广告
加载中

杭州冲出一个芯片IPO 背后是“龙芯之父”的亲哥哥

薛皓皓 2026-06-05 15:14
薛皓皓 2026/06/05 15:14

邦小白快读

EN
全文速览

本文核心介绍了即将冲刺港股IPO的国产芯片企业力积存储的创业历程、发展现状与业绩情况,核心干货如下:

1. 力积存储2020年成立于浙江,核心业务是利基型DRAM芯片设计销售,采用无晶圆厂模式,代工主要依托中国台湾力积电,核心技术来自并购的日本老牌存储设计公司ZentelJapan,直接获得从110纳米到25纳米六代制程技术,跳过了漫长研发试错期。

2. 创始人应伟是资深投资人,也是“龙芯之父”胡伟武的亲哥哥,他抓住美国制裁后DRAM国产化的窗口,通过跨界并购拿下成熟技术入场券,不同于常规从零组建团队的芯片创业路径。

3. 2025年力积存储营收11.05亿元,同比增长70.9%,成功扭亏为盈,按利基型DRAM收入计,国内无晶圆厂企业排第四,全球排第十二,此次IPO募资将用于研发扩团队、产能建设和并购。

本文为存储领域品牌商提供了产业趋势参考与发展路径干货,具体如下:

1. 市场与消费趋势:当前全球三星、SK海力士、美光三大存储巨头,纷纷将产能转向高端主流DRAM,逐步退出利基型DRAM市场,给国产厂商留下巨大替代空间,利基型DRAM占全球DRAM市场7%,主打定制化、高可靠性,面向汽车、工业、医疗等增量需求赛道,增长空间广阔。

2. 发展路径参考:力积存储避开巨头锋芒,深耕利基赛道的定位值得后进品牌参考,通过并购成熟技术快速切入赛道,绑定上游核心代工厂获得优先产能保障,构建了自身核心护城河。

3. 经营与人才参考:力积存储通过拓展内存模组业务优化收入结构,逐步降低供应链和客户集中度分散风险,针对外籍核心技术人才搭建离岸持股平台做股权激励,解决了行业人才绑定的痛点,经验可借鉴。

本文针对存储芯片领域从业者,整理了相关机会、风险和可学习干货如下:

1. 市场机会:当前全球存储行业格局变化叠加美国对中国半导体制裁,DRAM国产替代进入历史性窗口,利基型DRAM赛道缺口大,随着汽车、工业、医疗电子行业发展,细分需求持续增长,是国产存储玩家的入场好时机。

2. 可学习的创业与经营经验:芯片创业不一定需要从零组建团队研发,可以通过整合并购成熟海外技术资产快速切入赛道,缩短研发周期降低试错成本;创业后要主动绑定上游核心代工厂,获得稳定产能支持,构建自身核心壁垒。

3. 风险提示:要警惕存储行业的周期性波动,同时要规避供应链和客户过度集中的风险,力积存储三年间将前五大供应商采购占比从88.6%降至68.7%,前五大客户收入占比从66.8%降至43.4%,这种分散风险的做法值得参考。

4. 资本机会:当前产业资本、地方国资都十分看好半导体国产化项目,融资支持力度大,国产存储项目更容易获得资本加持。

本文为芯片领域相关工厂整理了需求、机会和发展启示干货,具体如下:

1. 产品需求与商业机会:利基型DRAM侧重定制化、长生命周期支持和高可靠性,面向汽车、通讯、工业、医疗等多个领域,当前三大国际巨头逐步退出该赛道,国产替代缺口大,给国内配套的封测厂、模组组装厂带来大量稳定订单机会。

2. 企业转型启示:中国台湾力晶原是DRAM制造商,2012年因行业价格崩盘负债千亿被迫退市,后来通过断臂转型晶圆代工,五年还清债务,还打造出力积电、晶合集成两大上市代工厂,这种困境转型的经验值得国内传统制造工厂参考。

3. 合作机会:力积存储此次IPO募资将用于扩产能、拓产品线,还计划开展战略性投资与收购,国内符合技术要求的存储相关生产工厂,可以对接寻求合作机会。

4. 技术布局提示:力积存储已经突破面向AI和高性能计算的高带宽堆叠DRAM(WoW 3D异构集成)技术,未来相关生产需求会逐步释放,相关工厂可以提前布局配套生产能力,抓住技术红利。

