广告
加载中

中国电信领投 觅蜂科技完成新一轮数亿元融资

亿邦动力 2026-08-17 11:14
亿邦动力 2026/08/17 11:14

邦小白快读

EN
全文速览

本文核心信息是一站式物理AI数据服务平台觅蜂科技完成新一轮数亿元融资,本轮由中国电信领投,老股东超额跟投,核心业务是解决物理AI发展的数据瓶颈,以下为核心干货:

1. 企业核心成果:觅蜂科技从数据采集、治理、评测三端发力,全栈自研重构具身智能数据供给体系,已经实现MEgo无本体采集终端量产,数据治理效率比传统人工提升10倍以上,在测试场景中每轮数据闭环可推动任务成功率提升5%-10%。

2. 融资用途与未来规划:本轮资金将用于建设具身智能数据平台供给基础设施,扩大产能,加速千万小时级物理交互数据落地,还会联合投资方打造标准化云网加具身数据一体化解决方案,推动具身智能落地。

本文能帮助布局AI相关业务的品牌商把握具身智能赛道趋势,获得商业机会,干货总结如下:

1. 产业消费趋势:当前具身智能正从实验室走向真实落地场景,产业处于高速增长期,核心痛点是缺乏大规模高质量真实物理交互数据,提前布局数据相关配套业务,能抢占赛道先发优势。

2. 产品研发启示:物理AI模型迭代需要持续更新数据,数据供给不是一次性交付,品牌商研发AI相关产品时,要重视搭建数据回流闭环,匹配模型迭代需求。

3. 合作机会:觅蜂科技目前联合中国电信、张江集团打造开放式生态,面向人形机器人、智能制造、智慧家居等领域企业提供数据服务,布局相关业务的品牌商可对接获取高质量合规数据资源。

本文能给布局AI赛道的卖家梳理新机会与风险,干货总结如下:

1. 新增长机会:当前具身智能产业处于发展初期,核心配套的高质量数据供给缺口极大,细分的场景数据采集、数据标注处理等环节都有大量未被满足的需求,卖家可结合自身资源切入对应赛道。

2. 可对接的合作方向:觅蜂科技的采集终端已经覆盖工厂、物流、商超、养老等数十个真实场景,拥有场景资源的卖家可以和觅蜂这类头部平台合作,开展场景数据采集业务,依托平台能力降低自身风险获得稳定收益。

3. 风险提示:物理AI对数据质量、精度要求远高于普通大语言模型,单一环节的小玩家很难满足客户需求,盲目入场容易因为数据质量不达标被淘汰,优先依托头部平台生态合作更稳妥。

本文给制造类工厂带来了具身智能时代的数字化转型与业务拓展机会,干货总结如下:

1. 新增商业机会:当前具身智能、智能制造落地加速,对工厂真实场景的高质量物理交互数据需求旺盛,工厂可以开放自身生产场景,参与数据采集合作,在不影响正常生产的前提下获得额外收益。

2. 数字化转型启示:工厂布局智能制造、工业AI的时候,训练模型需要大量真实生产场景数据,觅蜂科技的轻量化无本体采集终端可低成本深入各类工厂场景获取高精度数据,工厂可借助这类成熟数据基础设施降低自身AI转型的数据成本。

3. 生态合作机会:目前觅蜂科技联合张江集团、中国电信共建具身智能数据公共服务载体,联合研发行业数据集,工厂可参与生态合作,对接下游智能制造、人形机器人企业的需求,拓展自身业务范围。

本文给AI相关服务商明确了具身智能赛道的发展方向与可参考方案,干货总结如下:

1. 行业发展趋势:具身智能是当前AI产业的核心增长赛道,物理AI要实现落地突破,必须先解决数据供给不足的问题,数据基础设施服务是当前赛道最紧缺的服务方向,市场空间大,资本关注度高。

2. 核心客户痛点:相比普通大语言模型,物理AI需要的数据有效信息密度更低、噪声更高,整体需求量大得多,传统人工数据处理模式效率低成本高,一次性数据交付无法满足模型持续迭代的需求,这些都是服务商可以切入的痛点。

