• 亿邦公众号
    • APP下载

深圳宝安又冲出一个IPO!50岁创始人 把多光谱AI做成了全国第一

作者:流光 文章来源:创业邦

2026-06-25 12:08

作者:流光 文章来源:创业邦

2026-06-25 12:08

邦小白快读

EN
全文速览

本文核心信息是深圳海清智元于6月22日在港交所挂牌上市,成为港股多光谱AI第一股,目前是国内多光谱AI赛道的头部企业。

1. 企业基本情况:创始人周波、陈永刚是拥有多年经验的安防老兵,属于二次创业,公司聚焦多光谱AI技术,融合不同波段光学信息与AI算法,探测传统视觉无法发现的安全风险,打造了从AI模组、智能终端到大模型云端平台的闭环业务体系,截至2025年底累计服务超2500家客户。

2. 发展与经营数据:按2025年收入计,海清智元多光谱AI整体市占率、多光谱AI大模型市占率均为全国第一,2023-2025年收入从1.17亿元增长至6.69亿元,大模型业务已经贡献超一半营收,早期投资机构最高回报超过113倍。

3. 行业前景:多光谱AI仍处于早期阶段,国内市场预计2025-2030年复合增长率达31.8%,增长空间广阔,目前尚未出现绝对龙头。

海清智元作为多光谱AI细分赛道的头部品牌,其发展路径对科技品牌有多方面参考价值。

1. 产品研发方向:品牌抓住传统可见光安防在复杂场景下“失明”的痛点,提前布局多光谱AI这一差异化细分赛道,逐步搭建底层技术,跟随客户需求从单一算法延伸到模组、终端再到大模型服务,形成三级产品矩阵,贴合市场需求持续升级。

2. 消费与产业趋势:当前多光谱AI已经从专业领域走向工业安全、智慧城市、数据中心运维等通用产业场景,行业整体从单一硬件向软硬件一体化方案演进,政策也明确支持云网边端协同AI技术发展,市场增长空间充足。

3. 品牌增长经验:深耕底层技术获得了资本的高度认可,多家知名机构加持,上市后进一步提升了品牌影响力,早期布局细分赛道获得了超百倍的投资回报,验证了细分赛道头部品牌的价值。

面向To B科技领域的从业者,本文清晰梳理了多光谱AI赛道的机会与风险,有较多干货参考。

1. 市场机会:多光谱AI是处于早期的高增长赛道,国内市场规模将从2025年的200亿元增长至2030年的794亿元,复合增长率达31.8%,传统视觉感知无法满足多场景对温度、热源等隐形信息探测的需求,市场缺口大;目前行业极度分散,头部企业市占率仅3.3%,前五大企业合计占比仅10.9%,新进入者仍有充足的成长空间。

2. 风险提示:发展过程中需要注意对赌风险,海清智元曾因业绩不达标触发补偿机制;同时高增长背景下要警惕现金流风险,2025年海清经营现金流为负,存货同比增长超150%,逾期应收账款接近7000万元,短期偿债压力较大。

3. 可学习路径:跟随客户需求从硬件向高附加值的大模型服务延伸,逐步提升营收规模和盈利能力,这个产品升级路径值得参考。

对涉感知硬件生产的工厂来说,本文明确了产品需求方向和新的商业机会,也提供了数字化转型的参考。

1. 产品生产设计需求:当前工业、安防等领域对感知设备的需求已经发生变化,传统单一可见光设备无法满足复杂场景需求,市场需要集成可见光、红外、紫外多类传感器,搭载AI芯片,可以同步采集分析多波段信息的多光谱硬件,工厂调整产品设计生产方向时可贴合该需求。

2. 商业机会:多光谱AI行业尚处早期,市场快速增长,目前头部品牌专注于技术研发和算法服务,对上游硬件生产加工有充足的合作需求,工厂可切入多光谱AI的供应链体系,获得新的增长增量。

3. 数字化转型启示:海清智元抓住产业升级的痛点,用数字化AI技术改造传统视觉感知产品,在细分赛道建立了领先优势,工厂推进数字化升级时,要聚焦具体场景痛点,打造差异化竞争力,抓住新产业升级的红利。

对AI和工业领域服务商来说,本文梳理了多光谱AI赛道的发展现状、痛点和机会,有较多干货。

1. 行业发展趋势:多光谱AI仍处于发展早期,国内市场未来5年复合增长率达31.8%,全球市场同样保持高增长,行业发展方向从过去销售单一传感器、硬件,向整合感知、分析、决策的软硬件一体化方案演进,政策支持云网边端协同推理技术发展,未来随着边缘计算、自动驾驶等场景发展,应用空间会进一步扩大。