本文为半导体领域服务商整理了行业趋势、客户痛点和相关干货,具体如下:

1. 行业发展趋势:当前DRAM国产化进程加速,越来越多国内玩家切入利基型DRAM赛道,这类企业对技术服务、资本服务、人力资源服务、合规服务的需求快速增长,给服务商带来新的业务增量。

2. 技术发展方向:除了主流的高端DDR5、HBM技术,利基市场的技术升级、面向AI场景的WoW 3D异构集成存储技术已经取得研发突破,未来会成为重要的技术方向,相关技术服务商可以提前布局。

3. 客户痛点与解决方案参考:当前国产芯片设计企业的核心痛点分别是缺成熟技术积累、缺稳定上游产能、难绑定外籍核心技术人才,对应的成熟解决方案分别是跨界并购海外成熟技术资产、绑定头部代工厂获取产能保障、搭建离岸持股平台实现外籍人才股权激励,可供服务商做方案参考。

4. 客户开发方向:当前不少地方政府积极引入半导体项目落地,冲刺IPO的国产芯片企业对投行、法务、财务等服务需求旺盛,服务商可以针对性拓展客户。

本文为半导体产业相关平台商整理了招商、运营、风险规避相关干货,具体如下:

1. 企业需求:国产存储芯片企业发展过程中,需要产业资本背书、稳定产能支持、地方政策配套,还需要合规方案解决非大陆籍技术人才的股权激励问题,对平台的综合产融服务能力要求较高。

2. 招商方向:利基型DRAM是当前国产替代的热门赛道,这类项目技术成熟、增长速度快,带动性强税收贡献大,是非常优质的招商项目,平台可以针对性对接引入,优化本地产业结构。

3. 风险规避提示:存储行业有明显的周期性波动,项目发展前几年普遍会陷入亏损,平台投资和引入项目时要做好长期布局的准备,同时要提醒企业分散供应链和客户集中度风险,避免被单一环节卡住发展。

4. 可参考的操作模式:杭州钱塘国资通过Pre-IPO轮入股的方式,引入力积存储将总部落地钱塘区,既支持了本土半导体产业发展,也获得了资本收益,这种产融结合引入项目的做法,值得其他产业平台参考。

本文为半导体产业研究者提供了新的产业动向、商业模式和研究素材,干货如下:

1. 产业新动向:国内DRAM国产化出现了全新的创业路径,不同于传统从零开始研发积累技术的模式,力积存储通过跨界并购整合海外成熟技术资产,整体平移到国内发展,快速实现技术落地和业绩增长,为半导体国产化提供了新的可行路径。

2. 创新商业模式:力积存储采用Fabless模式,深度绑定上游力积电获得技术和产能支持,同时构建了包含创始人团队、产业资本、地方国资、财务投资的多元股权结构,兼顾了技术、产能、资本多维度支持,这种模式适合细分赛道后进芯片企业参考,具备研究价值。

3. 产业格局新变化:当前全球存储产业竞争格局重构,三大国际巨头逐步退出利基型DRAM市场,该细分赛道不再是边缘市场,反而成为国产存储实现突破的重要切入点,改变了以往国产存储从高端突破的单一思路。

4. 行业新问题解决案例:如何合法合规绑定外籍顶尖技术人才,是芯片行业的共性问题,力积存储搭建离岸持股平台的实践,为解决该问题提供了可研究的现实案例。

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 provides an overview of Pmei Memory, a Chinese domestic DRAM chipmaker that is preparing for its Hong Kong IPO, covering its founding journey, current operations and financial performance. Key takeaways are as follows:

1. Founded in Zhejiang in 2020, Pmei focuses on the design and sales of niche DRAM chips, operating a pure-play fabless model. It primarily relies on Taiwan-based Powerchip Semiconductor Manufacturing (PSMC) for manufacturing. Its core technology comes from the acquisition of Zentel Japan, a long-established Japanese memory design firm, through which it acquired six generations of process technology ranging from 110nm to 25nm, skipping a lengthy period of R&D and trial-and-error.

2. Its founder Ying Wei is a veteran investor and the elder brother of Hu Weiwu, known as the "Father of Loongson". Ying seized the window for DRAM localization created by U.S. sanctions, gaining access to mature technology via cross-border acquisition — a fundamentally different path from the conventional chip startup model of building a team from scratch.