3. 可复制的解决方案:可参考觅蜂科技的三端全栈方案,采集端用轻量化无本体终端实现多场景低成本高精度采集,治理端打造自动化闭环将效率提升10倍以上,评测端搭建数据回流闭环持续更新数据,该模式已经得到头部AI企业验证,可行性较强。

本文给AI相关平台商梳理了具身智能数据赛道的发展方向与运营要点,干货总结如下:

1. 当前产业核心需求:具身智能上下游企业都急需大规模、合规、高质量的真实场景物理数据,现有供给无法满足行业需求,平台商搭建标准化的具身数据供给基础设施,能解决行业核心痛点,获得增长空间。

2. 可借鉴的运营做法:可以对接产业资本、地方产业平台整合资源,依托投资方的云网能力、场景资源,搭建覆盖采集、治理、评测全链路的核心能力,同时开放生态对接上下游企业,联合打造垂直行业解决方案,实现生态共赢。

3. 风险规避要点:做数据基础设施平台需要搭建全链路核心能力,不能只布局单一环节,否则缺乏核心竞争力,同时要格外重视数据合规性,提前构建合规可控的真实场景数据资源池,避免合规风险。

本文为AI与数字产业研究者提供了具身智能赛道的最新研究素材,干货总结如下:

1. 产业最新动向:当前具身智能数据基础设施成为资本关注的热点,觅蜂科技创立半年内连续完成数轮融资,投资方囊括国资产业平台、头部市场化基金、产业链龙头企业,说明该方向已经成为产业与资本共同看好的核心赛道。

2. 产业新问题:当前物理AI发展面临数据荒漠的核心瓶颈,物理AI的数据特性和大语言模型差异极大,有效信息密度更低、噪声更高,需求量也大得多,传统数据供给模式完全无法适配,数据供给体系重构是当前产业需要解决的核心问题。

3. 创新商业模式参考:觅蜂科技打造的平台型数据供给基础设施,是全新的供给范式,数据供给不再是一次性交付,而是伴随模型部署持续更新的闭环模式,该模式已经验证可有效提升模型任务成功率,具备很高的研究价值。

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 covers that Mifeng Technology, a one-stop physical AI data service platform, has completed a new multi-hundred-million RMB funding round led by China Telecom, with over-subscription follow-on investments from existing shareholders. The company focuses on addressing the data bottleneck hindering the development of physical AI. Key takeaways are as follows:

1. Core achievements: Mifeng Technology has built a fully self-developed end-to-end embodied intelligent data supply system spanning data collection, governance, and evaluation. Its MEgo ontology-free data collection terminals have entered mass production, and the company’s data governance efficiency is more than 10 times higher than traditional manual processes. In test scenarios, each round of data closed-loop iteration boosts task success rates by 5% to 10%.

2. Use of funds and future plans: Proceeds from this round will be used to build embodied intelligent data supply infrastructure, expand production capacity, and accelerate the rollout of 10 million hours of physical interaction data. The company will also cooperate with its investor to develop a standardized integrated solution combining cloud-network infrastructure and embodied data to advance the commercial adoption of embodied intelligence.

This article helps brands布局 AI-related businesses grasp industry trends in the embodied intelligence sector and identify new business opportunities. Key takeaways are as follows:

1. Industry and consumer trends: Embodied intelligence is currently moving from laboratories to real-world commercial scenarios, and the industry is in a period of rapid growth. Its core pain point is the lack of large-scale, high-quality real physical interaction data. Early布局 in data-related supporting businesses can help brands capture first-mover advantage.

2. Product development insights: Physical AI model iteration requires continuous data updates, and data supply is not a one-time delivery. When developing AI-related products, brands should prioritize building a closed-loop data return system to match the needs of model iteration.

