广告
加载中

谁在豪赌具身智能?

定焦One团队 2026-06-18 22:59
定焦One团队 2026/06/18 22:59

邦小白快读

EN
全文速览

本文梳理了2025年7月到2026年6月国内具身智能领域的融资发展情况,核心干货信息如下:

1. 整体融资热度远超预期,一年内国内一级市场融资达503起,总金额超960亿元,单笔融资额不断提升,诞生了它石智航单轮4.55亿美元的国内最高融资纪录。

2. 当前行业融资呈现三大趋势:资金总量增长但融资项目数量减少,资金不断向头部集中,腰部及以下玩家加速出局,估值逻辑从过去看团队背景、愿景,转变为看复购率、故障率等实实在在的运营数据。

3. 目前资本主要押注全栈能力整机、具身大脑、数据采集三大方向,财务VC、互联网大厂、制造业资本、国资、早期专业机构四类玩家均已下场,不过当前行业刚需场景仍不明确,商业化落地仍待验证。

当前具身智能行业处于资本热潮期,产业和市场层面出现了很多新变化,对品牌商布局相关业务的参考干货如下:

1. 产业与消费趋势:受老龄化劳动力短缺、制造业回流的推动,中美都将发展具身智能作为核心方向,2026年被行业视作机器人量产元年,商业化落地已经从愿景变成明确预期,多家头部企业即将登陆公开市场,行业进入关键转折期。

2. 产品研发方向:市场当前偏好具备“硬件本体+智能算法”全栈自研能力的项目,行业核心缺口是足够智能的具身大脑,高质量真实交互训练数据十分稀缺;优先做透细分专用场景、做实投入产出比的项目更易获得市场认可。

3. 竞争提示:当前资金高度向头部集中,百亿估值已经成为进入主流融资圈的入场券,只会讲故事没有实际落地数据的玩家会被快速淘汰,品牌商布局需要聚焦核心能力,拿出真实运营数据才能获得资本和市场认可。

近一年国内具身智能赛道资本大量涌入,赛道格局快速变化,带来了新的机会与需要警惕的风险,核心干货整理如下:

1. 机会层面:当前整机、具身大脑、数据采集三大细分方向持续升温,是资本疯抢的核心增长赛道;财务VC、产业资本、国资、早期专业机构四类玩家全部下场,产业资本除资金外还能提供场景、供应链等资源,具备硬数据的头部项目更容易获得资本支持。

2. 风险提示:关节模组、传感器等细分赛道已经降温,腰部及以下玩家正在悄然离场;行业融资门槛大幅抬高,百亿估值成为主流融资圈入场券,资金高度向头部集中,估值逻辑已经从讲概念转向看硬数据,没有实际落地成果的项目很难获得融资。

3. 方向建议:优先聚焦细分专用场景,把单一场景做透做实投入产出比,更容易获得资本和市场认可;通用路线虽然更受资本追捧,但目前刚需场景仍不明确,商业化存在较大不确定性。

具身智能行业的快速发展,给制造工厂带来了新的商业机会和数字化转型启示,核心干货整理如下:

1. 产品生产与设计需求:2026年被视作机器人量产元年,行业对机器人本体、核心零部件的需求将持续增长;当前市场更需要满足工业特定场景,比如工业拣选、物流搬运的专用机器人,对机器人的连续运行时长、故障率、复购率等硬指标有明确要求。

2. 商业机会:不少制造业头部企业已经下场布局具身智能,将其作为主业的新增量,工厂可以依托自身对产线痛点的深入了解,要么切入具身智能本体、零部件生产领域,要么对接具身智能创业项目,凭借自身供应链、场景优势获得产业资本的支持,挖掘新增长空间。

3. 转型启示:具身智能是下一代制造业的核心基础设施,工厂可以提前对接布局具身智能技术,推进自身生产的智能化升级,应对劳动力短缺的行业问题,抓住制造业智能化升级的风口。

