--- title: "A robot-brain startup raises ≈¥3 billion in a month as Lei Jun and Jack Ma team up" date: 2026-10-24 category: Humanoids & Robotics site: NeuroAI canonical: https://neuroai.site/a/na-robot-xsquare-brain-funding language: en --- # A robot-brain startup raises ≈¥3 billion in a month as Lei Jun and Jack Ma team up > X Square Robot (千寻智能) closed a ¥1 billion round co-led by Lei Jun's Shunwei and Jack Ma's Yunfeng in April 2026 — part of roughly ¥3 billion raised in about a month — betting that the scarce resource is real-world data, not hardware. A robot making coffee in a JD Mall store in Hangzhou is not the headline. The headline is that the same machine keeps getting better at it — not because an engineer reprogrammed it overnight, but because it is quietly collecting the one resource the whole industry says it cannot get enough of: real-world experience. ## Key takeaways - X Square Robot (千寻智能) closed a ¥1 billion (≈ US$140M / HK$1.1B) round on April 7, 2026, co-led by Shunwei Capital (顺为资本, Lei Jun) and Yunfeng Fund (云锋基金, Jack Ma) — reported as the first time the two have jointly backed an embodied-AI company. - That round was the capstone of roughly ¥3 billion (≈ US$420M / HK$3.3B) raised across multiple closings in about a month, pushing the startup's valuation above ¥10 billion. - X Square positions itself as a builder of the robot "general brain" (通用大脑) — a software-first player in a field crowded with hardware makers. - Its core asset is claimed to be data: over 200,000 hours of real interaction collected, with a target of 1,000,000 hours in 2026. ## The round, and who showed up On April 7, 2026, X Square announced a ¥1 billion (≈ US$140M / HK$1.1B) financing co-led by Shunwei Capital (顺为资本) and Yunfeng Fund (云锋基金). Securities Times reported this as the first occasion on which Lei Jun and Jack Ma's vehicles jointly backed an embodied-intelligence startup. The round followed a nearly ¥2 billion (≈ US$280M / HK$2.2B) cluster of closings in February–March 2026, bringing the roughly month-long total to about ¥3 billion (≈ US$420M / HK$3.3B) and lifting the company past a ¥10 billion valuation — a so-called "百亿独角兽." The investor roster reads like a map of Chinese tech power. JD led a nearly ¥600 million (≈ US$85M / HK$660M) Pre-A+ round in July 2025, and a ¥528 million (≈ US$74M / HK$580M) Pre-A round in March 2025 was led by Prosperity7 Ventures (P7), the growth fund of Aramco Ventures. Securities Times reports the cap table has since drawn in Huawei-affiliated funds, the China Internet Investment Fund, Shunwei, Yunfeng and HongShan — a mix of internet, industrial and state-linked capital. The throughline is strategic, not just financial: these are precisely the kinds of companies that own retail floors, delivery fleets and cloud infrastructure where robots could actually earn their keep. ## Why a "brain" instead of a body Where peers like Unitree or Agibot race to ship metal, X Square calls itself a builder of the robot "general brain" (通用大脑) and stays deliberately light on hardware manufacturing. The thesis: the bottleneck to a robot that can work is no longer the actuators — it is scarce, expensive, hard-to-scale real-world data. The company says it has accumulated more than 200,000 hours of multi-type real interaction data (internet video, teleoperation, wearable capture) and plans to pass 1,000,000 hours in 2026. Its self-developed wearable data-collection gear is reportedly on its fifth generation and has cut capture cost to about one-tenth of traditional methods. On the model side, X Square cites a fusion of VLA (视觉—语言—动作, vision-language-action) and a world model (世界模型); Securities Times noted a recent global embodied-intelligence benchmark in which the company's model ranked first, though that claim originates from the company and the publication, not an independent lab. ## Where the robots actually work X Square's "Moz" (墨子) robot has been placed in JD Mall smart-retail settings for coffee-making, product explanation and demonstration — a live loop where commercial use generates the data that improves the model. Late in 2025, a "Xiao Mo" (小墨) unit was deployed on a CATL (宁德时代) battery line at its Zhongzhou base, handling end-of-line and DC-internal-resistance high-voltage test plug insertion. Those are unglamorous, repeatable tasks, which is exactly the point: boring work produces dense, clean training data. ## The bet and its risks The data-flywheel story is attractive and logically consistent: put robots in real sites, harvest experience, improve the model, deploy again. It also explains why retail and battery factories — not living rooms — are the first beaches. But the strategy lives or dies on whether the collected data is diverse and clean enough to generalize, and on whether the company can keep funding the deployment burn until the model advantage is real and defensible. ## Honest limitations Funding figures and the "first joint Lei Jun–Jack Ma embodied bet" come from Securities Times reporting and company statements, not from audited filings (X Square is private). The ¥3 billion / month total aggregates rounds from February–April 2026 and depends on how those windows are counted; I have described it as "roughly" for that reason. The benchmark "first place" claim is company-cited, not independently verified. Valuation figures are press-reported. Deployment descriptions (JD Mall, CATL) are company announcements; I have not confirmed unit counts or uptime in the field. No forward-looking revenue or profitability estimate is included. ## What readers can do now - If you follow embodied intelligence (具身智能), track the "data flywheel" race between model-first players like X Square and body-first players — the winner may be decided by who owns the dullest, highest-volume real workflows. - For enterprises piloting robots, insist on a data-return clause and a pilot-before-purchase plan; X Square's retail and battery-line deployments are useful templates for what "real" looks like versus a trade-show demo. - For observers of Chinese tech capital, the convergence of Lei Jun's and Jack Ma's vehicles on one startup is a cleaner signal of where strategic money expects robots to earn than any single product launch. --- Published by NeuroAI (https://neuroai.site/) — https://neuroai.site/a/na-robot-xsquare-brain-funding Free to quote with attribution and a link to the original.