广告
加载中

前华为高管再创业 AI记忆公司MemoraX AI三个月累计斩获数亿元融资

亿邦动力 2026-08-20 17:24
亿邦动力 2026/08/20 17:24

邦小白快读

EN
全文速览

本文核心信息是前华为高管创办的AI记忆领域创业公司忆纪元科技,成立三个月就完成三轮数亿元融资,是当前AI大模型细分赛道的热门项目,核心干货信息如下:

1. 团队背景:创始人郝建业是前华为诺亚方舟决策与推理实验室主任,现任天津大学智算学部教授、博士生导师,学术和产业背景都十分扎实,技术可信度高。

2. 技术成果:公司构建了完整的Agent记忆技术体系,在两个多模态记忆权威榜单拿到第一名,还将Token消耗降低了30%,已经推出编程场景、交互式娱乐场景的落地产品,技术已经实现场景验证。

3. 行业前景:根据行业报告,该赛道未来五年年复合增长率超过110%,市场规模将从14.4亿元增长到642.5亿元,属于增长潜力极大的新兴赛道。

本文披露了AI大模型细分赛道的最新发展动态和消费技术趋势,对布局AI相关业务的品牌商有较高参考价值,核心干货如下:

1. 赛道趋势:大模型长期记忆是当前资本高度看好的热门细分赛道,头部创业公司短时间内拿到三轮数亿元融资,资本认可度极高,赛道已经进入快速发展阶段。

2. 市场增长潜力:行业数据显示,中国智能体记忆市场2025到2030年规模将从14.4亿元增长至642.5亿元,年复合增长率超过110%,到2030年占智能体基础设施的比例将接近26%,成为第二大独立基础设施市场,增长空间巨大。

3. 落地方向参考:目前赛道已经实现编程工具、交互式娱乐场景落地,还在拓展医疗健康、智能家居、智能穿戴等领域,品牌商可结合自身业务嵌入标准化记忆模块,升级用户交互体验,已有头部客户合作案例可参考。

本文透露了AI智能体赛道的新增长机会,对想要切入AI相关领域的卖家有诸多干货参考,核心内容如下:

1. 蓝海市场机会:大模型长期记忆是目前大模型产业中稀缺的细分赛道,市场增速极快,未来5年市场规模增长超40倍,年复合增长率超过110%,属于早期高增长蓝海赛道,早期入场的卖家可获得更多红利。

2. 落地场景参考:目前已经验证了编程效率工具、交互式娱乐两个成熟落地场景,同时还在拓展医疗健康、智能家居、智能穿戴、具身智能等多个领域,卖家可围绕这些场景寻找配套的产品和合作机会。

3. 合作与风险提示:当前赛道头部公司已经推出标准化的记忆模块,还和华为云达成了战略合作,卖家可对接合作,将技术模块嵌入自身产品提升竞争力;同时赛道资本热度高,资源投入快,卖家需要抓住早期窗口提前布局。

本文为传统工厂推进数字化、AI智能化转型提供了新的方向和商业机会,核心干货如下:

1. 产品升级需求方向:当前AI大模型和智能硬件产业快速发展,对长期记忆模块的需求进入爆发期,市场规模即将迎来高速增长,如果工厂布局智能家居、智能穿戴、具身智能相关硬件产品,就需要配套长期记忆能力完成产品升级。

2. 对接合作商机:目前专注该赛道的忆纪元已经研发出成熟的标准化记忆模块,可开放给各类硬件厂商合作,工厂不需要投入高额资金自主研发,可通过对接现成技术模块快速完成产品智能化升级,降低转型门槛。

3. 数字化转型启示:AI基础设施细分赛道的机会越来越多,工厂可结合自身的业务场景,对接成熟的AI技术供应商,打造差异化的智能产品,提升产品附加值,避开传统硬件的价格竞争。

本文披露了智能体记忆赛道的最新发展动向,为AI相关服务商提供了行业趋势、客户痛点和解决方案的参考,核心干货如下:

1. 行业发展趋势:智能体记忆已经从大模型的附加功能,变成了独立的基础设施赛道,未来在智能体基础设施中的占比将提升至接近26%,成为仅次于模型的第二大细分市场,年复合增长率超过110%,赛道增长空间极大,是服务商未来可拓展的核心方向。

2. 行业普遍客户痛点:当前大模型Agent普遍存在长程任务上下文断档、多轮交互记忆混乱、Token消耗成本高的问题,这些都是服务商服务客户过程中经常遇到的共性痛点,有明确的市场需求。