当前具身智能行业处于资本热潮期,行业发展呈现明确的新趋势,也浮现出大量待解决的痛点,给相关服务商带来了新机会,核心干货整理如下:

1. 行业发展趋势:近一年国内具身智能领域融资总金额超960亿元,资本大量涌入,行业即将进入量产元年,商业化落地进程加速;资金不断向头部项目和全栈整机、具身大脑、数据采集三大主线集中,不同背景的资本纷纷入局,行业规模快速扩张。

2. 核心客户痛点:当前具身智能企业最核心的痛点是缺乏高质量的真实交互训练数据,训练具身大模型的数据采集成本极高;其次腰部非头部项目融资难度大幅提升,普遍缺乏资本和产业资源的对接渠道;估值逻辑重构后,企业需要符合资本要求的硬运营数据才能获得融资,不少创业企业缺乏相关梳理能力。

3. 可布局方向:可以切入数据采集、数据标注处理赛道,为具身大脑研发企业提供高质量训练数据服务;也可以为创业企业提供资本对接、商业化咨询服务,帮助企业对接产业资本和落地场景,匹配资本的要求。

具身智能行业的快速发展,给相关产业平台、创投平台带来了新的需求和变化,核心干货整理如下:

1. 行业对平台的新需求:当前具身智能行业处于早期分化阶段,创业企业需要对接资本、产业资源、落地场景,各类投资机构也需要平台筛选优质标的、对接行业资源,平台的对接服务需求明显增长。

2. 风向规避要点:当前行业资金高度向头部集中,腰部及以下项目加速出清,关节模组等早期热门细分赛道已经降温,平台招商需要优先筛选具备硬运营数据、核心技术壁垒、全栈能力的项目,避开只有概念没有实际落地成果的标的。

3. 运营优化方向:可以围绕全栈整机、具身大脑、数据采集三大热门赛道打造专项招商和服务体系,对接四类资本的不同投资需求:针对国资的产业培育需求,对接符合地方战略布局的项目;针对产业资本的需求,对接能匹配主业增量的创业项目,提升平台的核心竞争力。

近一年国内具身智能行业融资高速增长,出现了很多新的产业动向和待研究的新问题,核心内容整理如下:

1. 产业新动向:2025年7月到2026年6月,国内一级市场具身智能领域融资达503起,总金额超960亿元,资本大量涌入;行业呈现三大新特征:资金总量扩容但向少数头部标的集中,行业门槛抬高腰部玩家加速出清,估值逻辑从看团队故事转向看复购率、故障率等硬运营数据;资本重点押注全栈整机、具身大脑、数据采集三大方向,财务VC、产业资本、国资、早期专业机构四类玩家纷纷下场,国资已经成为不可忽视的核心投资力量。

2. 待研究的新问题:当前具身智能估值存在逻辑矛盾,传统静态估值框架无法适配具身智能价值复利增长的动态特性,投资人陷入兼顾硬数据和动态增长价值的两难;行业目前真正的刚需落地场景仍然不够明确,多数头部产品仍偏科研展示,技术路线上通用路线和专用路线仍存在分歧,未来发展路径尚不清晰。

3. 商业模式层面:目前行业仍未跑出成熟可复制的商业化模式,最终市场规模和发展方向仍需要企业探索验证,是值得长期跟踪研究的新兴领域。

返回默认

声明:快读内容全程由AI生成,请注意甄别信息。如您发现问题,请发送邮件至 run@ebrun.com 。

我是 品牌商 卖家 工厂 服务商 平台商 研究者 帮我再读一遍。

Quick Summary

This article summarizes financing developments in China's embodied intelligence sector from July 2025 to June 2026, with key takeaways as follows:

1. Overall financing heat has far exceeded expectations. Over the 12-month period, China's primary market saw 503 financing deals worth a total of over ¥96 billion, with average deal size growing steadily. The sector set a new domestic single-round financing record of $455 million for Tashi Zhihang.

