The Zhitong Finance App learned that after meeting with the company's management, the investment agency BNP Paribas believes that Meta Platforms' (META.US) computing power assets have considerable room for commercialization.
Analyst Nick Jones wrote in a report to clients: “Management believes that compared to simply selling raw computing power, the economic benefits of commercializing AI intelligence are much better. Meta will only use computing power for cloud service deployment when internal demand falls short of expectations, and such partnerships will focus on short-term or strategic projects.”
Jones further stated that in view of promising market opportunities, Meta is actively investing in a number of AI projects and continuing to expand its computing power scale. However, management is also aware that due to the macropolitical environment, the supply side may face certain constraints in 2028 and beyond.
Looking deeper, Jones believes that Meta's core advertising business will maintain “sustained” growth, while other revenue streams, including potential cloud services, will also drive the company's overall revenue to achieve a “steady” upward trend.
“Although capital expenditure may continue to expand until 2027, we believe that in the long run, these investments will generate sufficient revenue to support its capital expenditure scale,” Jones added. “On the basis of cost control, we expect the company's profit margin to return to an expansion path.”
Jones set a price target of $855 for Meta and gave it an “outperforming market” rating.
Last month, it was reported that Meta is preparing to build a cloud computing business with the aim of selling its surplus computing power.
Meta CEO Mark Zuckerberg told investors at the shareholders' meeting in May this year that the sale of excess computing power is already under consideration. “This is really within the scope of discussion,” Zuckerberg said. But at the same time, he added that Meta has yet to act because the company believes it still has actual demand for this computing power.