3. 成熟解决方案可对接:目前忆纪元已经突破核心技术,在权威榜单取得第一,将Token消耗降低30%,推出了可规模化落地的标准化记忆模块,覆盖多个应用场景,服务商可整合该方案,为自身客户提供AI能力升级服务,拓展自身业务边界。

本文透露了当前大模型生态对平台的新需求,为平台商的招商、生态布局和风险规避提供参考,核心干货如下:

1. 生态布局新需求:当前大模型生态不断细化,长期记忆是目前稀缺的核心基础设施,市场需求呈爆发式增长,平台需要引入这类细分赛道的优质企业,才能完善自身大模型生态布局,提升平台竞争力。

2. 优质招商标的特征:该赛道的优质标的具备硬核技术背景,创始人同时具备大厂从业经验和学术研究背景,技术已经获得权威榜单认可,有多个落地场景和头部客户合作案例,还获得资本多轮投资,成长确定性高,是平台优质的招商对象。

3. 风向与风险提示:当前AI大模型赛道概念化项目较多,聚焦底层核心技术的细分赛道增长确定性更高,平台布局这类企业可有效规避概念项目风险;目前头部云平台华为云已经引入该赛道企业达成战略合作,其他平台可跟进布局,抢占生态先机。

本文给出了大模型产业发展的最新产业动向,对AI产业研究者有较高的研究价值,核心干货如下:

1. 产业新动向:当前大模型产业已经进入细分分化阶段,底层基础设施逐步拆分,长期记忆已经从大模型的附属功能,变成独立的垂直赛道,诞生了专注该赛道的创业公司,并且获得资本高度认可,成立三个月就完成三轮数亿元融资,资本对赛道的认可度极高。

2. 明确的市场规模数据:行业研究机构给出了该赛道未来五年的增长预测,中国智能体记忆市场年复合增长率超过110%,2030年市场规模将达到642.5亿元,占智能体基础设施的比例接近26%,成为仅次于模型的第二大独立基础设施市场,明确验证了赛道的商业价值。

3. 技术和商业模式新动向:该赛道已经形成“底层核心技术研发+标准化记忆模块输出+对接多场景B/C端客户”的商业模式,已经解决了长上下文记忆、高Token消耗的行业痛点,和多个领域头部客户达成合作,可作为新产业孵化的典型研究案例。

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 Yijiyuan Technology, an AI memory startup founded by a former Huawei executive, which has completed three financing rounds totaling hundreds of millions of yuan just three months after its founding, making it one of the hottest startups in a generative AI niche. Key takeaways are as follows:

1. Team background: Founder Hao Jianye previously served as director of Huawei Noah's Ark Laboratory's Decision and Inference Lab, and currently holds a position as professor and PhD supervisor at Tianjin University's Intelligent Computing Department. His deep academic and industry background lends strong technical credibility to the startup.

2. Technical progress: The company has built a complete Agent memory technology stack, claimed the top spot on two authoritative multimodal memory benchmarks, and reduced token consumption by 30%. It has already launched products for coding and interactive entertainment use cases, completing technical validation for real-world applications.

3. Industry outlook: Industry projections show this niche will achieve a compound annual growth rate (CAGR) exceeding 110% over the next five years, with total market size expanding from 1.44 billion yuan to 64.25 billion yuan, making it an emerging high-growth segment.

This article outlines the latest developments and consumer technology trends in a fast-growing generative AI niche, offering valuable insights for brands building AI-powered businesses. Key takeaways are as follows:

1. Trend analysis: Long-term memory for large models is a hot AI niche that has attracted intense capital interest, with a leading startup securing three hundreds-of-millions-yuan financing rounds in a very short time. High capital inflow confirms the segment has entered a phase of rapid growth.

2. Growth potential: Industry data projects China's agent memory market will grow from 1.44 billion yuan in 2025 to 64.25 billion yuan in 2030, representing a CAGR of over 110%. By 2030, it will account for nearly 26% of all agent infrastructure, making it the second-largest independent infrastructure market with enormous room for growth.

3. Go-to-market reference: The segment already has live deployments in coding tools and interactive entertainment, with expansion ongoing into healthcare, smart home, and smart wearables. Brands can integrate standardized memory modules into their existing products to upgrade user experience, with proven collaboration cases with leading clients already available for reference.