2. The industry currently presents three core trends: total capital deployed has grown while the number of funded projects has shrunk, as capital increasingly concentrates in top-tier players and lower-tier and mid-tier players are exiting at an accelerating pace. Valuation logic has also shifted: instead of focusing on team background and long-term vision, investors now prioritize tangible operational metrics such as repurchase rates and failure rates.

3. Capital is currently betting primarily on three areas: full-stack integrated robots, embodied brains, and data collection. Four types of investors—financial VC firms, large internet companies, manufacturing capital, state-owned assets, and early-stage specialized institutions—have all entered the space. However, clear mass-market刚需 use cases have not yet emerged, and commercialization remains unproven.

The embodied intelligence sector is currently experiencing a capital boom, with many new developments taking shape across industry and markets. Below are key insights for brands looking to enter the space:

1. Industrial and consumer trends: Driven by aging populations and labor shortages, as well as manufacturing reshoring, both China and the U.S. have identified embodied intelligence as a core strategic priority. The industry views 2026 as the first year of mass robot production, with commercialization shifting from a long-term vision to a clear near-term expectation. Multiple leading players are preparing for public listings, putting the industry at a critical inflection point.

2. Product R&D priorities: The market currently favors projects with full-stack in-house capabilities covering both hardware body and intelligent algorithms. The industry's core gap is sufficiently capable embodied brains, and high-quality real-world interaction training data is extremely scarce. Projects that prioritize deep penetration into niche use cases and deliver solid return on investment are far more likely to win market acceptance.

3. Competitive notes: Capital is now heavily concentrated in top players, and a RMB 10 billion valuation has become the minimum entry ticket for the mainstream financing circle. Players that only pitch narratives without tangible on-the-ground data are being淘汰 rapidly. Brands looking to enter the sector need to focus on core capabilities and deliver real operational data to win support from both capital and the market.

Massive capital has flowed into China's embodied intelligence track over the past year, reshaping the sector landscape rapidly and bringing both new opportunities and notable risks. Key takeaways are below:

1. Opportunities: Three core segments—full-stack integrated robots, embodied brains, and data collection—have continued to heat up, and are the core growth tracks that capital is aggressively chasing. Four types of investors—financial VCs, industrial capital, state-owned assets, and specialized early-stage institutions—have all entered the space. Beyond capital, industrial capital can also provide access to use cases and supply chain resources, and top projects with solid operational data are far more likely to secure capital backing.

2. Risk warnings: Segments such as joint modules and sensors have already cooled, and mid-tier and lower-tier players are quietly exiting the market. Financing thresholds have risen sharply, with a RMB 10 billion valuation becoming the entry ticket for mainstream financing. Capital is heavily concentrated in top players, and valuation logic has shifted from concept pitching to hard performance data, making it extremely difficult for projects without tangible commercial progress to secure funding.

3. Strategic recommendations: Prioritize focusing on niche, specialized use cases, and achieve deep penetration with solid return on investment, which will make it far easier to win recognition from capital and the market. While general-purpose embodied intelligence is more popular among investors, clear刚需 use cases have not yet emerged, leaving commercialization with significant uncertainty.

The rapid growth of the embodied intelligence sector has brought new commercial opportunities and digital transformation insights for manufacturing factories. Key takeaways are as follows:

1. Product design and manufacturing demand: 2026 is widely seen as the first year of mass robot production, and industry demand for robot bodies and core components will continue growing. The market currently has strong demand for specialized robots tailored to specific industrial scenarios, such as industrial picking and logistics handling, with clear requirements for hard performance metrics including continuous runtime, failure rate, and repurchase rate.

2. Commercial opportunities: Many leading manufacturing enterprises have already entered the embodied intelligence space, positioning it as a new growth driver for their core businesses. Factories can leverage their in-depth understanding of production line pain points to either enter the robot body or core component manufacturing segments, or partner with embodied intelligence startups, leveraging their supply chain and on-the-ground scenario advantages to win industrial capital support and unlock new growth.

