Shares in Meta Platforms inventory sank by over 10% on Thursday after the tech big’s newest earnings report revealed extra AI spending. Meta’s third-quarter earnings beat expectations on income however fell brief on earnings per share resulting from a one-time, tax-related cost. Nevertheless, the true dip got here after Meta revealed elevated capital expenditures this yr and in 2026 resulting from its AI buildout.
Mark Zuckerberg’s firm has been one of many largest traders in AI in the course of the ongoing AI wave in 2025. In comparison with Amazon, Microsoft, and Apple, that are all fairly large traders as effectively, Meta has been on a large spending spree in latest months. The corporate has spent billions to rent expertise and construct the information facilities mandatory to satisfy its demand for AI. Nevertheless, if EPS is falling brief prefer it did in Q3 resulting from elevated spending on AI, traders seemingly imagine investing within the firm isn’t well worth the threat.
Moreover, Meta says it’s additional rising its capital expenditures estimates for 2025 to $70 billion to $72 billion, versus its prior outlook of $66 billion to $72 billion. “Our present expectation is that capital expenditures greenback progress shall be notably bigger in 2026 than in 2025,” Meta CFO Susan Li mentioned in her commentary. “We additionally anticipate complete bills will develop at a considerably quicker proportion price in 2026 than in 2025, with progress pushed primarily by infrastructure prices, together with incremental cloud bills and depreciation.”
Whereas Wall Avenue seems nervous about Meta inventory, Mark Zuckerberg isn’t. “Meta Superintelligence Labs is off to a terrific begin, and we proceed to guide the business in AI glasses,” he mentioned in the course of the earnings name If we ship even a fraction of the chance forward, then the subsequent few years would be the most enjoyable interval in our historical past.” The Meta CEO seems to be banking on large positive aspects sooner or later to make up for the excessive spending now. He additionally provided a contingency plan, noting that if AI doesn’t develop as anticipated, the corporate can repurpose infrastructure for different workloads “in a really worthwhile means.“
