Technology

Spirit AI Raises $145M as Humanoid Robots Hit Factory Production Lines

BEIJING: Spirit AI has raised $145 million in a new funding, bringing total funding across all rounds to approximately $420 million. The fresh round was co-led by Shunwei Capital and Yunfeng Fund. The Beijing startup is building what it calls a universal AI brain for general-purpose humanoid robots.

Founded in February 2024 by Han Fengtao, Spirit AI develops vision-language-action models that power physical robots in real-world environments. Its Moz1 humanoid robot features 26 degrees of freedom and full-body force control. The company’s Spirit v1.5 foundation model was open-sourced in January 2026 and topped the RoboChallenge global leaderboard.

Spirit AI has transitioned from research to real-world deployment, with its Xiaomo robot already operating on battery assembly lines at CATL, the world’s largest battery manufacturer.

Also read: Apptronik powers ahead in humanoid race with $520M funding

The robot manages flexible wire harnesses, once considered too complex for automation, with a 99% success rate comparable to skilled human workers. Its Moz robots are also being used in JD.com retail environments.

What differentiates Spirit AI is its data strategy. Instead of relying on curated datasets, the company uses “dirty data”, raw human video and wearable sensor inputs from real-world interactions. This approach cuts data acquisition costs by up to 90% compared to traditional teleoperation.

So far, Spirit AI has collected over 200,000 hours of interaction data and aims to reach one million hours by the end of 2026.

The company expects its revenue to surpass 100 million yuan this year, with new funding directed toward scaling its embodied AI models and expanding data pipelines. While competitors like Google DeepMind and Physical Intelligence are advancing in this space, Spirit AI stands out as the only player currently deployed on a large-scale industrial production line.

Amita Parul

Amita Parul is an Independent journalist with experience in reporting and commentary on current events and sociopolitical developments. She contributes original reporting and analysis that aligns with Tea4Tech’s editorial standards for accuracy, transparency, and context, focusing on business and technology trends. || Amita covers emerging news stories and provides explanatory insights that help readers understand both the events and their implications.

Recent Posts

OpenAI Says Its AI Model Went Rogue, Hacked Startup Hugging Face

San Francisco: OpenAI has disclosed what it is calling an "unprecedented cyber incident," in which…

2 days ago

OpenAI Presence Corporate Software Launches to Help Enterprises Deploy Trusted AI Agents

New Delhi: OpenAI has announced the launch of OpenAI Presence Corporate Software, a new enterprise-grade…

3 days ago

OpenAI Launches ChatGPT Work to Turn Goals Into Finished Work

SAN FRANCISCO: OpenAI has released ChatGPT Work, a new agent mode designed to take a…

4 days ago

Google Drops Three New Gemini Models, But Flagship 3.5 Pro Still Missing

California: Google DeepMind has released three new AI models, Gemini 3.6 Flash, Gemini 3.5 Flash-Lite,…

4 days ago

Alibaba Previews Qwen 3.8, Claiming Second Place Behind Fable 5

HANGZHOU, China: Alibaba unveils Qwen 3.8, a 2.4 trillion-parameter model its team ranks "second only…

6 days ago

Rippling Co-Founder Launches Vorflux With $15M for Autopilot Coding

SAN FRANCISCO: Prasanna Sankar, co-founder and former CTO of Rippling, launches Vorflux with $15 million…

6 days ago