3. Transformation insights: Embodied intelligence will serve as core infrastructure for the next generation of manufacturing. Factories can proactively partner with and布局 embodied intelligence technology to推进 the intelligent upgrade of their own production, address industry-wide labor shortages, and capture the wind of intelligent manufacturing upgrade.

The embodied intelligence sector is currently in a period of booming capital inflows, with clear new industry trends and a large number of unaddressed pain points, creating new opportunities for relevant service providers. Key takeaways are as follows:

1. Industry development trends: Over the past year, total financing in China's embodied intelligence sector exceeded ¥96 billion, with massive capital inflows. The sector is about to enter its first year of mass production, with commercialization accelerating. Capital is increasingly concentrating on top projects and three core tracks: full-stack integrated robots, embodied brains, and data collection. Investors with diverse backgrounds have all entered the space, driving rapid industry expansion.

2. Core customer pain points: The most critical pain point for embodied intelligence companies is the shortage of high-quality real-world interaction training data, as data collection for large embodied models carries extremely high costs. Second, mid-tier non-leading projects face sharply higher financing difficulty, and generally lack access to connection channels for capital and industrial resources. After the valuation logic restructuring, companies need hard operational data that meets investor requirements to secure financing, and many startups lack the capability to organize and present this data.

3. Tracks worth entering: Service providers can enter the data collection and data annotation processing segment to provide high-quality training data services for embodied brain R&D companies. They can also provide capital connection and commercialization consulting services for startups, helping these companies connect with industrial capital and commercial落地 scenarios, and meet the requirements of investors.

The rapid development of the embodied intelligence sector has brought new demand and changes for relevant industrial platforms and venture capital platforms. Key insights are as follows:

1. New industry demand for platforms: The embodied intelligence sector is currently in an early stage of market differentiation. Startups need support connecting with capital, industrial resources and commercial落地 scenarios, while investment institutions need platforms to screen high-quality targets and connect with industry resources. Demand for platform connection services has grown notably.

2. Risk avoidance guidance: Capital is currently heavily concentrated in top projects, with mid-tier and lower-tier projects exiting at an accelerating pace. Early hot segments such as joint modules have already cooled. When sourcing projects, platforms should prioritize screening projects with solid operational data, core technical barriers, and full-stack capabilities, and avoid targets that only offer concepts without tangible commercial progress.

3. Operation optimization directions: Platforms can build specialized investment recruitment and service systems focused on the three hot tracks: full-stack integrated robots, embodied brains, and data collection. They can also match the different investment needs of the four major types of capital: for state-owned investors with industrial cultivation goals, connect projects that align with local strategic layout; for industrial capital, connect startups that can deliver new growth to their core businesses, to enhance the platform's core competitiveness.

Over the past year, financing for China's embodied intelligence sector has grown rapidly, bringing many new industry developments and new questions for research. Key insights are as follows:

1. New industry developments: From July 2025 to June 2026, China's primary market recorded 503 financing deals in embodied intelligence, with total deal volume exceeding ¥96 billion, driven by massive capital inflows. The sector now shows three new characteristics: total capital has expanded but is increasingly concentrated in a small number of top targets; industry barriers have risen and mid-tier players are exiting at an accelerating pace; valuation logic has shifted from team narratives to hard operational metrics such as repurchase rate and failure rate. Capital is primarily focused on three tracks: full-stack integrated robots, embodied brains, and data collection. Four types of investors—financial VC, industrial capital, state-owned assets, and specialized early-stage institutions—have all entered the space, with state-owned capital now becoming an core investment force that cannot be ignored.

2. New research questions: There is currently a logical contradiction in embodied intelligence valuation. Traditional static valuation frameworks cannot adapt to the dynamic compound growth characteristics of embodied intelligence value, leaving investors caught in the dilemma of balancing hard data and dynamic growth value. The sector still lacks clearly defined刚需 commercial落地 scenarios, most leading products are still focused on research and demonstration, and there is still disagreement over whether general-purpose or specialized technology routes are superior, leaving the future development path unclear.