2. 核心技术方向:行业核心技术是多波段光学信息与AI算法的融合,当前基于多光谱感知数据训练的行业大模型是核心增长点,端云协同的技术架构是主流方向,海清智元的多光谱AI大模型已经成为第一大收入来源,验证了技术的商业化可行性。

3. 客户痛点与解决方案:传统设备无法在夜间、烟雾、强光等复杂场景探测隐形风险,工业客户需要从感知到决策的完整闭环服务,服务商可参考该路径打造一体化解决方案,满足客户需求。

对科技投资、产业孵化类平台商来说,本文提供了多光谱AI赛道的布局参考和风险规避方向。

1. 企业对平台的核心需求:硬核科技初创企业发展过程中,需要持续的资本投入和政策扶持,海清智元从A轮到D轮累计融资2.6亿元,获得了高新投等多家产业平台的支持才实现快速技术扩张,对平台的融资赋能需求较高。

2. 平台招商与布局方向:多光谱AI是高增长的新兴蓝海赛道,目前行业尚未出现绝对龙头,大量优质初创企业分散在各产业链环节,已经验证了商业化可行性,平台可将该赛道作为重点招商布局方向,挖掘优质项目。

3. 风向规避:该赛道前期研发投入高,企业容易出现业绩不达预期、现金流紧张、存货和应收账款高增等问题,平台在服务过程中要提示企业管控对赌条款风险,合理控制扩张节奏,帮助企业管理流动性风险,避免出现资金链问题。

对科技产业研究者来说,本文提供了多光谱AI这一新兴细分赛道的最新一手研究素材,有较多参考价值。

1. 产业新动向与新商业模式:多光谱AI是AI+视觉领域的新兴细分方向,已经实现规模化商业化落地,当前行业处于早期增长阶段,海清智元作为目前的行业第一,已经形成模组、终端、大模型服务三级产品体系,大模型服务占比超50%,验证了“硬件+AI服务”新商业模式的可行性,行业整体正从单一硬件向闭环解决方案演进。

2. 行业新问题:当前行业竞争格局极度分散,2025年前五大企业合计市占仅10.9%,尚未出现绝对龙头;头部企业仍面临研发投入大导致的净利润下滑、现金流为负、存货和应收账款高增等问题,早期融资的对赌条款也会给创始人带来经营压力。

3. 政策启示:当前工信部已经出台政策支持相关技术发展,未来需要进一步加大对细分硬核科技企业的扶持,帮助企业渡过研发投入期,推动行业整合升级。

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 highlights that Shenzhen Haiqing Zhiyuan listed on the Hong Kong Stock Exchange on June 22, becoming the first "multispectral AI" stock traded in Hong Kong and the current leading player in China's multispectral AI sector.

1. Company fundamentals: Founders Zhou Bo and Chen Yonggang are veteran industry professionals with years of experience in the security sector, and this is their second entrepreneurial venture. The company focuses on multispectral AI technology, which integrates optical information across different wavelengths with AI algorithms to detect safety risks invisible to traditional computer vision. It has built a closed-loop business ecosystem covering AI modules, smart terminals and large model cloud platforms, and had served more than 2,500 clients as of the end of 2025.

2. Growth and operational metrics: By 2025 revenue, Haiqing Zhiyuan holds the largest market share in both China's overall multispectral AI market and the multispectral AI large model segment. Its revenue grew from RMB 117 million in 2023 to RMB 669 million in 2025, with large model services contributing over half of total revenue. Early investors have seen returns of more than 113x on their investment.

3. Industry outlook: The multispectral AI industry is still in an early stage of development. The domestic market is projected to achieve a compound annual growth rate (CAGR) of 31.8% between 2025 and 2030, indicating significant room for growth, and no absolute market leader has emerged yet.

As a leading brand in the niche multispectral AI sector, Haiqing Zhiyuan's growth journey offers multiple takeaways for technology brands.

1. Product R&D direction: The company addressed the pain point of "blindness" that traditional visible-light security systems suffer in complex scenarios, and pre-emptively positioned itself in the differentiated niche of multispectral AI. It built up underlying technology step by step, expanded its offering from single algorithms to modules, terminals and eventually large model services in line with client demand, and formed a three-tier product matrix that evolves continuously to match market needs.

