Alibaba: The primary goal is AGI

First, the main title: In the next 3 years, AI infrastructure investment will exceed the past 10 years of infrastructure investment..... Approximately 380 billion was invested in the past 10 years. Investing this much in 3 years means a baseline of 120 billion per year. Don't argue about whether it's beyond expectations, it must be beyond expectations. Because in the past, everyone was talking about 2025 as the year , now telling you about continuous investment over three years is even more certain than North American CSP.

A few days ago, I bothered a guy from Alibaba Cloud's strategy department, asking, "What is your capex?" He sent me this chart:

The implication is: If we benchmark against the average capital expenditure density overseas in the past 5 years, it is also 12.4%, and Alibaba's revenue in 2025 is expected to be around 140 billion USD, which corresponds to about 120 billion RMB in capital expenditure. But if we compare it to the investment density in 2025, that would be 20%... which means 200 billion in capital expenditure. My personal understanding is that the future 3 years will be within this range.

Another title circulated at the meeting — Alibaba states that the primary goal of its AI strategy is: "Pursuing AGI, pursuing the boundaries of model intelligence capabilities. This goal exceeds any application scenarios we can currently see. The current scenarios are merely opportunities arising from the process of AI enhancement. The standard definition of AGI is over 80% of human capabilities, from this perspective, globally 50% of expenditures are labor costs (both mental and physical), if AGI is achieved, then AI will become the largest industry globally, potentially replacing 50% of the current GDP composition."

Adding Capex, this is quite a bold statement... What did we read from this? The "back of the head" meeting was not in vain. I remember Zheng Shitang published an article before the meeting, which was very insightful — "The tiger descends the mountain." We have previously mentioned that now the interests of the internet and the upper echelons have reached a high degree of consensus on AI. Investing in AI now is politically correct. Moreover, it is also profitable. Today's internet not only needs to invest in AI domestically but also to go overseas to seize the AI market (before each country establishes its own firewall).

Other key points from the meeting:

• 90% of AI tokens will be generated and output on cloud computing networks, with the highest efficiency only available in the cloud.
• Quarterly expenditure: The annual cycle is a relatively average expenditure arrangement , each quarter will have fluctuations, arranged according to supply chain conditions (therefore, a sequential increase may be a bit aggressive, but can be considered as an upside case)
• Model commercialization? The business model of the base model itself is indeed unclear. Open source and closed source are similar, benefiting cloud computing companies, as they all need hosting on cloud computing networks. Whether it is intelligent homogenization or intelligence itself being the biggest commodity, we are well prepared.
• What is the largest AI application within Alibaba? The model's capabilities are becoming stronger, and the evolution speed is accelerating. It's hard to say. However, there are some obvious opportunities internally: Taobao is developing many projects (user interaction, transaction efficiency, transaction conversion), Quark (our AI search, with the highest user volume for AI search in China), DingTalk (the most important tool for B2B), Amap (becoming a gateway for life services through AI).
• Jiang Fan mentioned in the earnings call that e-commerce will continue to invest to maintain market share, improve the business environment for merchants, and acquire new users.
• 60% of customers use it for inference, and the demand for inference during the Spring Festival is explosive.
• Alibaba Cloud is well prepared, adapting to a variety of chips, and is ready regardless of how the ban policies change (Alibaba currently confirms it has: NVIDIA cards, self-developed Tianshu, Ascend, and Cambricon).

This article is sourced from: Information Equality, original title: "Alibaba: The Primary Goal is AGI"

Risk Warning and Disclaimer

The market has risks, and investment requires caution. This article does not constitute personal investment advice and does not consider the specific investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investing based on this is at one's own risk