0 products received investment from AMD's CEO Lisa Su. Liu Wei, a former top AI expert at Tencent, started a video startup and raised another $80 million.

QbitAI has learned that there is new progress in the startup project of Liu Wei, a former top AI expert at Tencent.

Just now, the AI video startup Video Rebirth announced that it has officially completed a financing round totaling $80 million (approximately RMB 550 million).

Liu Wei is a former outstanding scientist at Tencent, an IEEE/AAAS Fellow, and was the core technology leader of Tencent's Hunyuan large model. He has profound experience in the fields of computer vision and multimodal AI. After leaving Tencent to start his own business at the end of 2024, he focused on the AI video generation track.

Notably, Video Rebirth has received significant capital investment in succession in just over a year since its establishment.

Currently, Liu Wei is leading a team to independently develop an industrial-grade video generation model called Bach, which has not been released to the public yet.

$80 million in new financing

QbitAI has learned that this $80 million financing not only attracted top Asian venture capital institutions but also strategic cross - border giants such as AMD Ventures and Hyundai. Many leading enterprises in the pan - entertainment, transportation, and technology fields participated in this investment.

The $80 million will be used for the commercialization of Video Rebirth's cutting - edge video generation product line, the Bach series of models, and the expansion into the global market.

Video Rebirth is headquartered in Singapore. Currently, its business only covers overseas markets, and its customer base mainly consists of enterprises and professional individual users.

QbitAI learned from a source that this financing round started in November 2025, and when many institutions confirmed their investment intentions, the company's products were still in the R & D stage.

"(Investment institutions) mainly look at the team. They are quite optimistic about the technical route and differentiated positioning of Video Rebirth's world model. Later, it will definitely depend on the products and users."

When asked about the reason for participating in this financing, Sagi Paz, the head of AMD Ventures, said that Video Rebirth builds its world model in a video - native way, which is highly consistent with AMD's strategic vision of enabling cutting - edge AI technologies.

"We are very honored to be their long - term underlying infrastructure partner, providing them with the most solid and high - performance computing power foundation to fully support Video Rebirth in bringing the 'industrial - grade digital world' to the global market."

Keith Noh, the head of Hyundai Motor Group's ZER01NE, said that they invested because the technological value of Video Rebirth far exceeds the scope of traditional creative media. "We are very optimistic about the commercial potential of this AI engine: it can build an extremely realistic digital world, providing an excellent simulation scenario for training Physical AI."

Bookyung Kim, the director of HIVEN under the CJ Group, which has a top - tier entertainment portfolio spanning dramas, movies, and music in South Korea, also said: "From the perspective of HIVEN, this investment aims to jointly explore real - world commercial application scenarios with Video Rebirth in a broader CJ ecosystem such as CJ ENM."

In the past six months, there have been frequent reports of financing for AI video generation companies in the industry. The industry has officially entered a critical stage of commercialization and large - scale expansion from the technology verification period.

Represented by Seedance 2.0, Sora, and PixVerse, AI video has solved core pain points such as audio - video synchronization, character stability, and controllable narrative.

In the view of industry insiders, after the technology maturity crosses the critical point, scenarios such as advertising and marketing, short - drama creation, e - commerce display, film and television pre - production, and game animation are rapidly expanding. AI video is upgrading from a 'novelty tool' to a new - generation infrastructure for the content industry.

One year of entrepreneurship by a top Tencent AI expert

As AI video generation is regarded as the next core breakout point in AIGC, Liu Wei and Video Rebirth are undoubtedly at the forefront of this wave of technological innovation.

Although the products have not been officially released, capital has already made early - stage investments. It has to be said that Liu Wei's academic background and technical resume have given capital a lot of confidence.

Liu Wei is an alumnus of Zhejiang University's Zhu Kezhen College. In 2007, he went to Columbia Engineering School to pursue a Ph.D. in Computer Science and Electrical Engineering.

After graduating with a Ph.D., he worked at IBM for three years. His main work was to conduct research and engineering projects on large - scale/parallel/distributed machine learning and big data analysis. He also participated in projects related to web image search, face recognition, and verification.

In June 2016, Liu Wei joined Tencent, where he was responsible for the R & D of computer vision, multimedia AI, and the Hunyuan large model at Tencent AI Lab.

In November 2024, it was reported that Liu Wei had ventured into AI entrepreneurship.

Liu Wei has served as a guest editor and reviewer for many international authoritative journals. In the year after leaving Tencent, he and his team have still published many papers. According to Google Scholar, his papers have been cited more than 59,000 times, and his h - index is as high as 116.

"Video generation is the most direct path to building an interactive world model, but the scaling law of the industry has not been fully explored," Liu Wei said when he received the first - round financing of $50 million.

What has he done in this year of entrepreneurship?

In response to the pain points such as "image distortion" or object deformation often found in existing AI models, Liu Wei led a team to develop Bach, an industrial - grade AI video engine.

Bach uses the Physical Native Attention (PNA) mechanism to accurately simulate spatio - temporal dynamics, ensuring absolute coherence of characters and physical logic throughout a long shot. With its unique MSSL innovation technology, Video Rebirth can provide smooth native 30fps video materials.

In addition, the unique Dual Diffusion Transformer (DDiT) architecture of the Bach engine can achieve extreme instruction following, solving the pain point of time - consuming repeated generation for professional creators.

At the recent GDC2026 event, the company rented a booth. Foreign media has evaluated that its products can compete with OpenAI's AI video generation model Sora.

The above - mentioned source told QbitAI that Video Rebirth will conduct a new large - scale financing round at the end of this month and has already contacted some investors on a small scale.

Reference links:

[1]https://scholar.google.com/citations?user=AjxoEpIAAAAJ&hl=en

[2]https://www.taihecap.com/transaction_show-9-1022.html#jqkdxw

[3]https://www.bloomberg.com/news/articles/2025-11-05/tencent-ai-veteran-snags-funds-for-year-old-rival-to-openai-sora

This article is from the WeChat official account "QbitAI". Author: Tian Yanlin. Republished by 36Kr with permission.