3. Business model observations: The sector has not yet produced a mature and replicable commercial model, and the final market size and development direction still need to be explored and verified by market participants. It is an emerging field that is worthy of long-term tracking and research.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

一年融资超500起,整机、大脑、数据采集被疯抢。

定焦One(dingjiaoone)原创

作者 | 王璐

编辑 | 魏佳

具身智能的融资速度,比所有人预期的都更猛、更快。

据「定焦One」从IT桔子获得的数据显示,从2025年7月到2026年6月,国内一级市场(不包含IPO与并购)具身智能领域的融资已达503起,平均每天超过1起,总融资金额超960亿元。

与此同时,单笔金额也越来越高。单笔10亿元及以上的融资在2026年上半年已超过25起,其中它石智航在4月单轮拿下4.55亿美元,刷新中国具身智能融资纪录,这个规模在全球机器人创业史上也鲜有前例。星尘智能则在三个月内连续完成3轮融资,总额超10亿元,估值突破百亿元。

但热潮之下,钱的流向并不平均。

整机、大脑、数据采集这几个方向持续升温,关键模组等细分赛道则在降温;头部公司融资一轮比一轮高,腰部及以下玩家在悄然离场;出手争抢头部项目的,也不只是财务VC,国家队、互联网巨头、产业资本都已下场。

钱流向哪里、谁在出手,是判断一个行业最直接的信号。梳理这一年具身智能的融资变化会发现,热钱仍在涌入,但资本对这条赛道的定价方式,已经和之前不同了,每一家公司都将接受更苛刻的审视。

01.钱涌进具身智能,但规则变了

回顾具身智能从2025年7月到2026年6月这一年多的融资节奏,「定焦One」与多位一线投资人交流后发现,大家的感受基本一致:钱越来越多,但要求越来越严格。上海浦昌股权投资基金合伙人王瀚用了三个词来概括这个周期里的变化:加速、分化、重构。

具体来看,这一年多的融资呈现出三个趋势。

第一个趋势是,钱在变多,项目却在变少。

2025年下半年,具身智能进入密集布局期。从整机到大脑,再到关节模组、灵巧手,整条产业链在集中融资,连触觉传感器、电子皮肤、谐波减速器、一体化关节等此前少有人关注的硬件细分品类,也几乎每隔几天就有一笔融资公告落地。

但有个反常之处是,这一轮的热潮是资金体量的快速扩张、向更少的标的集中,而非数量的全面爆发。据投中研究院统计,2025年下半年具身智能融资事件数量同比下降31.7%,与此同时,单笔融资额均值同比上涨46.8%。

驱动这波热潮的,是2026年被视作机器人量产元年的预期,大脑、本体、零部件等各环节都必须提前布局补齐,于是各路资金从天使轮到Pre-A轮密集入场。

英诺基金布局具身智能已有数年,松延动力、云深处、加速进化、鹿明、灵足等一批头部公司都在被投名单上。在英诺基金管理合伙人周全看来,这个市场足够大,未来能跑出一批千亿市值公司,“一边是老龄化与劳动力短缺,一边是制造业回流的压力,这是中美都必须发展的方向。”

到2026年第一季度,这轮热潮被推向顶峰。短短三个月内,具身智能领域完成210起融资,总额超300亿元,不少投资人甚至感慨“赛道已过热”。

但并不是所有环节都在同步加速。第二个趋势是,门槛被抬高,腰部玩家正在出局。

「定焦One」观察到,进入2026年后,关节模组、传感器等零部件赛道的新项目融资数量已开始回落。

长期关注硬件赛道的投资人刘明指出,这类赛道项目质量本就参差不齐,经过2025年下半年的集中布局后,真正值得持续跟进的公司已经不多。

王瀚也持相似观点:“2025年下半年那波入场,更像是一次集体卡位,大家都怕错过窗口期,但并不是每一笔钱都押对了方向。”