3. Cooperation opportunities: Mifeng Technology is building an open ecosystem in partnership with China Telecom and Zhangjiang Group, offering data services to companies in fields including humanoid robots, intelligent manufacturing, and smart home. Brands布局 related businesses can connect with the platform to access high-quality, compliant data resources.

This article helps sellers布局 the AI sector identify new opportunities and risks. Key takeaways are as follows:

1. New growth opportunities: The embodied intelligence industry is still in an early stage of development, and there is a massive gap in the supply of high-quality core supporting data. Unmet demand is abundant across segmented links including scenario-specific data collection and data annotation processing. Sellers can leverage their existing resources to enter these segmented tracks.

2. Potential cooperation directions: Mifeng Technology’s collection terminals already cover dozens of real-world scenarios including factories, logistics, retail supermarkets, and elderly care. Sellers with access to scenario resources can cooperate with leading platforms like Mifeng to conduct scenario-based data collection business, reducing risks and securing stable returns by relying on the platform’s capabilities.

3. Risk warnings: Physical AI has far higher requirements for data quality and accuracy than general large language models. Small players focusing on only one single link can hardly meet customer needs, and blind market entry can easily lead to elimination due to substandard data quality. Prioritizing cooperation through leading platform ecosystems is a more reliable approach.

This article presents digital transformation and business expansion opportunities for manufacturing factories in the era of embodied intelligence. Key takeaways are as follows:

1. New business opportunities: As the adoption of embodied intelligence and intelligent manufacturing accelerates, demand for high-quality physical interaction data collected from real factory scenarios is booming. Factories can open their production scenarios to participate in data collection cooperation, earning additional revenue without disrupting normal production.

2. Digital transformation insights: When factories布局 intelligent manufacturing and industrial AI, model training requires large volumes of data from real production scenarios. Mifeng Technology’s lightweight ontology-free collection terminals can access all types of factory scenarios at low cost to obtain high-precision data. Factories can leverage this type of mature data infrastructure to reduce data costs for their own AI transformation.

3. Ecosystem cooperation opportunities: Mifeng Technology is co-building a public embodied intelligent data service carrier and jointly developing industry-specific datasets with Zhangjiang Group and China Telecom. Factories can participate in the ecosystem cooperation, connect with demand from downstream intelligent manufacturing and humanoid robot enterprises, and expand their business scope.

This article clarifies development directions and reference solutions for AI-related service providers in the embodied intelligence sector. Key takeaways are as follows:

1. Industry development trends: Embodied intelligence is currently the core growth track of the AI industry. For physical AI to achieve commercial breakthroughs, the problem of insufficient data supply must be resolved first. Data infrastructure services are the most urgently needed direction in the current sector, with large market space and high capital attention.

2. Core customer pain points: Compared with general large language models, physical AI requires data with lower effective information density, higher noise, and a much larger overall volume. Traditional manual data processing models are inefficient and costly, and one-time data delivery cannot meet the needs of continuous model iteration. All these pain points represent entry points for service providers.

3. Replicable solutions: Service providers can reference Mifeng Technology’s three-end full-stack solution: on the collection end, lightweight ontology-free terminals enable low-cost, high-precision data collection across multiple scenarios; on the governance end, an automated closed-loop system improves efficiency by more than 10 times; on the evaluation end, a closed-loop data return system enables continuous data updates. This model has already been validated by leading AI enterprises and proven highly feasible.

This article sorts out development directions and operational key points for AI-related platform players in the embodied intelligence data track. Key takeaways are as follows:

1. Current core industry demand: Upstream and downstream enterprises in the embodied intelligence industry all have an urgent need for large-scale, compliant, high-quality physical data from real scenarios. Existing supply cannot meet industry demand. If platform players build standardized embodied data supply infrastructure, they can solve the core pain point of the industry and unlock significant growth space.

2. Reference operational practices: Platform players can partner with industrial capital and local industrial platforms to integrate resources, leverage investors’ cloud-network capabilities and scenario resources to build end-to-end core capabilities covering collection, governance and evaluation. Meanwhile, they can open their ecosystem to connect upstream and downstream enterprises and co-develop vertical industry solutions to achieve ecosystem-wide win-win results.

