Kai-Fu Lee is raising an AI 2.0 company — and it won't just be a Chinese ChatGPT clone.

By Lixin He

Edited by Lili Yu and Jing Liu

"This will be China's first chance to compete for a platform in the AI space — one ten times larger than mobile internet."

AnYong Waves has learned that Project AI 2.0, the company being assembled by Kai-Fu Lee, has just been formally announced. In a WeChat Moments post, Lee stated: "I am personally assembling Project AI 2.0, a global company dedicated to building an AI 2.0 platform from the ground up and AI-first productivity applications."

This marks Lee's latest move following his recent declaration that "AI 2.0 has arrived, and will give birth to new platforms that rewrite all applications." It is also the seventh company to emerge from Sinovation Ventures AI Institute's "T0 incubation" program.

In Lee's view, the AI 2.0 era will produce "a platform opportunity ten times larger than mobile internet," one that will not only rewrite existing software, interfaces, and applications, but also spawn a new generation of AI-first applications and catalyze AI-led business models.

Project AI 2.0 is not simply aiming to build a Chinese ChatGPT. From Sinovation's perspective, AI 2.0 is neither merely a powerful chat tool nor simply AIGC generation for text and images — even Co-pilot and the applications we see today represent only the beginning of AI 2.0 capabilities.

Currently, Project AI 2.0 is recruiting globally for talent focused on large models, NLP, multimodal systems, algorithms, distributed architecture/infrastructure, and related fields, while simultaneously conducting its first funding round.

AI has long been a flagship focus for Sinovation. During the AI 1.0 era defined by deep learning, the firm nurtured and invested in ten unicorn companies. In Sinovation's framing, if AI 1.0 was akin to the invention of electricity, AI 2.0 is the power grid. The AI 1.0 era already brought recognition capabilities that transformed internet services, commerce, and physical retail, alongside the emergence of robotics and autonomous driving.

The AI 2.0 paradigm has three defining characteristics: First, it requires no human annotation, capable of self-supervised learning from massive text corpora. Second, it operates at enormous scale, demanding thousands of GPUs for training and consuming vast amounts of data and capital. Third, it possesses cross-domain capabilities, able to adapt to different tasks through fine-tuning at relatively low cost, extending from text into other modalities.

The first breakout application of the AI 2.0 era is Generative AI — what has been widely discussed domestically over the past year as AIGC. Generative AI enables self-supervised learning without annotation. AI will evolve from "assisting" humans to gradually "replacing" human labor, and all user interfaces will be redesigned and rewritten. Lee offered an analogy: imagine having AI read the first nine chapters of a book and then "guess" the tenth, then comparing its guess against the actual text. After reading tens of millions of books this way, the model continuously optimizes and iterates. Through this process, AI grows increasingly precise, ultimately forming foundational large models applicable across domains.

In Lee's assessment, "The arrival of the AI 2.0 era represents a massive platform opportunity, and this will be China's first chance to compete for a platform in the AI space."

In recent years, as capital market valuation frameworks have shifted and the overall pool of capital has adjusted, beyond some funds extending into earlier stages, early-stage incubation-style investing has also become more prevalent. In fact, beyond traditional fund investing, Sinovation has consistently cultivated companies through its T0 incubation model. Project AI 2.0 follows AInnovation, Langboat Technology, MetaNovas Biotech, and five other companies as the seventh to emerge from Sinovation AI Institute's T0 incubation program.

So-called T0 incubation means Sinovation must identify technological trends at the earliest possible stage. It then sets a specific technical thesis, deeply mining talent, research papers, and the latest industry developments in that domain. From there, it hand-holds founders through team building, product refinement, business development, go-to-market, and fundraising.

According to sources, in the initial launch phase of Project AI 2.0, Lee will devote considerable energy to personally leading the assembly of the new company, concentrating on recruiting a world-class management team. Once a CEO is appointed for Project AI 2.0, he will balance his time between the company and fund management. This mirrors the model used during AInnovation's T0 incubation in 2017–2018, when Lee led the company in its early days before handing full management authority to CEO Hui Xu upon his arrival.

Image source: Visual China

Layout: Du Meng