3. In 2025, Pmei posted 1.105 billion yuan in revenue, up 70.9% year-on-year, and turned a net profit for the first time. Ranked by niche DRAM revenue, it is the fourth-largest domestic fabless player and the 12th-largest globally. Proceeds from its IPO will be used for R&D expansion, capacity building and future acquisitions.

This article shares industry trend insights and development path takeaways for storage brands, as outlined below:

1. Market and consumer trends: The "Big Three" global memory players — Samsung, SK hynix and Micron — are all shifting production capacity to high-end mainstream DRAM and gradually exiting the niche DRAM segment, creating huge substitution space for domestic Chinese manufacturers. Niche DRAM accounts for 7% of the global DRAM market, characterized by custom specifications and high reliability, and serves growing end markets including automotive, industrial and medical applications, offering substantial room for growth.

2. Development path reference: Pmei’s strategy of avoiding direct competition with industry giants and focusing on the niche DRAM segment is a valuable reference for late-entrant brands. By acquiring mature technology to quickly enter the market and securing priority capacity access by partnering with a core upstream foundry, the company has built strong competitive moats.

3. Operational and talent management insights: Pmei optimized its revenue structure by expanding into memory module business, gradually reducing supply chain and customer concentration to diversify risk. It also addressed the common industry challenge of retaining core foreign technical talent by setting up an offshore shareholding platform for equity incentives, a practice other brands can learn from.

This article summarizes opportunities, risks and actionable insights for practitioners in the memory chip industry, as follows:

1. Market opportunity: The ongoing restructuring of the global memory industry landscape, combined with U.S. semiconductor sanctions on China, has created a historic window for domestic DRAM substitution. The niche DRAM segment faces large unmet demand, and growing demand from automotive, industrial and medical electronics has created an ideal entry point for domestic memory players.

2. Actionable startup and operational insights: Chip startups do not need to build R&D teams and technology from scratch. Acquiring mature overseas technological assets through integration and M&A allows players to enter the market quickly, shortening development cycles and cutting trial-and-error costs. After launching, companies should proactively partner with core upstream foundries to secure stable capacity and build core competitive barriers.

3. Risk warning: Companies should stay alert to the cyclical volatility of the memory industry and avoid excessive concentration of suppliers and customers. Pmei reduced procurement from its top five suppliers from 88.6% to 68.7%, and cut revenue from its top five customers from 66.8% to 43.4% over three years — this risk diversification approach is a good reference for the industry.

4. Capital opportunity: Industrial capital and local state-owned assets are now strongly bullish on semiconductor localization projects, offering extensive financing support, making it easier for domestic memory projects to secure capital backing.

This article compiles demand, opportunity and development insights for chip-related manufacturing factories, as follows:

1. Product demand and business opportunities: Niche DRAM prioritizes customization, long lifecycle support and high reliability, serving automotive, communications, industrial, medical and many other sectors. With the three international giants gradually exiting this segment, large substitution gaps have opened up, bringing substantial stable order opportunities for domestic supporting packaging & testing facilities and module assembly factories.

2. Enterprise transformation insights: Taiwan’s Powerchip was originally a DRAM manufacturer. It was forced to delist in 2012 with over 100 billion yuan in debt after an industry price crash. It later refocused on wafer foundry services through a major restructuring, paid off all its debt in five years, and built two listed foundries — PSMC and Nexperia Integrated. This turnaround experience offers valuable lessons for traditional Chinese manufacturing factories facing difficulties.

3. Cooperation opportunities: Pmei’s IPO proceeds will be used for capacity expansion, product line expansion and strategic investments and acquisitions. Domestic memory-related manufacturing factories that meet Pmei’s technical requirements can reach out to explore partnership opportunities.

4. Technology layout tips: Pmei has already developed its own high-bandwidth stacked DRAM technology (WoW 3D heterogeneous integration) for AI and high-performance computing, and related production demand will gradually release going forward. Relevant factories can pre-position supporting production capacity to capture upcoming technological dividends.

This article organizes industry trends, client pain points and key insights for semiconductor service providers, as follows:

1. Industry development trends: Domestic DRAM localization is accelerating, with a growing number of Chinese players entering the niche DRAM segment. These companies have rapidly growing demand for technical services, capital services, human resources services and compliance services, creating new business growth opportunities for service providers.

