Ex-Qwen Lead Lin Junyang Launches AI Startup

Former Qwen Chief Lin Junyang Launches $2 Billion AI Venture

Lin Junyang, the former technical lead of Alibaba’s Qwen large language model team, has officially entered the AI startup arena. He is raising funds for a new company with a valuation target of approximately $2 billion (20亿美金). This move signals a significant shift in the global generative AI landscape, as one of China's most influential model architects steps out from a tech giant to build independent infrastructure.

The venture focuses on cutting-edge areas including world models and embodied intelligence (often referred to as 'embodied brains'). These fields represent the next frontier in artificial intelligence, moving beyond static text generation to dynamic, interactive systems that understand physical reality. Lin’s departure from Alibaba marks the end of an era for the Qwen project, which he helped elevate to global prominence.

Key Facts About the New Venture

• Founder: Lin Junyang, former head of Alibaba Qwen R&D.
• Valuation: Targeting ~$2 billion USD in initial funding rounds.
• Core Focus: Development of world models and embodied AI architectures.
• Team Composition: Recruiting talent from ByteDance, Tencent, and overseas tech hubs.
• Investor Interest: Engaging with top-tier firms like Sequoia China and Gaorong Capital.
• Timeline: Launched two months after his March 2026 resignation announcement.

The Strategic Pivot from Corporate Giant to Startup

Lin Junyang’s decision to leave Alibaba was abrupt yet decisive. In early March 2026, he announced his departure on social media platform X, stating, 'me stepping down. bye my beloved qwen'. This public farewell followed an internal message to his team where he expressed regret over not being able to continue leading them. The speed of his exit suggests a premeditated plan to pursue independent innovation rather than a reactive career change.

During his three-year tenure starting in 2022, Lin transformed Qwen into a major player in the open-source community. Under his leadership, Qwen adopted a strategy of releasing models across various sizes, making it accessible to developers worldwide. This approach contrasted sharply with competitors who kept their largest models proprietary. Lin’s vision prioritized ecosystem growth over immediate monetization, a strategy that resonated deeply with the global developer community.

His new venture now seeks to apply this same open, collaborative philosophy to more complex AI domains. World models require AI to simulate how the world works, predicting outcomes based on sensory input. Embodied AI involves giving these models a 'body' or interface to interact with the physical world, such as robots or autonomous agents. These are significantly more resource-intensive and technically challenging tasks than standard language modeling.

Competitive Landscape and Investor Confidence

The $2 billion valuation places Lin’s new startup in the upper echelon of AI ventures globally. For context, many established AI companies struggle to reach such valuations without proven enterprise revenue streams. This high valuation reflects investor confidence in Lin’s technical pedigree and the strategic importance of the sectors he is entering.

Top-tier venture capital firms are actively participating in the fundraising process. Sequoia China and Gaorong Capital are among the key investors in discussions. Their involvement signals strong institutional belief in the potential of world models and embodied AI. These investors have previously backed successful tech unicorns in Asia, suggesting they see Lin’s team as capable of achieving similar scale.

The recruitment strategy further underscores the ambition of the project. Lin is assembling a team with diverse backgrounds, pulling talent from major Chinese tech giants like ByteDance and Tencent. Additionally, he is hiring professionals with overseas experience. This hybrid team structure aims to combine local engineering efficiency with global research perspectives, a critical advantage in competing with Western labs like OpenAI and Anthropic.

Why World Models Matter Now

• Beyond Text: Current LLMs predict the next word; world models predict the next state of reality.
• Simulation Capability: They allow AI to test scenarios in virtual environments before real-world application.
• Robotics Integration: Essential for creating robots that can navigate unpredictable physical spaces.
• Data Efficiency: Potentially reduce the need for massive labeled datasets by learning from raw video/sensor data.
• General Intelligence: A key step toward Artificial General Intelligence (AGI) by understanding cause and effect.

Implications for the Global AI Market

Lin Junyang’s move highlights the intensifying competition in the Asian AI sector. While Western companies like NVIDIA, Microsoft, and OpenAI dominate headlines, Chinese innovators are rapidly closing the gap in specialized architectures. The focus on embodied intelligence is particularly notable, as it aligns with China’s strengths in manufacturing and robotics hardware.

This development also impacts the open-source community. Lin’s previous work ensured that high-quality models were available to researchers outside big corporations. If his new startup continues this tradition, it could accelerate global progress in robotics and simulation. However, if the technology remains proprietary, it may consolidate power within a few well-funded entities.

For businesses and developers, the emergence of advanced world models offers new opportunities. Industries ranging from logistics to autonomous driving stand to benefit from AI that understands physical constraints. Unlike current chatbots, these systems could manage complex supply chains or operate machinery with minimal human oversight. The integration of language understanding with physical action represents a paradigm shift in automation.

Looking Ahead: What Comes Next for Lin’s Team

The coming months will be critical for Lin Junyang’s new company. Securing the full $2 billion in funding will require demonstrating early technical milestones. Investors will look for prototypes that prove the viability of their world model architecture. Given the rapid pace of AI development, delays could allow competitors to capture market share.

Furthermore, the regulatory environment in both China and the West is evolving. Startups working on embodied AI must navigate safety guidelines and ethical considerations carefully. Lin’s experience in managing large-scale projects at Alibaba will be invaluable in establishing robust governance frameworks. His ability to balance innovation with compliance will determine the long-term sustainability of the venture.

As the AI industry moves from the 'big model' phase to the 'agentic' phase, leaders like Lin Junyang are redefining what is possible. His transition from corporate executive to founder illustrates the dynamic nature of the tech sector. For the global audience, this story serves as a reminder that innovation is not confined to Silicon Valley. The next breakthrough in artificial intelligence may very well emerge from the collaboration between Eastern engineering prowess and Western theoretical frameworks.