---
title: "The 1992-born founder steering China's Kua Fu humanoid robot to an IPO"
date: 2026-11-09
category: Builders
site: NeuroAI
canonical: https://neuroai.site/a/na-people-leng-xiaokun-leju-robotics
language: en
---

# The 1992-born founder steering China's Kua Fu humanoid robot to an IPO

> Leng Xiaokun (冷晓琨), who founded Leju Robotics (乐聚机器人) at 24, has taken its KUAVO (夸父) humanoid from a lab project to nearly 1.5 billion yuan in Pre-IPO funding and an IPO filing.

In January 2025, a humanoid robot named KUAVO (夸父) became the "No. 0 torchbearer" at China's National Games, the first 5G-A humanoid to carry a flame. A few months later, at Mobile World Congress in Barcelona, Spain's King Felipe VI shook its hand. The company behind it, Leju Robotics (乐聚机器人), was founded less than ten years earlier by a then-24-year-old from a second-tier city in Shandong.

Leng Xiaokun (冷晓琨) was born in 1992 in Weifang, Shandong. As a student he competed in robotics competitions and was admitted to the Harbin Institute of Technology (哈尔滨工业大学, HIT), where he earned bachelor's, master's, and doctoral degrees and joined the university's robot team. In 2016, with a handful of HIT classmates, he moved south to Shenzhen and started Leju — at a time when humanoid robotics was still a cold, unfashionable niche.

## From a small robot that paid the bills

Leng's first move was deliberately modest. Rather than chase a full-size humanoid immediately, he built Aelos (AELOS), a small bipedal education robot launched in 2017. It was cheap enough for schools and gave the company real revenue. In 2018, Aelos appeared in the "Beijing 8 Minutes" segment at the PyeongChang Winter Olympics closing ceremony, a rare global stage for a Chinese startup robot.

That "small first, full-size later" strategy bought time. Leju used the income to fund the harder project: KUAVO (夸父), a full-size humanoid. The first generation began development in 2018 with almost all core parts imported and localization below 10%. By the third generation, launched in December 2023, localization had passed 90%, and the robot could walk on varied terrain and jump.

## The Huawei and 5G turns

Two partnerships changed KUAVO's trajectory. In June 2024, Leju equipped KUAVO with OpenHarmony (开源鸿蒙, KaihongOS), making it one of the first large humanoids on an open-source operating system. Months later it integrated Huawei Cloud's Pangu embodied-intelligence model (盘古具身智能大模型), giving the robot a cloud "brain" for task planning. At MWC 2025, Leju, China Mobile (中国移动), and Huawei jointly showed a 5G-A humanoid that can locate itself precisely across large spaces and be controlled remotely in real time.

Industrial customers came next. KUAVO now works on flexible manufacturing and inspection tasks at factories including FAW Hongqi (一汽红旗), NIO (蔚来), and BAIC Off-road (北汽越野). In January 2025, Leju delivered its 100th full-size humanoid to BAIC — a milestone the company says made it the Chinese humanoid vendor with the most deliveries at the time.

## Heading to the public market

The business is scaling fast. Per the company's prospectus (招股书), reported by Southern+, Leju's KUAVO series generated about 178 million yuan (≈ US$24.7M / HK$194M) in sales in 2025, roughly 69% of that year's revenue and about twelve times the prior year. Full-year 2024 revenue was about 83.65 million yuan (≈ US$11.6M / HK$91M). According to IDC-cited market data, Leju ranked third globally in humanoid shipments in 2025 and second in full-size bipedal shipments.

In July 2025, Leju closed a Pre-IPO round of nearly 1.5 billion yuan (≈ US$210M / HK$1.63B), one of the largest single financings in China's humanoid sector that year; the post-money valuation was about 4.327 billion yuan (≈ US$600M). The company has since filed for IPO guidance with the Shenzhen regulator, with Orient Securities as sponsor. Leju has completed six funding rounds in total, backed by Shenzhen Capital Group (深创投), Tencent (腾讯), and others.

Leng has said the cost of a single robot has fallen from about 3 million yuan to several hundred thousand, and localization is now around 90%. He has received China's Youth May 4th Medal (中国青年五四奖章), and pitches an open-source ecosystem — Leju runs training centers such as a 1,500-square-meter facility in Suzhou that produces about 6,000 data entries a day — to feed real-world data back into models.

## What readers can do now

- If you are in manufacturing, watch Leju's deployments at FAW, NIO, and BAIC as a real test of whether humanoids can do flexible, non-repetitive tasks — not just dances.

- If you invest, separate demo hype from delivery counts: Leju's "100th robot" and IDC shipment rank are more meaningful than stage shows.

- If you study robotics, Leju's open-source and training-center model is a concrete example of how Chinese firms are trying to solve the real-world-data bottleneck.

## Honest limitations

This article is based on Leju's official site, Xinhua and Shenzhen government reporting, IDC-cited shipment figures, and the company's prospectus as reported by Southern+. The financial figures (2024 revenue, 2025 KUAVO sales, valuation) come from the prospectus via media and should be treated as preliminary pre-IPO disclosures subject to revision. Exact per-robot cost and localization percentages are company statements, not independently audited here. Competition is intensifying: industry reports note Unitree (宇树科技) and AgiBot (智元机器人) together may account for a large share of China's 2026 humanoid shipments, so Leju's current rank could shift quickly.

## Key takeaways

- Leng Xiaokun founded Leju at 24 and built KUAVO from <10% to >90% component localization in about five years.

- Partnerships with Huawei (OpenHarmony + Pangu) and China Mobile (5G-A) turned a lab humanoid into an industrial and public-facing product.

- With nearly 1.5 billion yuan (≈ US$210M) in Pre-IPO funding and an IPO filing, Leju is one of the Chinese humanoid players closest to the public market.

---

Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-people-leng-xiaokun-leju-robotics
Free to quote with attribution and a link to the original.
