Just Registered and Becomes a Unicorn, Embodied Startup Kunlunxing Raises Billions in 90 Days | 36Kr Exclusive 刚注册就成独角兽，具身创企昆仑行90天融资数十亿元｜36氪独家

TL;DR

• Former Alibaba Cloud president Ren Geng launches humanoid robotics startup Kunlunxing.
• Company achieved unicorn valuation (> $1B) within 90 days of incorporation.
• Three completed funding rounds total "tens of billions" of yuan.
• Top-tier investors like Hillhouse Capital made continuous follow-on investments.
• Kunlunxing targets a "body + brain" integrated strategy to rival Tesla's Optimus.

Key Data

Deep Analysis

The birth of Kunlunxing reads less like a startup launch and more like a meticulously orchestrated capital symphony. A valuation cresting $1 billion before the company even finishes its registration? That's not venture capital; that's a bidding war for pedigree. The market is starving for a credible champion in the humanoid race, and it's throwing money at the résumé instead of the product.

The duo of Ren Geng and Lang Xianpeng is being sold as a "rare combination," and it is—for the wrong reasons. It's a classic marriage of a top-tier commercial operator (Ren, the cloud infrastructure and enterprise sales maestro) and a celebrated technical lead (Lang, the autonomous driving prodigy). Capital loves this combo because it ticks the boxes: scalability and innovation. But it's a template for a consulting firm, not a deep-tech hardware company. The real test isn't whether they can build an organization that scales, but whether they can solve the fundamental engineering and materials science problems that keep humanoid robots as science fair prototypes.

Let's talk strategy. The "body + brain" dual-drive, full-stack self-research mantra is the current vogue. It sounds comprehensive, but it's also the most capital-intensive and complex path possible. They're not just building a robot; they're attempting to vertically integrate from fundamental AI models (the Kunlun World Model) down to custom actuators and sensor fusion. The comparison to Tesla is telling. Tesla has a multi-decade head start in manufacturing, supply chain, and massive AI compute infrastructure. For Kunlunxing to claim parity based on a funding round is audacious, bordering on fantasy. The brutal reality is in the components: a dexterous hand that fails in two months is not a supply chain problem, it's a fundamental design failure that capital cannot magically solve.

What's truly happening here is the financialization of a hype cycle. The embodied intelligence sector, jolted by Tesla's Optimus and the IPO of Unitree, is experiencing a classic speculative bubble. Investors, fearing they'll miss the "next big thing," are stampeding into anything with a impressive logo and a credible founding story. Kunlunxing isn't being valued on its technology, which is nascent, but on the exit potential of its founders' networks and the sheer size of the theoretical addressable market ($5 trillion by 2050). This is venture capital as momentum trading.

The critical flaw in the narrative is the leap from automotive autonomy to general-purpose robotics. While the perception-decision-action pipeline shares DNA, the embodiment is radically different. A car operates on structured roads with 2D planning; a robot must navigate unstructured, 3D human environments with manipulation as a core function. Lang's "lateral entry" success in ADAS is impressive, but it doesn't automatically translate to solving robotic grasping, balance, or material fatigue. The team is being judged on past performance in a adjacent field, which is a risky proxy for future performance in a harder one.

Ultimately, Kunlunxing represents the peak of the "dream team" investment thesis. It has an impeccable team sheet, a war chest, and a target on the biggest prize in physical AI. But it enters an arena littered with decades of failed attempts and colossal incumbents. Its success hinges not on another funding round, but on a breakthrough in robotics that eluded some of the world's best-funded labs. The hype is deafening, but the sound of a humanoid robot reliably performing a simple household task is still silent.

Industry Insights

1. Hardware Will Be the Bottleneck: Startups focused solely on AI "brains" will fail. The winners must co-design software with novel, durable hardware, driving M&A in component manufacturing.
2. Capital Concentration will Sharpen: The market is bifurcating. A few "star teams" will absorb the majority of funding, forcing smaller players to specialize in niche applications or become subcontractors.
3. The "Hybrid CEO" Model is Emerging: The most sought-after founding teams pair a proven enterprise scaling executive with a deep-tech CTO, mirroring the Kunlunxing blueprint to de-risk both business and technology for investors.

FAQ

Q: How is Kunlunxing different from other humanoid robotics startups?
A: Its primary differentiation is the exceptionally deep commercial and operational experience of its founders from Alibaba Cloud and Li Auto, which capital believes will accelerate commercialization beyond pure R&D teams.

Q: Why was it able to secure funding so rapidly?
A: The startup leveraged the extreme hype in the embodied AI sector and the star power of its founders. Investors, fearing missing out, prioritized team credibility over technological maturity in a scarce market of "investable" projects.

Q: What is the biggest challenge facing Kunlunxing?
A: Moving beyond prototypes to mass production. This requires solving fundamental hardware reliability issues (like component lifespan) and building a robust supply chain, which is a far greater challenge than securing initial funding.