2. Technology development direction: Beyond mainstream high-end DDR5 and HBM technologies, technical upgrades for the niche market and AI-oriented WoW 3D heterogeneous integrated memory technology have already achieved R&D breakthroughs, and will become an important technology direction going forward. Relevant technical service providers can prepare and position for this trend in advance.

3. Client pain points and solution references: The core pain points for domestic chip design firms are the lack of mature technology accumulation, the lack of stable upstream capacity, and the difficulty of retaining core foreign technical talent. The proven solutions are: acquiring mature overseas technological assets via cross-border M&A, partnering with leading foundries to secure capacity, and setting up an offshore shareholding platform to deliver equity incentives for foreign talent — these can serve as references for service providers building client solutions.

4. Client development direction: Many local governments are actively attracting semiconductor projects, and domestic chip companies preparing for IPO have strong demand for investment banking, legal and financial services. Service providers can target this segment for client expansion.

This article organizes insights on investment promotion, operations and risk mitigation for semiconductor industry platforms, as follows:

1. Enterprise demand: As domestic memory chip companies scale, they need industrial capital endorsement, stable capacity support, supporting local policies, and compliance solutions to deliver equity incentives to non-Chinese technical talent — all of which require strong integrated industrial and financial service capabilities from industry platforms.

2. Investment promotion focus: Niche DRAM is a high-potential segment for domestic substitution. Projects in this space have mature technology, fast growth, strong industrial spillover effects and solid tax contributions, making them high-quality investment targets. Platforms can actively target and attract such projects to optimize local industrial structure.

3. Risk mitigation tips: The memory industry has clear cyclical volatility, and projects typically operate at a loss for their first several years. Platforms need to prepare for long-term布局 when investing in and attracting projects, and advise portfolio companies to diversify supply chain and customer concentration to avoid getting stuck from over-reliance on a single link.

4. Reference operating model: Hangzhou Qiantang State-owned Assets invested in Pmei in its pre-IPO round and attracted the company to locate its headquarters in Qiantang District. This model both supports the development of the local semiconductor industry and generates capital gains, offering a reference for other industrial platforms looking to attract projects through industry-finance integration.

This article provides updates on industry dynamics, business models and research material for semiconductor industry researchers, with key takeaways as follows:

1. New industry dynamics: A new startup path has emerged for domestic DRAM localization. Unlike the traditional model of developing technology from scratch, Pmei acquired and integrated mature overseas technological assets via cross-border M&A, then relocated the full technology portfolio to China, achieving rapid technology commercialization and revenue growth. This offers a new viable path for semiconductor localization.

2. Innovative business model: Operating a fabless model, Pmei has deep ties with upstream PSMC to secure technology and capacity support, while maintaining a diversified ownership structure including the founding team, industrial capital, local state-owned assets and financial investors. This structure balances support across technology, capacity and capital, making it a valuable research subject and a reference for late entrants to niche chip segments.

3. Shifts in industry landscape: As the global memory industry restructures, the three international giants are gradually exiting the niche DRAM segment, transforming this once-marginal segment into a key entry point for domestic memory players to break through. This shifts the previous conventional wisdom that Chinese memory players must first target the high-end market.

4. Case study for a common industry problem: Legally and compliantly retaining top foreign technical talent is a shared challenge across the chip industry. Pmei’s practice of setting up an offshore shareholding platform provides a real-world, researchable case study for solving this problem.

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。

作者丨薛皓皓

编辑丨关雎

浙江金华走出的投资人应伟,用一场跨越两岸三地的并购,把一家日本老牌存储芯片设计公司的技术资产整体平移进了中国大陆。这家公司,就是2020年成立的力积存储。

力积存储的核心业务,是利基型DRAM芯片的设计与销售,采用Fabless(无晶圆厂)模式,晶圆代工主要依托中国台湾的力积电完成。力积电全称力晶积成电子制造股份有限公司,是台湾仅次于台积电、联电的第三大专业晶圆代工厂。

力积存储的技术基因,来自2020年并购整合的日本ZentelJapan——这家由力晶体系孵化的设计公司,为力积存储带来了从110纳米到25纳米六代制程的完整产品迭代能力,让公司直接跳过了漫长的研发试错期。