3. Risk mitigation: Building a data infrastructure platform requires developing full-stack core capabilities rather than focusing on only one single link. Partial布局 will lead to a lack of core competitiveness. In addition, players must attach great importance to data compliance, and build a compliant and controllable real-scenario data resource pool in advance to avoid compliance risks.

This article provides the latest industry insights for researchers focusing on AI and the digital industry in the embodied intelligence track. Key takeaways are as follows:

1. Latest industry developments: Embodied intelligent data infrastructure has recently become a hot spot for capital attention. Mifeng Technology has completed multiple funding rounds within just six months of its founding, with investors including state-owned industrial platforms, leading market-oriented funds, and leading industrial chain enterprises. This confirms that this direction has become a core track jointly favored by both industry and capital.

2. New industry challenges: The core bottleneck currently limiting physical AI development is a "data desert". The data characteristics of physical AI differ greatly from those of large language models: it has lower effective information density, higher noise, and requires a far larger total volume. Traditional data supply models are completely unable to adapt to these requirements, and reconstructing the data supply system is the core problem the industry currently needs to solve.

3. Reference for innovative business models: The platform-based data supply infrastructure built by Mifeng Technology represents a brand-new supply paradigm. Under this model, data supply is no longer a one-time delivery, but a continuous closed-loop updating model that accompanies model deployment. This model has been proven to effectively improve model task success rates, and has high research value.

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.

【亿邦原创】日前,一站式物理AI数据服务平台觅蜂科技完成新一轮数亿元融资。本轮融资由中国电信领投,张江集团跟投,红杉中国、元启创新等老股东超额追加投资。自2026年初创立以来,觅蜂科技在约半年时间内连续完成数轮融资,投资方囊括国资产业平台、头部市场化基金和产业链龙头企业。

本轮资金将用于持续冲破行业 “数据荒漠” 瓶颈,建设具身智能数据平台型供给基础设施,扩大MEgo无本体产品量产规模,夯实数据采集、数据治理、闭环评测全栈能力,加速千万小时级物理交互数据产能落地,重构物理世界大模型迭代范式。

物理AI要真正迎来“GPT时刻”,实现对环境变化、物理规律的掌握和更高阶的因果推理,离不开大规模、高质量真实物理交互数据。作为参照,主流大语言模型的预训练数据已达100万亿token,等效于100亿小时的数据。而物理AI需要的数据有效信息密度更低、噪声更高,对数据的需求量更大。面对行业瓶颈,觅蜂科技从数据采集端、治理端、评测端三端发力,全栈自研重构具身智能数据供给体系,打造具身智能数据“平台型供给”基础设施。

在采集端,觅蜂科技自研的MEgo系列无本体采集终端已实现量产交付,并已在工厂、物流、超市、家庭、酒店、餐饮、安防、养老、文旅等真实场景采集数据。MEgo系列终端实现宏观环境感知与微观手部操作数据的精准同步采集,无需依赖机器人本体,凭借轻量化便携硬件深入多样化场景,实现全域、高保真、低成本规模化的数据采集。在此基础上,觅蜂科技构建了分层数据供给金字塔,覆盖真机遥操、无本体数据采集等全类型。

在治理端,觅蜂科技推出的MEgo Engine数据治理服务平台,已打通预处理、空间重建、多模态标注、质量评测完整自动化闭环,将传统人工数据处理效率提升10倍以上。平台底层搭载自研VIO高精度位姿捕捉算法,支持人机技能迁移、长时序动作自动切分、原子操作智能标注,大幅降低企业、科研机构的数据处理成本。

在评测端,面向闭环迭代构建评测体系,通过多维数据洞察,优化采集策略,持续提升部署ROI。觅蜂科技通过部署回流数据飞轮,将任务执行、数据回流、案例挖掘、数据更新连接成自动化闭环——数据供给不再是一次性交付,而是伴随模型部署持续更新。在部分应用场景测试中,每轮数据闭环可推动任务成功率提升5%至10%。也正因如此,觅蜂科技被认为是"最懂模型和场景需求"的物理AI数据基础设施平台。

