On November 16, Alibaba Group released its Q2 2023 financial report, with Alibaba Cloud, its cloud business, reporting a 2% year-on-year growth in revenue to RMB 27.65 million (USD 37.9 million) compared to the same period last year.
In spite of a challenging market, Alibaba Cloud’s profits surged from RMB 387 million (USD 54.0 million) in the previous quarter to RMB 1.41 billion (USD 196.7 million), representing a 264% increase. This jump was primarily attributed to the company’s strategic optimizations.
In the Chinese cloud computing industry where profit margins are typically thin, a substantial increase in the market leader’s profits is undoubtedly a positive signal.
However, Alibaba Cloud’s latest financial achievements have been overshadowed by another piece of news. Six months after Alibaba Group announced its intention to list Alibaba Cloud as an independent entity, it has just announced its decision to terminate the Alibaba Cloud spin-off.
What investors are seemingly concerned about is whether this represents the termination of the IPO as well. During the financial report conference call, Joseph Tsai, chairman of Alibaba Group, did not provide a clear answer, only stating that “we will focus on developing a sustainable growth model based on emerging AI-driven demand for networked and highly scaled cloud computing services.”
In March this year, Alibaba Group announced its “1+6+N” organizational transformation, categorizing its internal businesses into one listed company, six business groups, and several independent companies. Alibaba Cloud was once part of the six business groups and was considered the business with the most potential for an independent listing. For many years, Alibaba Cloud, maintaining double-digit business growth, consistently played a crucial role in driving the overall growth of the group.
The decision to terminate the spin-off could be attributed to changes in the overall environment.
By the end of 2022, ChatGPT gained global popularity, causing a surge in demand for massive computations to support an array of generative artificial intelligence applications. Cloud providers selling computing power experienced another boom as a result. However, chips are often known as the Achilles’ heel of computing power for a reason. Tsai mentioned during the conference call that the decision to terminate the spin-off was due to “uncertainties created by recent US export restrictions on advanced computing chips,” bringing uncertainty to the business and making it impossible to increase shareholder value as the group had originally envisioned.
In fact, the motivation for Alibaba Cloud’s spin-off and listing was more heavily driven by the group’s organizational restructuring. Considering its stage of development, Alibaba Cloud’s decision to go public at that time was already a risky move.
Globally, the parent companies of the top three cloud providers—Amazon Web Services, Microsoft Azure, and Google Cloud—have not spun off their cloud businesses. Alibaba Cloud, ranking fourth, had just turned profitable and faced the significant challenge of surviving independently. Even if it entered the capital market, the current stock market conditions suggest that its post-IPO performance would not be very positive.
Internally at Alibaba Group, the motivation for the spin-off and listing of Alibaba Cloud has also diminished. Daniel Zhang, former chairman and CEO of Alibaba Group, was the main force behind the bold move to split the group. However, with Zhang stepping down, there are now uncertainties about how Alibaba Cloud’s ongoing organizational reform will continue.
“The biggest issue for Alibaba Cloud right now is not a business problem or a market problem, but an organizational problem,” said a recently departed mid-level manager at Alibaba Cloud to 36Kr.
With various factors at play, both internally and externally, the current environment is not the best time for an IPO. It may therefore seem prudent for Alibaba Group to bide time and wait.
However, the external environment is at best a superficial reason. The more pragmatic approach is to accept that generative AI will play a key role in the future of business, and that cloud providers will need more time and resources than previously envisioned if they wish to truly benefit from the AI wave.
For example, tech giant Microsoft is finding it challenging to sustain its heavy investment in AI. Since 2019, Microsoft has invested over USD 13 billion in OpenAI. After integrating its own business with the AI assistant GitHub Copilot, it is likely not profitable. According to a Wall Street Journal report released in October this year, for every user served by Copilot, Microsoft incurs a monthly loss of at least USD 20, which can sometimes go as high as USD 80.
In comparison, Chinese cloud providers, lacking sufficient computing power compared to their American counterparts, have to budget more carefully and take it slower. Even Alibaba Cloud, having recently turned profitable, has not yet escaped the stage of low growth and low profits. Remaining under the umbrella of Alibaba Group may therefore be advantageous for maintaining its early investment in the AI business.
Ultimately, for Chinese cloud providers, the most crucial step in fulfilling the business potential of large models is to ensure a stable foundation.
Just a few days before the release of this quarter’s financial report, Alibaba Cloud experienced a disastrous failure. On November 12, Alibaba Cloud’s global service crashed for nearly three hours, affecting not only its customers but also all internal apps of Alibaba Group, including Taobao, DingTalk, Hema, and more. Notably, this has occurred less than a year after the last large-scale failure in December 2022 when Alibaba Cloud experienced an outage in Hong Kong, interrupting its services in the region for over 12 hours.
Alibaba Cloud has yet to provide a post-mortem and response to this recent incident, but undoubtedly, this has sounded an alarm for the entire Chinese cloud market.
Compared to the era of general-purpose computing power, the age of large models demands exceptionally high-quality computing power. For instance, each training session of a large model requires the initiation of a massive computing cluster. When a service crashes, the entire computing task must be restarted, resulting in losses that can be exponential. According to the estimation of OneFlow, an AI startup, the cost of training an earlier version of GPT-3 is nearly USD 1.4 million per iteration.
Establishing robust infrastructure and putting technology investments to effective use will be a challenge for the entire cloud industry to collectively tackle in the future.
That being said, the silver lining is that the industry has the ability to quickly prioritize what is most crucial.
In the case of Alibaba Cloud, it is continually increasing its proportion of revenue earned from running public cloud services. In this quarter’s financial report, its project-based orders have noticeably decreased, and it proactively relinquished many businesses that inflated its revenue figures but had poor profitability.
At its recently concluded Apsara Conference, Alibaba Cloud declared that it aims to become the most open cloud in the era of AI, emphasizing integration and positioning itself as the “engine” of the AI era. This aligns with Alibaba Cloud’s original mission upon its founding, aiming to serve as the fundamental infrastructure akin to “water, electricity, and coal.” From basic computing power to AI platforms and model services, Alibaba Cloud is undergoing a new wave of product and technological innovation, preparing for the future demands of AI computing.
Over the past few years, Chinese cloud providers have gone through a period of extensive growth, acting as integrators and conquering new territories. During that time, strategies like “integration” and “building an ecosystem” might have been seen as mere slogans, but today, they are more likely to be regarded differently.
In relation, this challenging period of transformation and exploration deserves more patience from the market.
KrASIA Connection features translated and adapted content that was originally published by 36Kr. This article was written by Yong Yi for 36Kr.