2. Consumer and industrial trends: Multispectral AI is now expanding from specialized fields to general industrial scenarios including industrial safety, smart cities and data center operation and maintenance. The industry as a whole is evolving from selling standalone hardware to integrated software-hardware solutions, and government policies explicitly support the development of cloud-edge-end collaborative AI technology, leaving ample room for market growth.

3. Brand growth lessons: Deep investment in underlying technology earned Haiqing Zhiyuan strong recognition from capital markets, with backing from multiple well-known institutions. Its public listing has further boosted brand influence. Its early positioning in the niche sector has delivered over 100x returns for early investors, validating the value of leading brands in specialized niche markets.

For B2B technology industry practitioners, this article clearly outlines the opportunities and risks in the multispectral AI track, with plenty of actionable insights.

1. Market opportunities: Multispectral AI is an early-stage high-growth track. China's market size is projected to grow from RMB 20 billion in 2025 to RMB 79.4 billion in 2030, representing a 31.8% CAGR. Traditional visual perception cannot meet demand for detecting invisible information such as temperature and heat sources across many scenarios, creating a large market gap. The industry is currently highly fragmented: the leading player holds only 3.3% market share, and the top five players collectively hold just 10.9%, leaving ample room for new entrants to grow.

2. Risk warnings: Practitioners need to be wary of the risks associated with valuation adjustment mechanism (VAM) agreements—Haiqing Zhiyuan triggered a compensation clause after failing to meet performance targets. Against the backdrop of high growth, companies should also watch cash flow risks: in 2025, Haiqing Zhiyuan reported negative operating cash flow, inventory grew more than 150% year-over-year, and overdue accounts receivable reached nearly RMB 70 million, leaving it facing significant short-term debt repayment pressure.

3. Actionable growth path: Expanding from hardware to high-value-added large model services in line with customer demand, to gradually scale revenue and improve profitability, is a product upgrade path worth referencing.

For factories engaged in perception hardware manufacturing, this article clarifies product demand trends and new business opportunities, and also offers reference for digital transformation.

1. Product design and manufacturing demand: Demand for perception equipment in industrial and security sectors has changed. Traditional single visible-light devices can no longer meet the requirements of complex scenarios. The market needs multispectral hardware that integrates visible, infrared and ultraviolet sensors, is equipped with AI chips, and can collect and analyze multi-band information simultaneously. Factories can align their product design and manufacturing adjustments with this demand.

2. Business opportunities: The multispectral AI industry is still in its early phase of rapid growth. Current leading brands focus on technology R&D and algorithm services, and have strong demand for cooperation with upstream hardware manufacturers. Factories can enter the multispectral AI supply chain to capture new growth increment.

3. Insights for digital transformation: Haiqing Zhiyuan addressed industrial upgrade pain points by transforming traditional visual perception products with digital AI technology, and built a leading position in its niche track. When advancing digital upgrades, factories should focus on pain points in specific scenarios, build differentiated competitiveness, and capture dividends from new industrial upgrading.

For AI and industrial service providers, this article sorts out the current development status, pain points and opportunities of the multispectral AI track, with abundant actionable insights.

1. Industry development trends: Multispectral AI is still in an early development stage, with a projected 31.8% CAGR in China over the next five years, and the global market is also growing rapidly. The industry is evolving from selling standalone sensors and hardware to integrated software-hardware solutions that combine perception, analysis and decision-making. Government policies support the development of cloud-edge-end collaborative inference technology, and the application space will expand further with the development of edge computing, autonomous driving and other scenarios in the future.

2. Core technology direction: The core technology of the industry lies in the integration of multi-band optical information and AI algorithms. Industry large models trained on multispectral perception data are currently the core growth driver, and a cloud-edge collaborative technical architecture is the mainstream direction. Haiqing Zhiyuan's multispectral AI large model has already become its largest source of revenue, validating the commercial feasibility of this technology path.

3. Customer pain points and solutions: Traditional equipment cannot detect invisible risks in complex scenarios such as night, smoke and strong light. Industrial clients need full closed-loop services from perception to decision-making. Service providers can follow this path to build integrated solutions that meet customer demand.

For technology investment and industrial incubation platforms, this article provides reference for layout strategy and risk mitigation in the multispectral AI track.

1. Core demand from startups: Hard tech startups need continuous capital input and policy support to grow. Haiqing Zhiyuan raised a total of RMB 260 million across Series A to D rounds, and achieved rapid technology expansion only with support from multiple industrial platforms including Gao Xin Tou. These startups have high demand for financing empowerment from platforms.

