched a global layoff plan affecting approximately 10% of its workforce and reassigned about 7,000 employees to AI projects. Zuckerberg had emphasized the need to rebalance between AI infrastructure and labor costs, noting that increased investment in computing power necessarily required cuts in other expenditures, making a moderate reduction in headcount a necessary measure.
At the same time, Meta has reportedly been planning to launch a cloud computing business to lease out its idle AI computing power resources. This news had once driven the share price up by about 10%, but some analysts believe this move precisely reflects that current AI computing power demand may not have fully met the market’s previously optimistic expectations. A similar situation has also emerged at SpaceX (SPCX), where a large data center built after its merger with xAI has begun leasing computing resources to Google (GOOGL) and Anthropic due to insufficient utilization of some computing capacity. Additionally, Palantir (PLTR) CEO Alex Karp publicly questioned the industry’s prevailing token-based pricing model, arguing that excessive focus on token usage volumes may lead developers to prioritize short-term call counts over long-term commercial value, which is not conducive to the healthy development of the AI software ecosystem.
Despite the aforementioned concerns, the market is not uniformly pessimistic about the outlook for AI computing power demand. French bank BNP Paribas analyst Stefan Slowinski pointed out that the short-term pricing environment for AI computing power remains strong, with SpaceX’s recent infrastructure cooperation agreements with Anthropic and Google, as well as Amazon’s (AMZN) AWS raising prices for GPU capacity reservation services by approximately 20%, both indicating that leading AI companies remain willing to pay a significant premium for quality computing power, and that current demand still visibly exceeds supply. Against this backdrop, the institution believes that AI cloud computing service provider NEBIUS (NBIS) stands to benefit as enterprises shift toward lower-cost open-source AI models, offering more cost-effective cloud services. As for CoreWeave (CRWV), although its share price performance this year has lagged behind NEBIUS, the second half of the year may present valuation recovery opportunities as new computing capacity comes online and pricing absorbs rising hardware costs.
In a comprehensive view, Wolfe Research analysts noted that for every gigawatt of computing capacity sold, Meta’s earnings per share could increase by 20%, but the company’s capital expenditures are projected to reach $200 billion in 2027, up from the previously estimated $160 billion, potentially requiring financing support. Currently, Meta’s stock price-to-earnings ratio stands at approximately 21 times, representing a significant discount compared to larger technology peers, offering investors an opportunity to gain exposure at a relatively discounted price. As the AI industry enters the commercialization implementation phase, market focus is shifting from model scale and computing power build-out to application efficiency, computing power utilization rates, and business model sustainability, with industry competition entering a new stage.