与此同时,留下来的钱进一步向头部集中。

以始终是融资热门的整机环节为例,据「定焦One」统计,2025年下半年至今,该领域的融资数量占整体的40%以上,但大额融资流向少数公司,其中银河通用25亿、自变量机器人近20亿、星动纪元超2亿美元,头部公司的融资规模一轮比一轮大。

王瀚对这种变化感触颇深。“2024年,一家公司只要沾上具身智能的概念,融资就相对容易,但从2025年下半年起,这扇门开始收窄。百亿估值逐渐成了进入主流融资圈的入场券,资金向头部集中的趋势越来越明显,雨露均沾的时代结束了。”

钱之所以更愿意往头部押,因为整个行业给具身智能估值的方式,正在被改写。这就要说到第三个趋势,具身智能的估值逻辑,从看故事变成了看数据。

过去,投资人更关注机器人创业团队的背景、技术路线和长远愿景。进入2026年,实实在在的数据指标成为估值基础。王瀚直言,他现在评估一家机器人公司,问的关键问题已变成“客户有没有复购?机器人能连续运行多少小时?故障率是多少?”他总结,这些问题过去可能只是尽调里的附加题,现在已经是决定投资的必答题。

转变的背后,是行业走到了关键转折期。一方面,2026年被普遍视为机器人量产元年,商业化落地从愿景变成明确预期;另一方面,宇树科技即将登陆科创板,云深处、乐聚机器人等已经递交招股书,当头部企业开始接受公开市场的审视,整个行业有了更明确的参照物。

但这里有一个矛盾,具身智能行业的公司估值逻辑,恰恰很难用传统的静态框架去衡量。

“过去评估一个项目,我们往往参照市场可比公司,或使用PE、PS等财务指标,结合销量、单价、渗透率等静态数据来推算其未来市场规模与估值。”王瀚说,“但具身智能的价值是动态的,今天看到的,和年底再评估的,可能完全是两回事。”

他举了个例子,“今天一个只会跳舞的机器人,随着数据积累与持续训练,明天可能突然掌握一项全新的智能,真正解决某个特定场景或人群的实际问题。这种能力的突破不是线性的,它的价值会随着时间和数据的积累,呈现复利式增长。”

这就把投资人推到了一个两难的位置:一边是公开市场要求看得见的数据,一边是机器人真正的价值还在动态变化。投资人能做的,就是一边用复购率、故障率这些硬指标筛掉只会讲故事的公司,一边为那些可能出现复利式增长的项目预留出想象空间。而最可能兼顾两者的,正是头部项目。

02.三大投资主线:整机、大脑、数据采集

从具体的细分赛道来看,这一年的钱主要押向了三个方向。

其中,整机仍是融资的主战场,但投资人的关注点变了。随着商业化落地成为核心要求,资金不再撒向广泛的技术尝试,转向“硬件本体+运动控制+智能算法”一体化的公司。此前,这一创业方向还相当分散,轮式、四足、小型化、工业专用等各种路线的公司各走各的路,到2026年,资金已明显向具备全栈能力的公司集中。

刘明解释,“机器人最终要落地,纯硬件没有大脑是空壳,纯算法没有本体是空中楼阁,只有两者都能自研的公司,才在商业化上拥有主动权。”星动纪元、自变量机器人、银河通用等头部玩家,也都将全栈自研作为核心竞争力之一。

以星尘智能为例,它最初凭借绳驱硬件技术形成了差异化优势,现在则将“AI模型-具身OS-绳驱本体”的全栈架构当成核心壁垒。它在继续发挥绳驱技术成本优势的同时,开始自研机器人的“大脑”与决策系统。它拿到的融资与订单,正是这种能力获得市场初步验证的体现。