2. Investment and layout direction for platforms: Multispectral AI is a high-growth blue ocean emerging track. No absolute market leader has emerged yet, and a large number of high-quality startups are scattered across all links of the industrial chain. Commercial feasibility has already been proven, so platforms can prioritize this track for investment and startup sourcing, and source high-quality projects here.

3. Risk mitigation: This track requires high upfront R&D investment, and companies often face issues including missed performance targets, tight cash flow, and rapid growth in inventory and accounts receivable. When serving portfolio companies, platforms should help them manage the risks of VAM clauses, control expansion pace reasonably, manage liquidity risks, and avoid broken capital chains.

For technology industry researchers, this article provides up-to-date first-hand research material on the emerging niche track of multispectral AI, with high reference value.

1. New industrial trends and business models: Multispectral AI is an emerging niche direction in the AI + computer vision field, and has already achieved large-scale commercialization. The industry is currently in an early growth stage. As the current industry leader, Haiqing Zhiyuan has formed a three-tier product system of modules, terminals and large model services, with large model services accounting for over 50% of revenue. This validates the feasibility of the new "hardware + AI service" business model, and the industry as a whole is evolving from standalone hardware to closed-loop solutions.

2. New industry issues: The current competitive landscape is extremely fragmented: the top five players held a combined 10.9% market share in 2025, and no absolute leader has emerged. Leading players still face challenges including net profit decline from heavy R&D investment, negative cash flow, and rapid growth in inventory and accounts receivable, while VAM clauses from early financing also create operational pressure for founders.

3. Policy implications: China's Ministry of Industry and Information Technology has already introduced policies to support the development of related technologies. Going forward, greater support is needed for specialized hard tech companies to help them get through the R&D-intensive phase and drive industry consolidation and upgrading.

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.

「IPO全观察」栏目聚焦首次公开募股公司,报道企业家创业经历与成功故事,剖析公司商业模式和经营业绩,并揭秘VC、CVC等各方资本力量对公司的投资加持。

作者丨流光

编 辑丨关雎

6月22日,深圳海清智元科技股份有限公司(下称“海清智元”,股票代码:01392.HK)在港交所主板挂牌上市,成为港股多光谱AI第一股。

海清智元开盘报29港元/股,市值224.52亿港元(约194亿元)。截至发稿,股价报25.9港元,最新市值200亿港元。

2013年4月,创始人周波与陈永刚在深圳创立了海清智元。公司专注于多光谱AI技术,其核心是将可见光、红外线、紫外线等不同波段的光学信息与AI算法融合,探测那些“看不见但正在发生”的风险,例如异常温升、电弧、早期火情和设备老化等。

围绕多光谱AI技术,公司打造出一套从感知到分析再到决策的闭环业务体系。底层是集成了多波段传感器的AI模组,中层是可独立部署的智能感知终端,顶层则是基于自研大模型的云端服务平台。这套体系广泛应用在工业安全、智慧城市、数据中心运维等领域中。截至2025年底,公司累计服务超2500家客户。

2023年至2025年,海清智元收入从1.17亿元增长至5.23亿元,再增至6.69亿元。

根据弗若斯特沙利文的报告,按2025年收入计,海清智元在中国多光谱AI公司中排名第一,市场份额3.3%。其多光谱AI大模型服务也位居全国第一,市场份额高达23.0%。

IPO前,朗科投资、浙商创投、高新投集团、小禾创投、毕沃德、科智华、盈科值得私募基金等机构都投资了海清智元。

上市后,浙商创投持股13.22%,是最大外部机构股东。朗科投资持股5.87%、毕沃德持股8.88%、高新投集团持股4.04%、科智华持股1.79%、盈科值得私募基金持股1.25%。

按照开盘市值224.52亿港元(约合194亿元人民币)计算,浙商创投所持份额价值约25.65亿元,回报倍数达25.65倍。

创始人、董事会主席、执行董事兼总经理周波直接持股5.31%,同时通过员工持股平台控制公司43.5%的投票权。执行董事兼技术专家陈永刚直接持股3.07%。

安防老兵二次创业

周波今年50岁,1999年毕业于西安矿业学院(现西安科技大学)自动控制专业。他的第一份工作是在成都科力电子研究所担任视讯研发工程师。这家机构专业生产安防系列产品,周波主要负责视讯技术的研发。此时陈永刚也在此从事研发工作。