TL;DR

• 前阿里云中国区总裁任庚创立昆仑行机器人，切入具身智能赛道。
• 公司注册不到90天完成3轮融资，累计数十亿元，估值超10亿美元成独角兽。
• 创始团队由任庚（商业运营）与郎咸朋（理想智驾技术负责人）组成“稀缺组合”。
• 公司战略对标特斯拉Optimus，坚持“本体+大脑”双轮驱动与全栈自研。
• 资本追捧源于团队过往成功经验及对具身赛道“好标的稀缺”的共识。

核心数据

深度解读

昆仑行的闪电式崛起，像一剂强心针，扎进了本就燥热的具身智能赛道。90天，10亿美元，这些数字勾勒出的不是一家公司的成长轨迹，而是一场资本在技术迷雾中的集体豪赌，以及对中国科技创业“路径依赖”的深刻讽刺。

资本追捧的，究竟是“机器人”还是“成功学”？ 投资机构们坦诚，“主要看团队是否够强”。这个“强”，指的是任庚带领阿里云做到千亿营收的商业战绩，是郎咸朋用千万级预算撬动百万辆智驾交付的“性价比奇迹”。这本质上是在为过去的成功支付溢价，赌他们能将上一场战争的胜利经验，平移到一个规则截然不同的新战场。但云计算市场的规模化、标准化打法，与需要极致硬件工程、场景深耕和漫长用户教育的机器人产业，存在本质鸿沟。资本押注的，更像是一套可复制的“操盘手方法论”，而非对机器人技术本身深刻理解的共识。这暴露了当下投资市场的某种焦虑：在看得懂的商业模式和看不懂的硬科技之间，他们选择了前者，并用估值来掩盖对后者的不确定性。

“对标特斯拉”的叙事，是勇气还是幻觉？ 昆仑行将标杆定为特斯拉Optimus，这堪称一场“品牌升维”的公关妙棋。但现实是骨感的。特斯拉拥有全球顶尖的制造体系、海量的真实数据闭环、以及马斯克个人IP带来的无限试错资本。昆仑行凭什么对标？凭的是“本体+大脑”全栈自研的战略PPT吗？当特斯拉用全球工厂数据训练模型时，昆仑行的数据从哪来？当灵巧手寿命还以“月”计时，谈论通用人形机器人的“终局形态”是否为时过早？这种对标，更像是在资本寒冬中，用一个宏大的叙事框架，来快速确立赛道卡位和估值锚点。它成功地吸引了眼球，但也可能让公司过早背负上不切实际的期望，忽略了从专用场景切入、积累真实世界数据的笨功夫。

“商业派+技术派”的组合，是完美互补还是潜在裂痕？ 文章极力渲染任庚的“端到端整体操盘能力”与郎咸朋的“AI技术大脑”属性，描绘了一幅珠联璧合的蓝图。但在中国的科技创业史上，商业领袖与技术核心因理念、节奏、资源分配产生冲突的案例比比皆是。任庚擅长的是体系搭建、资源运营和规模化商业落地，这要求清晰、可控、可预测；而前沿AI和机器人研发，本质是探索未知、容忍失败、迭代试错。当资本要求快速看到“可落地应用场景”和“商业闭环”时，是坚持长期技术攻坚，还是屈从于短期项目变现？这个组合的韧性，将在公司的第一个生存危机中面临真正的考验。

总而言之，昆仑行的故事，是时代情绪的缩影：技术理想主义在资本助推下的急速膨胀。它拥有一个梦幻开局，但其最终价值，不取决于它融了多少钱，喊了多大的口号，而在于它能否在特斯拉的阴影下，走出一条属于自己的、扎扎实实的从“演示级”到“量产级”的爬坡之路。这需要的不只是操盘手的运筹帷幄，更是工程师面对物理定律时，那份最原始的谦卑与坚韧。

行业启示

1. 投资逻辑需分化：对具身智能的投资，应从“赌明星团队”转向更细分地评估其在硬件工程、场景数据获取、成本控制等具体环节的真实壁垒。
2. 警惕“估值前置”风险：资本催熟下，企业估值严重领先于产品里程碑，可能导致团队动作变形，应更关注其首个可商用场景的落地时间表与可行性。
3. 构建“T型能力”结构：成功的具身智能公司可能需要“垂直技术深度（如运控、传感）”与“横向系统整合（软硬件、供应链）”相结合的T型能力，而非单一维度的“全栈”口号。

FAQ

Q: 为什么昆仑行能如此快速成为独角兽？
A: 核心在于“热钱寻找出口”与“强团队信用背书”的结合。具身智能赛道火热，优质标的稀缺，而任庚和郎咸朋拥有已验证的、堪称“稀缺”的跨领域成功履历，极大降低了投资人的信任成本，导致份额被哄抢。

Q: 从智能驾驶跨界到机器人，技术真的能直接迁移吗？
A: 理论框架（感知-决策-执行）高度同源，但实践挑战巨大。智驾的环境相对结构化（道路），而机器人在非结构化环境中操作，对物理世界的理解、柔性控制和安全性要求呈指数级增长。郎咸朋的经验更多是在“研发体系”和“成本控制”上，底层技术仍需大量创新。

Q: 这家公司面临的最大风险是什么？
A: 最大的风险是“愿景与落地能力的脱节”。对标特斯拉、押注通用人形机器人是极高难度的目标。如果无法在1-2年内在某个垂直场景（如工业、物流）证明其硬件可靠性和算法有效性，并跑通小规模商业模式，那么前期靠叙事支撑的高估值将难以为继，面临巨大的资本和市场压力。