Alibaba and Its New AI Sandbox: Why the Giant Needs Kechuang Kezao

Alibaba is doing what it does best again — multiplying entities to maintain flexibility. While Western analysts are guessing how powerful the next iteration of their language model will be, a new company has quietly appeared in Hangzhou — Kechuang Kezao. The name sounds ambitious, something like 'capable of creating and building,' but behind the façade lies a fairly pragmatic calculation. If you thought the giant had settled down after major restructuring, the registration of this 'subsidiary' proves otherwise.

Let's examine the facts. Registered capital of one million yuan — by the standards of Jack Ma's empire, this is not even a budget for a corporate party, but rather pocket change for coffee for the development department. However, the list of business activities deserves closer attention. There's development of applied AI software, work with basic software, and data storage services. In fact, Alibaba is building yet another testing ground for technologies that will become the foundation of their cloud services tomorrow.

Why is this happening right now and in this particular format? After the company switched to the '1+6+N' management scheme, the parent structure finally transformed into an incubator. Instead of inflating payrolls within existing departments, it's far more profitable for them to create compact, legally separate units. This gives them the very maneuverability that corporate leviathans so sorely lack. In a world where new neural network architectures emerge monthly, any bureaucratic delay means defeat.

Interestingly, the founders are Alibaba (China) Network Technology and Hangzhou Alibaba Advertising. This is a classic pairing for internal technology projects. Hangzhou remains the 'capital' of the empire, and the establishment of a new growth point here emphasizes that the company has no intention of relocating its think tanks, despite all the regulatory storms of recent years.

The new Kechuang Kezao structure will likely engage in 'rough' work — creating the very basic tools on which loudly advertised consumer products will subsequently be launched.

Why should we follow this? Because such seemingly minor moves often precede the launch of new infrastructure solutions. Basic AI software is not about chatbots for generating images. It's about the architecture on which cloud computing and complex data processing algorithms for retail and logistics are built. Alibaba is clearly preparing the ground for something larger than just another mobile app update.

Perhaps we are looking at a future supplier of solutions for the Chinese government sector or specialized tools that need to be kept separate from the main cloud platforms due to sanctions risks.

In any case, the emergence of Kechuang Kezao is a symptom that Alibaba is shifting to a tactic of small strike teams. They need to code quickly, test quickly, and rapidly shut down failed projects without exposing the main brand. In the context of fierce competition with Baidu and their Ernie model, as well as Hunyuan from Tencent, such a strategy appears to be the only correct one.

We're watching to see how this 'million-yuan' startup transforms into something larger. Key point: Alibaba continues to fragment its structure for the sake of speed in developing fundamental AI software. Did Claude and GPT force the Chinese to run twice as fast?