2001年“9·11”事件后,全球安防需求激增。这一年,成都科力创始人张少锋南下深圳创办了图敏实业,周波和陈永刚也随之加入。

该公司早期专注于视频监控硬件产品研发,并在2008年与成都科力达成并购。2015年公司转型智慧交通业务,更名深圳市图敏智能视频股份有限公司。

从成立时间看,图敏实业比现在的安防龙头企业海康威视还早了2个月。不过,图敏实业后续发展并未能跟上市场节奏,周波与陈永刚等核心成员也选择离开。

之后,周波与陈永刚在2005年联手创办了深圳市黄河数字技术有限公司,主营网络化视频监控,并参与了电信“平安城市”“全球眼”等大型国家级项目。2009年6月,黄河数字被加拿大上市公司加创系统收购。

2013年4月,周波与陈永刚再次联手,在深圳宝安区创立海清智元。

此时,传统安防依赖可见光的局限已经显现——夜间、烟雾、强光等场景下,摄像头几乎“失明”。周波判断,融合多光谱感知与AI算法将是下一代视觉技术的方向,于是公司从一开始就瞄准多光谱AI这条细分赛道。

之后,海清智元围绕多光谱感知底层技术进行研发。2015年,公司发布AI视觉场景训练技术。从这一年开始,公司逐步摆脱单纯的算法研发阶段,开始将技术嵌入具体的应用场景。

直到2020年,海清智元发布多光谱AI感知计算整合技术。至此,可见光、红外、紫外等多波段信号的采集、融合与分析,被完整地纳入同一技术框架。

早期投资人回报达113倍

刚成立时,海清智元注册资本100万元,周波持有全部股权。2014年,周波通过一系列股权转让,将股权分别转给了陈永刚、员工持股平台众诚创展以及毕沃德和朗科投资。

2016年,周波在一次行业展会上结识了湖南乐泊科技董事长夏东。乐泊科技主营智慧停车解决方案,正在推进AI与停车场运营场景结合。

双方在2017年底推动了一项股权置换计划。根据安排,海清智元原有股东将所持股权全部出售给乐泊科技,以换取乐泊科技部分股权。2017年12月,乐泊科技成为海清智元唯一股东,公司管理层则继续留任经营。

不过,由于双方在产品与服务的长期发展方向上存在差异,原本预期的协同效应和战略收益并未实现。2020年1月,双方决定解除全部股权置换安排,乐泊科技退回海清智元股权,海清智元原有股东则恢复持股。

从这之后,海清智元开始独立融资。

2020年11月,公司完成A轮融资,引入韬略新能源、高新投、成都中小担、朗科投资等机构,投后估值约3.5亿元。

此后五年间,海清智元又陆续完成B轮、C轮和D轮融资,并持续扩充员工持股平台。招股书显示,公司累计融资金额约2.6亿元,投资方包括浙商创投体系、高新投、成都中小担、朗科投资以及多家产业和市场化基金。

随着多光谱AI概念升温,公司估值一路上涨。到2025年7月完成D轮融资时,海清智元估值已达到35.5亿元,较A轮增长超过10倍。

不过,估值上涨并不意味着发展一路顺风。

招股书显示,A轮融资时,公司曾与投资机构约定业绩承诺条款。由于2020年至2022年累计净利润未达到约定目标,相关补偿机制被触发。

2025年5月,周波和陈永刚与投资机构达成协议,以名义对价人民币1元向11家机构投资者转让部分股份作为补偿。其中,周波转让约9.85万股,陈永刚转让约6.28万股,合计超过16万股。

上市后,海清智元主要投资机构持股分别为:浙商创投13.22%、朗科投资5.87%、高新投集团4.04%、科智华1.79%、盈科值得私募基金1.25%。

按照开盘市值224.52亿港元(约合194亿元人民币)计算,上述机构的回报倍数分别为25.65倍、113.88倍、17.42倍、16.77倍、4.85倍。

年入近7亿

大模型业务贡献过半营收

海清智元最早的产品,是多光谱模组。相比传统视觉设备主要依赖可见光信息,多光谱AI模组集成了可见光、红外、紫外等多种传感器,并搭载AI芯片,能够同步采集和分析不同波段的信息。

2025年,该业务实现收入2.09亿元,占总收入31.3%。按收入计算,公司在中国多光谱AI模组市场排名第四。

随着客户数量增加,海清智元逐渐发现,仅仅提供感知能力并不足够。在电力、煤矿、数据中心等场景中,客户更关心的并不是设备拍到了什么,而是这些信息意味着什么。

因此,公司开始推出多光谱AI感知终端。这类产品建立在模组基础之上,将感知、分析和预警能力集成到整机设备中,可直接部署于电力巡检、煤矿安全监测、工业设备运维等场景。2025年,该业务实现收入9326万元,占总收入13.9%。