目前,觅蜂科技已建立20多个国内城市创新中心,并布局超过5个海外节点,并与国内外头部互联网、具身智能、AI企业等达成战略合作,为蚂蚁灵波LingBot-VLA 2.0等具身基座模型提供高质量训练数据。

本轮融资的投资方均拥有丰富的产业与场景优势,将进一步强化觅蜂科技在数据采集、治理到应用的生态协同能力,构建真实、合规、高质量的具身场景数据资源池。中国电信拥有强大的云网基础设施、广泛的客户渠道及政企客户资源。依托中国电信在5G网络、云端存储、算力调度、场景连接等优势,结合觅蜂科技在具身智能全链路数据采集、数据集生产、数据治理、软硬件一体化技术能力,双方拟联合面向人形机器人企业、智能制造厂商、科研院所、智慧园区和智慧家居政企客户,联合打造标准化“云网 + 具身数据”一体化解决方案,推动具身数据的高效采集、安全存储、规模治理与标准化供给。

张江科学城是世界级科创产业载体,在产业集聚、应用场景开放和创新资源配置等方面积累深厚。作为张江科学城建设主力军,张江集团将开放园区、工业、商业等多元实景应用场景,以及产业政策、空间载体及上下游企业对接资源,与觅蜂科技从硬件、技术、场景和生态等层面共同突破具身智能数据供给瓶颈。双方将共建具身智能数据公共服务载体,并联合研发高质量行业数据集,推动数据标准落地与生态企业集聚。

张江集团副总经理林晨表示:“觅蜂科技作为国内稀缺的平台型具身数据基建企业,拥有完整硬件采集、自动化治理、规模化数据供给全链路能力。技术路线与产业发展方向高度契合,期待双方持续深挖数据要素价值,推动更多具身智能创新成果在张江转化落地。”

觅蜂科技董事长兼CEO姚卯青表示,具身智能从实验室走向真实世界,关键在于能否持续获得真实、合规、高质量的数据。本轮融资带来的不仅是资本支持,更是云网基础设施、产业场景与生态资源的深度协同。觅蜂科技将依托已形成的“采集—治理—供给—评测”全链路能力,进一步提升规模化数据供给和场景服务能力,推动高质量数据更高效地转化为模型能力与产业价值。

文章来源:亿邦动力

广告
微信
朋友圈

FAQ回顾

觅蜂科技是做什么的?

觅蜂科技是一站式物理AI数据服务平台,打造具身智能数据“平台型供给”基础设施,自研MEgo系列无本体采集终端、MEgo Engine数据治理服务平台,具备数据采集、治理、闭环评测全栈能力,可为具身基座模型提供高质量训练数据。

物理AI发展面临的核心数据瓶颈是什么?

物理AI要实现高阶因果推理离不开大规模高质量真实物理交互数据,其所需数据有效信息密度更低、噪声更高,需求量远高于大语言模型,行业普遍存在“数据荒漠”瓶颈,传统人工数据处理效率低、成本高,难以满足发展需求。

觅蜂科技本轮融资的资金用途是什么?

本轮融资资金将用于建设具身智能数据平台型供给基础设施,扩大MEgo无本体产品量产规模,夯实数据采集、治理、闭环评测全栈能力,加速千万小时级物理交互数据产能落地,重构物理世界大模型迭代范式。

觅蜂科技的数据治理服务有哪些优势?

觅蜂科技全栈自研重构具身智能数据供给体系,其MEgo Engine数据治理服务平台可打通预处理、空间重建、多模态标注、质量评测完整自动化闭环,将传统人工数据处理效率提升10倍以上,还搭载自研高精度位姿捕捉算法,大幅降低数据处理成本。

这么好看,分享一下?

朋友圈 分享

APP内打开

赞 +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