如果说整机是具身智能的身体,那么具身大脑便是真正的天花板。不止一位从业者认为,机器人大脑一旦做成,对整个赛道的重塑,不亚于大语言模型对软件行业的冲击。

这种预期体现在融资金额上。近一年,大脑赛道的单笔融资额,在整个具身智能产业链里最高。

代表案例是它石智航,2026年4月,这家专注具身智能大脑方向的公司完成4.55亿美元(约30亿人民币)的Pre-A轮融资,由高瓴创投、红杉中国联合领投,美团战投等20余家机构参与,创下中国具身智能有史以来最高单轮融资纪录。此外,千寻智能两轮合计融资近30亿元、智平方B轮融资超10亿元。

王瀚将大脑赛道的走热,归结为一个水到渠成的过程,“过去几年,资本的扶持让做本体的公司把机器人身体问题基本解决了。但机器人迟迟无法大规模商业化,核心原因是大脑还不够聪明。”

在他看来,2026年最大的投资主题就是大脑,“这有点像在赌具身智能的iPhone时刻,当通用大脑能真正驱动不同形态的机器人完成各种任务,且效率足够高时,基本上就能替代人的工作。”

周全的布局路径,也是这种市场判断的缩影。早两年大家把钱主要押在硬件本体上,如今重心转移。“本体说到底还是制造业,竞争一定越来越激烈,毛利往下走。”他的策略因此分成两层:本体类项目格局已定;而真正资本价值更高的,是偏软的大脑端。今年英诺几乎完全转向世界模型,一口气投资了流形空间、VAST、厘清等等。

但具身大脑还存在路线上的分歧。押注通用路线的公司认为,只有做成通用,机器人才能适配不同本体和不同场景,护城河才足够宽;走专用路线的公司则认为,通用是愿景,落地才是现实,先把一个场景跑通,比什么都重要。

王瀚目前更倾向于专用路线,“机器人不需要什么都会,只需要把工业拣选、物流搬运某一件事做好、做透。把账算清楚,把ROI做实,才更容易判断它到底能不能真正走进工厂。”

但他同时也承认,市场重视通用路线有其逻辑,“本质上是在赌一个更大的未来,如果通用大脑这条路走通了,会是更加变革性的存在。”从融资结果看,市场目前将真金白银更多押注了通用路线。

具身大脑也是烧钱最快的方向。训练一个具身大模型,需要海量真实交互数据,这些数据无法通过互联网爬取,只能靠机器人在真实世界中不断操作来积累,成本极高。谁能率先建立起数据优势,谁才有机会跑出来。

正是这种对于高质量训练数据的迫切需求,把数据采集也推成了2026年投资人关注的新方向。要让机器人变得更聪明,当前最稀缺的已不是算法,而是真实、高质量、可用的训练数据。

一个信号是,智元孵化的数据平台觅蜂科技,在今年2月成立当月就完成数亿元种子轮与天使轮融资,由红杉中国领投,鼎晖VGC、BV百度风投、云锋基金、慕华科创跟投,产业方均普智能、灵初智能也参与其中。头部机器人公司将数据业务单独拆出来融资,本身就说明了这个环节的重要性。

三条主线看下来,一个共同的变化是,市场正在从具身智能行业的广撒网转向精准下注。而无论押在哪个环节,能够持续获得融资的公司,都需要回答其技术如何以及何时能变现的问题。

03.四类投资者,谁在豪赌?

