Bank of America: Meta Sells Computing Power, Aiming to Tell a Good AI Investment Return Story

Meta is planning to monetize its massive AI computing power assets. This move is not only the early shape of a new business line, but also a strategic signal in response to investors’ doubts about the returns on high capital expenditures.

According to reports, Meta is drawing up plans to launch a cloud infrastructure business, selling external customers access to AI computing power and model services. After the news was released, Meta’s share price jumped by about 10% in a single day—well above the roughly 0.25% gain of the S&P 500 over the same period—showing a positive market reaction to this potential new business line.

Citing information from the Zhui Feng Trading Desk, BofA Securities analysts Justin Post and Nitin Bansal said in a research report published on July 1 that advancing the cloud business could help highlight the potential value of Meta’s computing power assets and model R&D, thereby alleviating, to some extent, investors’ concerns that the company keeps pouring more into AI infrastructure but still has not seen returns for a long time. BofA maintained a Buy rating on Meta, with a target price of $835.

Meta Cloud Plan Emerges: Two Paths Moving Forward in Parallel

According to Bloomberg, citing people familiar with the matter, Meta’s cloud business plan currently has two directions: first, offering AI model hosting services that allow developers to access a variety of models running on Meta’s existing AI infrastructure, including its Muse Spark series models, and charging based on access volume. This model is similar to Amazon AWS’s Bedrock; second, selling raw computing power directly—positioning itself closer to emerging cloud computing service providers such as CoreWeave.

The above plan is part of a Meta internal strategic initiative called “Meta Compute,” focusing on building and operating AI infrastructure. Meta’s CEO has previously publicly hinted at business opportunities in the enterprise market and said the company is likely to sell computing power externally at prices higher than its construction costs.

In its report, BofA pointed out that, from a more macro perspective, if Meta’s 2026 capital expenditure scale can support up to 3GW of computing power buildout (estimated at roughly $40–45 billion per GW), then establishing a cloud business platform in the near term would give the company greater strategic flexibility—once there is excess computing power, it could be leased externally at $10–15 billion per GW per year, providing positive support for the company.

Competitive Positioning in Question, Strategy Debate Hard to Avoid

Despite the strong market reaction, BofA also explicitly pointed out potential doubts. Meta’s progress on internally developed chips appears to be behind that of mature hyperscale cloud service providers such as Amazon, Microsoft, and Google. At the same time, the company is still actively procuring computing power through third-party agreements—including a recent 1.6GW procurement agreement signed with Crusoe.

This has prompted the market to question Meta’s strategic logic: can a company that still needs to buy computing power externally build a convincing computing power resale business? And how would it position its competitiveness in the hyperscale cloud market?

BofA believes that whether Meta can gain stronger market recognition in this arena depends, to some extent, on the frontier development of its large language models (LLMs)—the higher the model capability, the stronger external demand for Meta’s computing power, and the more solid the commercial logic of the cloud business becomes.

AI Unit Economics Improves—A Double-Edged Sword for Cloud Service Providers

Beyond Meta’s cloud plan, there are also notable signals on the AI computing power cost side. According to The Information, OpenAI has reportedly discovered a system-level optimization scheme that reduces inference costs for certain models by about half. The optimization achieves this by using existing server infrastructure more efficiently, without adding new hardware or introducing new model architecture. According to the report, OpenAI has applied this optimization to unauthenticated ChatGPT traffic, and the related traffic can be supported by only a few hundred Nvidia GPUs. It remains unclear what the specific mechanism of this method is, and it is also uncertain whether it can be extended to logged-in users, API workloads, or compute-intensive inference products.

BofA believes that improvements in computing power cost efficiency are generally positive for large internet companies as a whole: if this technology can be rolled out across the industry, it would expand the effective output of existing computing power without increasing hardware investment, reduce the urgency of additional capital expenditures, and improve the unit economics of AI businesses. As agentic application scenarios drive a significant increase in token consumption, the strategic value of computing power optimization will become even more prominent.

However, for hyperscale cloud service providers, the decline in inference costs also creates some downside price-pressure risk. At the same time, a better gross margin structure and a broader addressable market are expected to drive continued growth in demand for AI workloads, and overall this remains positive.

The above insightful content is brought to you by the Zhui Feng Trading Desk.

For more detailed interpretations, including real-time analysis, frontline research, and more, please join【**Zhui Feng Trading Desk ▪ Annual Membership**】

![](https://img-cdn.gateio.im/social/moments-460a44862d-112f3e2057-8b7abd-7649e1)

Risk Warning and Disclaimer

          

            There are risks in the market, and investment requires caution. This article does not constitute personal investment advice, nor does it take into account any particular investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their specific circumstances. Any investment made based on this article is at your own risk.