This article highlights new growth opportunities in the AI agent space, offering actionable insights for sellers looking to enter AI-related sectors. Key takeaways are as follows:

1. Blue ocean opportunity: Long-term memory for large models is a currently underserved niche in the broader large model industry, with extremely rapid market expansion. The market will grow more than 40x over the next five years, with a CAGR exceeding 110%, making it an early-stage, high-growth blue ocean. Sellers that enter early stand to capture significant first-mover advantage.

2. Use case references: Two mature use cases—coding productivity tools and interactive entertainment—have already been validated, with expansion underway into healthcare, smart homes, smart wearables, embodied intelligence and other sectors. Sellers can pursue complementary product and partnership opportunities around these high-potential areas.

3. Collaboration and risk notes: Leading players in the segment already offer standardized memory modules, and the top player has secured a strategic partnership with Huawei Cloud. Sellers can partner with these providers to integrate ready-made modules into their own products to boost competitiveness. At the same time, the segment is seeing high capital enthusiasm and rapid resource deployment, so sellers need to seize the early window to establish their position.

This article outlines new directions and business opportunities for traditional factories pursuing digital and AI-powered transformation. Key takeaways are as follows:

1. Product upgrade demand: The rapid growth of large AI models and the smart hardware industry has sparked an explosion in demand for long-term memory modules, and the market is about to enter a phase of hyper-fast growth. Factories producing smart home, smart wearable or embodied intelligence hardware will need to integrate long-term memory capabilities to upgrade their products.

2. Collaboration opportunities: Yijiuan Technology, the leading player focused on this segment, has already developed mature standardized memory modules open for collaboration with hardware manufacturers. Factories do not need to invest heavily in independent R&D; they can complete intelligent product upgrades quickly by integrating ready-made modules, lowering the barrier to transformation.

3. Transformation insights: Opportunities in AI infrastructure niches are multiplying. Factories can align AI opportunities with their own business scenarios, partner with mature AI technology providers to build differentiated smart products, boost product value-added, and avoid the price competition common in traditional hardware markets.

This article covers the latest developments in the agent memory segment, offering insights on industry trends, customer pain points and available solutions for AI-focused service providers. Key takeaways are as follows:

1. Industry trend: Agent memory has evolved from an add-on feature of large models to a stand-alone infrastructure segment. It will account for nearly 26% of total agent infrastructure by 2030, becoming the second-largest niche after foundation models, with a CAGR exceeding 110% and enormous growth potential. It is a core expansion direction for service providers looking for new growth.

2. Common customer pain points: Current large model Agents普遍 suffer from broken context in long-horizon tasks, inconsistent memory across multi-turn interactions, and high token consumption costs. These are widespread pain points service providers regularly encounter when serving clients, representing clear unmet market demand.

3. Deployable ready-made solutions: Yijiuan Technology has already achieved core technological breakthroughs, ranking first on authoritative industry benchmarks, cutting token consumption by 30%, and launching scalable standardized memory modules that cover multiple application scenarios. Service providers can integrate this solution to deliver upgraded AI capabilities to their own clients and expand their business scope.

This article outlines new demands from the evolving large model ecosystem for platform operators, offering guidance for investment sourcing, ecosystem layout and risk mitigation. Key takeaways are as follows:

1. New demand for ecosystem building: As the large model ecosystem continues to specialize, long-term memory has emerged as a scarce core infrastructure with exponentially growing market demand. Platforms need to onboard high-quality players from this niche to complete their large model ecosystem layout and improve their overall competitiveness.

2. Characteristics of high-quality investment targets: Top players in this segment boast solid technical credentials, founders with both big tech and academic experience, technology validated by top industry benchmarks, proven deployments across multiple use cases, existing partnerships with leading clients, and multiple rounds of capital backing, resulting in high growth certainty. These make ideal targets for platform onboarding.

3. Trend and risk notes: The broader large model space is crowded with conceptual, unproven projects; niche segments focused on core underlying technology have far higher growth certainty. Onboarding players from this segment allows platforms to effectively mitigate the risk of backing unviable conceptual projects. Leading cloud platform Huawei Cloud has already formed a strategic partnership with a top player in this space; other platforms can follow suit to capture first-mover advantage in ecosystem building.

This article presents the latest industry developments in the large model space, offering high research value for AI industry analysts. Key takeaways are as follows:

1. New industry development: The large model industry has now entered a phase of segmentation and specialization, with underlying infrastructure gradually splitting into independent verticals. Long-term memory has evolved from an auxiliary function of large models to a stand-alone vertical segment, spawning dedicated startups that have already won strong capital backing—one startup completed three hundreds-of-millions-yuan financing rounds just three months after founding, reflecting extremely high capital confidence in the segment.

