UBS: Meta selling AI computing power may not be bad news, and could instead ease earnings pressure.

BlockBeats News, July 2 - UBS believes that if Meta sells AI computing power or model access services to external customers, it should not necessarily be interpreted by the market as a negative signal of "AI infrastructure overcapacity." On the contrary, this could become a path for Meta to convert its massive AI investments into revenue more quickly.

In a First Read report released on July 1, UBS mentioned that Meta is reportedly considering two commercialization methods: one is selling "raw" computing power to external companies, and the other is providing access to AI models hosted on Meta's infrastructure. The report stated that Zuckerberg has previously mentioned similar options in public, so this is not entirely new news.

However, this direction may still unsettle some investors. Meta's long-term growth opportunities previously outlined to the market mainly include advertising, more immersive content experiences, business messaging, Meta AI, AI devices, etc., rather than directly becoming a cloud computing or computing power provider. Therefore, if the company actually sells computing power externally, the market may question: Is this active monetization, or a passive digestion after excessive AI capital expenditure?

UBS's judgment is more pragmatic. The bank believes that one of the core issues Meta currently faces is the excessively long AI investment cycle and unclear revenue realization timeline. Compared to waiting for Meta AI chatbots or enterprise intelligent agent businesses to gradually scale, selling cloud computing power or model access could bring in nearer-term revenue, thus alleviating investor concerns about EPS remaining flat or even declining in 2027.

The report maintains a Buy rating on Meta with a 12-month price target of $865, while the stock price listed in the report is $601.85. UBS estimates Meta's diluted GAAP EPS for 2026 and 2027 to be approximately $32.6 and $33.0 respectively, and stated that it will not adjust earnings forecasts before the company confirms the relevant news. Its price target is still based on 26 times the expected full-year diluted GAAP EPS through the first quarter of 2028.

The implication of this report is that Meta's AI narrative is entering a new phase: the market is no longer just asking how much it will spend on GPUs and building data centers, but is beginning to demand to see how these investments will generate returns. For UBS, selling computing power is not a strategic retreat, but rather adding a cash flow outlet to the AI investment cycle.