Securities Times: Meta 'crashes' tech stocks? Industry insiders: Overcapacity in computing power is a misinterpretation!

On July 2, A-share tech stocks plummeted, with sectors such as semiconductors, computing hardware, and memory chips experiencing significant declines. Taking the "dual innovation board," where tech stocks are concentrated, as an example, the ChiNext Index and the STAR Composite Index fell by 5.71% and 5.64%, respectively.

From a market perspective, the sharp drop in A-share tech stocks stemmed from news that Meta is selling computing power, which the market interpreted as an oversupply of computing resources. However, multiple industry insiders believe this interpretation is a misunderstanding; Meta's sale of computing power is not the end of AI capital expenditure, but rather signals the maturation of the AI infrastructure business model.

"Selling computing power is positive for Meta. Previously, whenever the company raised its AI CapEx, the market worried about how these investments would be recouped. Now, if Meta can sell part of its AI computing power externally, CapEx is no longer just a cost but could become a direct source of revenue." Tianfeng Securities said, "The market previously penalized Meta for high CapEx, but now it may reassess these assets in reverse."

Zhang Qiyao, chief strategy analyst at Industrial Securities, also believes that Meta is a unique entity among hyperscalers. Its primarily B2C business means that its AI monetization relies mainly on advertising. Exploring cloud services could enhance shareholder returns and cash flow.

As one of the companies most capable of internally digesting AI computing power globally, Meta has a large number of usage scenarios such as advertising and recommendations. If Meta rents or sells part of its computing power externally, investors will undoubtedly worry: Have major companies engaged in phased advance procurement over the past two years? Have some H100/H200 chips transitioned from the scarcest training resources to inference assets that need to find external monetization?

In response, Tianfeng Securities believes that Meta's move into AI cloud does not mean admitting a full-scale GPU oversupply. It is more likely that Meta is allocating AI computing power of different generations to different economic uses: the latest GB200/GB300/Rubin resources prioritize next-generation model training; while the previous generation H100/H200 is increasingly shifted toward inference and external computing power sales. Meanwhile, Google once failed to fully meet Meta's access demand for Gemini due to capacity constraints, which actually indicates that frontier models and high-quality inference capacity remain tight. Therefore, this is not the end of AI CapEx transactions, but rather the evolution of a business model from "pure cash-burning infrastructure" to "chargeable platform assets."

Zhongtai Securities also stated that Meta's rental and sale of computing power is merely the implementation of "Meta Cloud," unrelated to "oversupply." Tracking Token and ARR slopes in the near term reveals that AI demand is still in the steep growth early stage. The backlog of orders for North American cloud vendors' cloud segments is about 14 times quarterly revenue. "Computing power is just being shifted from Meta to first-tier large models like Anthropic; the supply-demand gap at the industry level will not change."

Tianfeng Securities emphasized that the market should not simply interpret Meta's external leasing and sale of computing power as "AI computing power demand peaking." "More accurately, this is Meta's attempt to transform AI infrastructure from a pure cost center into rentable, chargeable, and platformable assets."

In fact, Meta's plan to sell computing power is not new information. At the shareholder meeting in May this year, it was hinted that if Meta builds more computing power than internally needed, it would consider selling excess computing power or launching API services.

Tianfeng Securities believes that for the hardware supply chain, what truly needs to be observed is not "whether Meta rents out computing power" itself, but whether the real token usage and ARR growth of OpenAI and Anthropic continue to rise. If the ARR and token usage of model companies like OpenAI/Anthropic continue to grow, and hyperscaler CapEx is not substantially revised downward, then the hardware main line remains valid.

"Short-term volatility provides another window for positioning. In late July, as the earnings season reaffirms the boom advantage and inflation data confirms a peak, tech stocks are expected to hit new highs," Zhang Qiyao said.

Source: Securities Times

Risk Warning and Disclaimer

        Market risk exists, and investment requires caution. This article does not constitute personal investment advice and does not consider the specific investment objectives, financial situations, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment decisions based on this article are at the user's own risk.