2. Definitive market size data: Industry research institutions project that China's agent memory market will achieve a CAGR exceeding 110% over the next five years, reaching 64.25 billion yuan in 2030, when it will account for nearly 26% of all agent infrastructure to become the second-largest independent infrastructure market after foundation models. This data clearly validates the commercial potential of the segment.

3. New technology and business model trends: The segment has already established a clear business model of "core underlying R&D + standardized memory module output + B/C end client coverage across multiple scenarios". It has solved the long-standing industry pain points of long-context memory retention and excessive token consumption, and already formed partnerships with leading clients across multiple sectors, making it a compelling case study for new industry incubation 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.

据光源资本披露,AI记忆基础设施创企深圳忆纪元科技有限公司(MemoraX AI)近日完成数亿元种子++轮融资,由北洋海棠基金、尚势资本、仁爱资本联合领投,光源资本担任独家财务顾问。这是MemoraX AI继今年4月种子轮、5月种子+轮融资后,短时间内斩获的第3轮大额融资。

本轮融资将重点投入大模型长期记忆底层核心技术攻坚,并推进B端、C端交互式产品的记忆模块标准化研发与规模化商业化落地。

MemoraX AI成立于2026年3月,专注于大模型长期记忆能力构建,其创始人兼CEO为现任天津大学智算学部教授、博士生导师、前华为诺亚方舟决策与推理实验室主任郝建业。

据天津大学深度强化学习实验室郝建业主页,他本科毕业于哈尔滨工业大学,博士毕业于香港中文大学,研究方向聚焦博弈智能。

业务方面,MemoraX AI构建了以“可学习记忆策略引擎、记忆基模、自演进Agent Harness”为核心的Agent记忆技术体系,在文本记忆领域形成超长上下文和时序推理记忆能力突破,在多模态记忆权威榜单Mem-Gallery和ATM-Bench上取得第一,并且Token消耗降低30%。

面向编程效率工具场景,MemoraX AI推出了面向Coding Agent的记忆产品,让Codex、Claude Code等编程Agent能持续理解项目上下文,在复杂、长程任务中减少重复解释和上下文断档。

面向交互式娱乐场景,MemoraX AI记忆引擎在上千轮次连续交互中,可实现动态记忆沉淀与召回,保障角色、关系与世界状态的长期稳定性与一致性。

目前该公司已与华为云正式签署战略合作,成为入选华为云大模型战略合作厂商中唯一聚焦长期记忆赛道的企业。此外,MemoraX AI已与医疗健康、智能家居、智能穿戴及具身智能等领域头部客户开展合作。

根据行业研究机构爱分析发布的《2026中国智能体记忆市场研究报告》,中国智能体记忆市场规模预计将从2025年的14.4亿元增长至2030年的642.5亿元,年复合增长率超过110%。到2030年,记忆在智能体基础设施中的占比将从不足10%提升至接近26%,成为仅次于模型的第二大独立基础设施市场。


亿邦持续追踪报道该情报,如想了解更多与本文相关信息,请扫码关注作者微信。

本文首发于 亿邦动力 官方网站

亿邦持续追踪报道该情报,如想了解更多与本文相关信息,请扫码关注作者微信。

文章来源:亿邦动力

广告
微信
朋友圈

FAQ回顾

中国智能体记忆市场的规模增长趋势是怎样的?

中国智能体记忆市场规模预计将从2025年的14.4亿元增长至2030年的642.5亿元,年复合增长率超过110%。到2030年,记忆在智能体基础设施中的占比将从不足10%提升至接近26%,成为仅次于模型的第二大独立基础设施市场。

忆纪元科技的融资资金主要用于哪些方向?

融资资金将重点投入大模型长期记忆底层核心技术攻坚,同时推进B端、C端交互式产品的记忆模块标准化研发,以及相关技术产品的规模化商业化落地。

大模型长期记忆技术有哪些落地应用场景?

大模型长期记忆技术可落地于编程效率工具、交互式娱乐场景,还可应用在医疗健康、智能家居、智能穿戴及具身智能等领域,能提升长程任务处理效率,保障交互场景下状态的长期稳定一致。

忆纪元科技的大模型记忆技术有什么核心优势?

忆纪元科技构建了三大核心支撑的Agent记忆技术体系,在文本记忆领域实现超长上下文和时序推理记忆能力突破,在多模态记忆权威榜单Mem-Gallery和ATM-Bench上排名第一,Token消耗可降低30%。

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