Alibaba to spin off cloud unit, take two others to IPO

Alibaba Group is making plans to carve out its cloud business unit and push two others to IPO, as it shifts up the gear on its organizational overhaul. 

The Chinese tech giant is aiming to spin off the cloud business within a year and will look to bring in external investors through private financing in the lead up. This new entity will become an independent publicly listed company, Alibaba said Thursday. 

Also: The top cloud storage services

IPO plans also are in the works for its retail supermart business Freshippo, also called Hema, and logistics unit Cainiao, subject to the usual conditions and approvals. 

Alibaba is hoping to push Cainiao to IPO in the next 12 to 18 months, while Freshippo's public listing is targeted for the next six to 12 months. 

The moves are part of a major organizational restructure announced in March, from which the Chinese vendor looks to gain agility and respond more quickly to market changes.  

Also: Cloud specialists get the big bucks, but there's still a role for generalists

These efforts will be necessary amid “challenges and opportunities” in an uncertain economic landscape, according to Alibaba chairman and CEO Daniel Zhang. 

Speaking during an earnings call with investors, Zhang outlined plans to drive the company's strategy in three key areas — cloud, consumption, and globalization — to tap market opportunities. 

In cloud, specifically, digitalization and the emergence of artificial intelligence (AI) have pushed demands for computing power, with foundational models expanding to support AI applications across all aspects of life, he said. 

Zhang noted that cloud business models, customer profiles, and development are fundamentally different from other consumer businesses within Alibaba's ecosystem. Spinning off its cloud unit will provide the full independence it needs to “further sharpen” its business strategy and optimize its operations, he said. 

Alibaba's cloud revenue dipped 2% year-on-year in the last quarter, ended March 31, which Zhang attributed to an adjustment of its revenue structure and changes in external market environments, as well as customer profiles. A top customer, for example, had moved away from the vendor's cloud services to an on-premises infrastructure for its international business. The customer's contribution to Alibaba's cloud revenue dropped 41% year-on-year. 

Also: How to use ChatGPT: What you need to know now

The resurgence of the COVID-19 pandemic in China also affected public cloud consumption and led to delays in the delivery of some hybrid cloud projects for the quarter, he said.

Beyond these short-term fluctuations in cloud revenue, Zhang expressed confidence in future growth and opportunities amid the rapid development of AI. 

“The emergence and broad application of AI, large models, and various vertical models have raised new requirements for computing power. This is a huge first-mover advantage for Alibaba Cloud as we have established sizeable IaaS and PaaS to provide stable, secure, high-performance, and cost-efficient computing services,” he said. 

Also: How to set up your own NAS for more reliable data backups

Zhang expressed hope that the company's cloud services could support not only its own generative AI model, but also training and services of other large models and vertical use cases. 

He pointed to further opportunities in offering “model-as-a-service”, where Alibaba would provide its proprietary foundation model to the general public and support its customers, partners, and developers in building customized vertical-specific AI models and services based on its own AI model.

Alibaba launched its large language model, Tongyi Qianwen, last month, which it said will be integrated with all its business applications, including e-commerce, search, navigation, entertainment, enterprise communication, and intelligence voice assistance. More than 200,000 customers have since applied for beta access and work has begun with several industry partners to develop vertical models based on Tongyi Qianwen, according to Zhang.

Also: How to install a cloud service at home: It's easy

Reiterating the company's plans to integrate AI across its businesses, he said: “Starting from Dingtalk, we believe all of our consumer-facing businesses can be reinvented with large models to offer a new AI-based service experience for our users. Qianwen, a large-language model, is just one member in the family of our Tongyi series of pre-training models. We plan to release some of the other large models in the Tongyi series in the near future.”

Alibaba last month also unveiled price cuts of between 15% and 50% for its core cloud products and services in China, as part of efforts to gain a wider footprint in its domestic public cloud market. 

Zhang said such moves would help expand its customer base and drive consumption of its public cloud services, as well as providing the high-performance computing power needed to train AI models and services. “These will provide a healthier and more sustainable growth driver for Alibaba Cloud's long-term development,” he said.