Tencent and Alibaba join forces! Hangzhou produces the "youngest unicorn"

【Introduction】Jizan Power raised a total of 2 billion yuan in six months, with Tencent and Alibaba investing simultaneously

After the Spring Festival Gala, investment in the robotics sector continues to be hot…

On March 9th, embodied intelligence company Jizan Power officially announced that within less than half a year, it completed five rounds of financing, totaling 2 billion yuan. Its post-investment valuation has exceeded 1 billion USD, making it the youngest unicorn in the sector.

Jizan Power’s financial investors include: Yuanjing Capital, BlueRun Ventures, Sequoia China, Junlian Capital, Zhongke Chuangxing, GaoRong Venture Capital; strategic investors are Tencent and Alibaba Group. The latest funding round was advised by Guangyuan Capital.

Most major shareholders, including Sequoia China, BlueRun Ventures, and Junlian Capital, have continued to increase their investments since the first round.

Former Ideal Auto executives lead the charge in embodied intelligence

It is reported that Hangzhou Jizan Power Technology Co., Ltd. was established in July 2025, with its core founding team from Ideal Auto. According to information, Jizan Power’s CEO Jia Peng (former head of R&D at Ideal智驾), Chairman Wang Kai (former CTO of Ideal), and COO Wang Jiajia (former head of mass production at Ideal智驾).

Investors see this team as having experienced the high-stakes competition in the intelligent driving sector, completing the “from zero to one” and scaling breakthroughs. The three cover all aspects of entrepreneurship, forming the core confidence for institutional investment.

Jia Peng is an expert in VLA models and world models in China, with experience managing core R&D at IBM, NVIDIA, and Ideal Auto. He is the pioneer behind the world’s first VLM+ end-to-end dual-system, and the first to develop and deliver VLA models. Wang Kai previously worked at Nokia and Visteon; in 2021, as CTO of Ideal Auto, he led the team to achieve a record of mass production of intelligent driving systems in just seven months. He later served as Venture Partner at Yuanjing Capital. Wang Jiajia was one of Bosch China’s youngest R&D directors and later led the full-process mass production of intelligent driving systems at Ideal Auto.

To date, the company has completed strategic layouts in Beijing, Shanghai, and Suzhou.

Next, Jizan Power will fully invest in training foundational models, core development and iteration, data collection, and algorithm research, expanding from factories and manufacturing to service industries, domestically and internationally. The company aims to continue technological innovation and industry cooperation, empowering the large-scale deployment of embodied intelligence through technology, accelerating its application across multiple scenarios.

Bringing car-making experience into “robots”

In BlueRun Ventures’ year-end review, CEO Jia Peng said: “Embodied intelligence is one of the most ‘competitive’ tracks in AI. AI applications can rely on small teams for rapid iteration, but embodied systems need to respect hardware’s own iteration cycle. Our way to improve efficiency is to learn from the new energy vehicle industry, introducing standardized processes and manufacturing techniques.”

Jizan Power seems to be applying the same “approach” from car manufacturing to building robots.

VLA, as the “brain” of robots, aims to solve physical world interactions. Whether the carrier is a vehicle or a robot, the underlying logic is consistent. Jizan Power’s LaST foundational model successfully transferred the vehicle VLA architecture to robots: by integrating physical prediction from world models with VLA’s fast-slow thinking, it endows robots with human-like “quick response, slow thinking” capabilities. This architecture aligns with its technological lineage at Ideal Auto, enabling reuse and upgrading from autonomous driving to embodied intelligence.

In terms of data training, Jizan Power deploys computing power on the edge, replicating Tesla’s “shadow mode,” allowing robots to complete data collection and training loops on the hardware itself, at low cost, capturing real-world data.

Jizan Power’s “Human data is all you need” robot learning paradigm divides the learning process into three stages: pre-training with large-scale human operation data to improve generalization; task exploration through human demonstrations; and online learning with real-time human guidance. This layered training approach mirrors the data closed-loop logic in autonomous driving.

In deployment scenarios, Jizan Power’s approach is clear: from closed to semi-open to fully open environments, from B-end to C-end, domestically and overseas. This gradual progression mirrors the evolution of autonomous driving from closed areas to semi-closed, then to full scenarios.

Can car-making experience truly be replicated in robotics? The answer may be revealed next year. As business models and technical routes converge, funding and resources will increasingly concentrate in leading companies.

“Staying at the table is the only way to survive,” said another unicorn robotics leader, “2026 is just the beginning of the real ‘battle’.”

(Source: China Fund News)