随着市场对具身智能的关注度越来越高,参与这场游戏的除了传统的财务VC,还包含国资、产业资本、早期专业机构等不同背景的玩家,从它们下注方式的差异,一定程度上可以看出行业里还剩哪些机会、哪些窗口已经关闭。

财务VC仍是当前最活跃的参与者之一。

以红杉、高瓴、IDG、源码、经纬等机构为代表,这些在消费互联网时代成名的机构,在具身智能上的打法是“多环节覆盖”。同一家机构,可能同时出现在整机、大脑、灵巧手公司的股东名单里。比如红杉就参与了整机(自变量机器人、宇树科技)、大脑(它石智航)以及核心部件(灵猴机器人)等多个环节的投资。

王瀚解释,机器人赛道格局未定,没人知道其最终的技术路线和商业模式会走到哪里。在大方向判断正确的前提下,跨赛道广覆盖是降低失误成本的理性选择。

但要注意,这些机构在每一个细分方向上瞄准的,通常是被验证过的头部团队或成型产品,而非任意早期项目。

这与移动互联网时代有着本质区别。过去VC可以用较低成本广泛覆盖早期项目,等待优胜者跑出来再追投,但在具身智能领域,硬件属性决定了早期验证成本极高,具备技术实力的团队往往早已估值不菲,投资机构从一开始就必须精选目标。

第二类是产业资本。

产业资本分为两类,均偏向于战略投资。

一类是互联网大厂,小米、字节、百度、阿里、美团几乎全部在场,它们把机器人当做自身AI能力的落地场景之一。其中,小米的布局尤其典型,从整机(自变量机器人、宇树科技)到关键零部件(曦诺未来、纬钛机器人),几乎形成了一条具身智能产业链投资版图。

另一类是制造业头部企业。比如宁德时代、比亚迪、立讯精密,它们入局更多是为自己的主业找增量。刘明表示,机器人是下一代制造业的核心基础设施,没有人比制造业企业更了解生产线上真实的痛点,此时入场,既是提前卡位未来,也是为现有主业打造新防线。

这一类资本展现出的机会相对具体,大厂能带来算力、数据和应用场景,制造业巨头带来的是供应链、产线,对于机器人创业公司而言,有时股东带来的资源比融资金额更值钱。谁能把自己嵌进这些产业方的真实需求里,谁就更可能拿到这笔带资源的钱。

第三类是近两年出手的主力,国资与政府引导基金。

这类资本紧扣国家政策方向,虽然对早期项目的不确定性有一定容忍度,但更倾向于在产业方向明确、项目进入成长期后,以较大体量集中进入。

王瀚表示,各地政府引导基金除了财务回报,也承担着招商引资与产业培育的任务,会积极投向符合地方战略布局的赛道。这类资金规模庞大、出手稳健,已是当前一级市场中不可忽视的力量。国资的加速入场,也是具身智能从科技概念升级为国家战略的直接体现。

第四类是早期专业机构,在行业格局最不清晰的阶段提前下注。

王瀚所在的机构便属于这类。他表示,在种子轮、天使轮阶段就进行布局,意味着要承担技术路线不明朗、本体形态未确定、商业模式不清晰的多重风险。但一旦押中,回报倍数远超中后期入场者。

这也是这类机构区别于其他玩家的核心特征。在他们看来,与其在后期挤破头争抢确定性,不如在早期押注可能性。周全认为,中国市场足够大、细分赛道足够多,能容得下很多家公司,最后胜者为王。

但这并不等同于盲目追着最热的概念跑。周全告诉「定焦One」,“在共识尚未形成、路径仍然模糊的时候,提前看懂技术变化、产业需求和人的可能性是关键,半导体、具身智能、AI行业都是如此。”

四类资本,四套逻辑,但不止一位投资者表示,目前来看,具身智能真正的刚需应用场景还不够明确,即使是像宇树这样备受关注的头部公司,仍偏向科研与表演展示,距离规模化、商业落地还有很长的一段路要走。

资本已经用真金白银,把具身智能推到了量产元年的门口。究竟有没有一个足够大的真实市场,仍要等这些公司一家一家用业绩去证明。

注:文/定焦One团队,文章来源:定焦One,本文为作者独立观点,不代表亿邦动力立场。

文章来源:定焦One

广告
微信
朋友圈

这么好看,分享一下?

朋友圈 分享

APP内打开

+1
+1
微信好友 朋友圈 新浪微博 QQ空间
关闭
收藏成功
发送
/140 0