Meta’s late move into enterprise AI compute faces a timing gap, Jefferies analyst says

Meta Platforms is pressing into a new target market for AI compute, but a Jefferies analyst says the company may be “years behind” the established leaders in cloud-based AI infrastructure. The view, laid out by Jefferies senior internet analyst Brent Thill on CNBC’s Squawk Box, frames Meta’s strategy as a potential play on a larger supply-and-demand imbalance, rather than a bet on immediate first-mover advantage.

Thill’s core argument is that a large backlog of AI compute demand remains unmet across the biggest cloud providers, particularly Google, Amazon, and Microsoft. He characterized the market as “sold out for six months,” adding that pricing has risen at those incumbents, a sign to him that capacity constraints and demand outstripping supply are still binding. In that setup, he suggested that a credible fourth entrant with capital could win a meaningful share even if it is not the earliest builder.

According to the report, Thill believes Meta can participate through a planned cloud infrastructure business designed to sell excess AI computing capacity. The idea is that Meta’s internal scale and infrastructure investments can be translated into an external service for enterprise customers looking for more AI training and inference compute than existing providers can deliver on current timelines. In his telling, the timing gap matters, but the market’s capacity crunch could still provide room to compete on terms other than price in the near term.

Financial capacity is a central part of Thill’s bullish case. He pointed to Meta’s reported first-quarter FY2026 performance and operating cash flow, and said management increased its full-year 2026 capital expenditure guidance. Capital expenditures, or capex, are spending on items like data centers and servers that typically rise ahead of new capacity and longer-term revenue opportunities in infrastructure businesses. In the report, Thill ties higher planned spending to factors including higher component pricing and additional data center costs.

The analyst also positioned valuation as supportive. The report says Thill described Meta as “cheap” relative to its earnings power, citing a forward price-to-earnings multiple of 18. A forward P/E uses analysts’ estimates of future earnings to compare a stock’s market price to expected profit, and it is often used as a quick heuristic for how much investors pay for each dollar of future earnings.

Even with that valuation argument, Thill acknowledged the strategic difficulty of entering a market where incumbents have built deep customer relationships and operating scale. The report says he estimated that Amazon controls close to half of the cloud market after decades of investment, and argued that Meta will likely need to win smaller business customers before it can compete broadly at scale. In his view, the work required to catch up is substantial, which is part of why Meta is characterized as “really late.”

What is still unclear from the CNBC-linked write-up is how Meta intends to structure this enterprise offering in practice, including where it will target first, what pricing models it may use, and what timeline it expects for meaningful revenue contribution. The report does not detail margins, capacity ramp milestones, or specific enterprise contracts, so investors will have to watch for more concrete updates from Meta on launch plans and performance metrics.