到2024年,海清智元开始商业化推广“智元起源大模型”。这一系统采用端云协同架构,可处理来自多光谱AI终端及第三方设备的数据,并结合历史案例库和行业知识库,对潜在风险进行识别、预警和管理。

招股书显示,智元起源大模型基于超过1000万个多光谱感知数据点、10万余起真实案例事件以及超过1万个安全工程领域知识图谱训练而成。在火灾早期特征识别场景中,模型准确率超过95%,推理响应时间低至50毫秒。

当年该业务收入为1.14亿元,到2025年已增至3.55亿元,同比增长212.3%,收入占比提升至53.1%,首次超过模组业务,成为公司最大的收入来源。

支撑这套产品体系的,是公司构建的“光感图算”技术架构。其中,“光”对应光学设计能力,“感”对应多光谱感算一体技术,“图”对应成像与数据处理,“算”则包括端侧AI计算平台及多光谱AI大模型。

业务增长也反映在财务数据上。2023年至2025年,海清智元收入分别为1.17亿元、5.23亿元和6.69亿元。

不过,利润表现没有和收入完全同步。2024年,公司实现净利润4041万元;2025年净利润降至2935万元,同比下降27.4%。招股书显示,利润下降主要与研发投入和管理费用增长有关。

与此同时,公司经营活动现金流净额为-1.3亿元。截至2025年底,公司账上现金及现金等价物为6560万元,短期借款为1.52亿元。

应收账款和存货规模也在增长。2025年末,公司应收账款达到1.73亿元,其中逾期应收账款6790万元;存货余额增至8020万元,较上年增长超过150%。

多光谱AI还没有跑出真正的巨头

多光谱AI行业,仍是一个处于早期阶段的新市场。

根据弗若斯特沙利文数据,中国多光谱AI市场规模预计将由2025年的200亿元增至2030年的794亿元,期间复合增长率为31.8%。同期,全球市场规模预计将从850亿元增长至3054亿元。

市场快速增长的背后,是传统视觉感知能力的局限正在不断暴露。

无论是工业设备巡检、电力运维,还是数据中心管理、城市安全治理,越来越多场景开始要求设备不仅能够“看见”,还能够识别温度、热源、材料变化等传统摄像头难以获取的信息。多光谱感知技术因此逐渐从专业领域走向更广泛的产业应用。

不过,与市场规模相比,行业竞争格局仍然十分分散,目前仍未出现绝对龙头。

弗若斯特沙利文数据显示,2025年中国多光谱AI市场前五大企业合计市场份额仅为10.9%。虽然海清智元以3.3%的市场份额排名第一,但领先第二名不足1个百分点。

无论是红外热成像企业、视觉AI企业,还是大模型公司,都在不同环节切入这一市场。

从产业发展路径来看,多光谱AI正在从单一硬件产品向软硬件一体化方案演进。

过去行业主要销售传感器、模组或终端设备,而随着大模型技术的发展,越来越多企业开始尝试将感知、分析和决策能力整合到统一平台之中。

2026年6月,工信部发布《“人工智能+信息通信”创新发展实施意见(2026—2028年)》,提出探索云网边端协同推理技术,增强网络边缘推理能力。

随着边缘计算、具身智能、自动驾驶等应用场景的发展,多光谱AI的应用空间有望进一步扩大。

注:文/流光,文章来源:创业邦(公众号ID:ichuangyebang ),本文为作者独立观点,不代表亿邦动力立场。

【版权提示】亿邦动力倡导尊重与保护知识产权。未经许可,任何人不得复制、转载、或以其他方式使用本网站的内容。如发现本站文章存在版权问题,烦请提供版权疑问、身份证明、版权证明、联系方式等发邮件至run@ebrun.com,我们将及时沟通与处理。

已点赞

+1

{{ praise_text }}

好文章,值得被更多人看到

广告

下一篇

跨境电商资源对接

《全球跨境电商服务商资源手册》第二期手册现已发布,触达更多资源,了解跨境服务商信息,欢迎点击链接下载。

了解详情

Flipkart已落地上千个微履约仓 加码印度即时零售布局

沃尔玛投资的Flipkart在印度建成1000个微履约仓支撑分钟配送,计划2026年底前扩至1500个。