2025年,力积存储实现营收11.05亿元,同比增长70.9%,经调整净利润2609万元,成功扭亏为盈。按2025年利基型DRAM收入计,公司在中国无晶圆厂企业中排名第四,全球排名第十二。

IPO前,应伟及其一致行动人合计控制公司约44.52%的股份;主要外部股东包括龙芯系(10.13%)、力晶创投(10.1%)、么贵师控制的天津灏鑫(8.45%)、杭州钱塘国资旗下杭州和达(4.06%)及鼎晖投资(3.32%)。

始于并购的创业

在芯片行业的创业故事里,大多是技术大牛出走大厂,或是海归博士带着专利归国。但力积存储的故事,却始于一位投资老兵的跨界并购。

现年59岁的应伟,身上的标签一直是“投资人”。作为鼎晖投资的执行事务合伙人,他曾主导过对神州细胞、微宏动力等明星科技企业的投资。但他最广为人知的一笔投资,是投出了龙芯中科(688047.SH)。龙芯中科是国内唯一基于完全自主指令集(LoongArch/龙架构)实现CPU设计、IP核、操作系统全栈自研的通用处理器企业。

龙芯中科的创始人、董事长胡伟武,被誉为“龙芯之父”。鲜为人知的是,应伟正是胡伟武同父同母的亲哥哥——兄弟俩出生在浙江永康的教师家庭,按照当地“兄随父姓、弟随母姓”的传统,哥哥应伟随父姓,弟弟胡伟武随母姓。在龙芯中科的发展历程中,应伟所在的鼎晖投资在B轮、C轮和D轮连续加注。

2019年,美国对中国半导体的制裁骤然升级,应伟敏锐地察觉到,在被国际巨头垄断的DRAM(动态随机存取存储器)市场,国产化正迎来历史性窗口。

他没有选择从零组建团队,而是直接买下了一张入场券。

通过一位圈内友人的牵线,应伟接触到了台湾内存IC设计公司爱普科技(AP Memory)。彼时,爱普科技正在进行战略转型,计划剥离旗下的标准型内存业务,专注高毛利的客制化AI和IoT存储。而这块被剥离的资产,正是日本老牌存储芯片设计公司——ZentelJapan。

ZentelJapan并非无名之辈。这家成立于2003年的公司,十余年间已经积累了从110纳米到25纳米六代制程的内存产品技术,产品线覆盖SDR到DDR4。更重要的是,ZentelJapan的背后,站着台湾半导体界的传奇人物——黄崇仁,以及他一手打造的“力晶”帝国。

黄崇仁被台湾媒体称为半导体界的“九命怪猫”。他拥有医学博士学位,却跨界创办了力晶半导体,成为台湾最早一批DRAM制造商,并在台湾上柜、卢森堡上市。2012年,力晶因DRAM价格崩盘负债1200亿新台币,被迫退市。

但黄崇仁硬是通过断臂求生和转型晶圆代工,在五年内还清了债务。通过对业务进行分割重组,公司更名为力晶创新投资控股(简称“力晶创投”),成为控股母公司,旗下则有力积电(台湾上市)和晶合集成(A股上市)两大晶圆代工厂。

而ZentelJapan,正是力晶体系在2003年亲手孵化的设计团队,其“Zentel”品牌也是由力积电创立。

应伟看中的,正是这层打断骨头连着筋的关系。他不仅要买下ZentelJapan,还要把力晶创投拉入局中。

2019年底,应伟说服了力晶创投和爱普科技,三方达成了一项巧妙的交易架构:应伟承诺最终将100%收购ZentelJapan,将其引入中国大陆发展DRAM业务。

作为第一步,2020年1月,力晶创投以1900万美元收购ZentelJapan 76%的股权,应伟则通过全资子公司EaglestreamHK拿下剩余24%。

同年3月,力积存储在浙江金华正式成立。通过一系列眼花缭乱的股权腾挪,力晶创投以“0代价”换取了力积存储约10%的股权,而ZentelJapan的核心业务、品牌和技术团队,则被完整地平移到了力积存储的体内。

这场跨越两岸三地的并购,为力积存储带来了立竿见影的业绩爆发。

ZentelJapan成熟的利基型DRAM技术,让力积存储直接跳过了漫长的研发试错期,成为国内少数拥有完整内存产品迭代能力的设计公司。

利基型DRAM是相对主流DRAM来说的。主流DRAM产品具有制程新、容量大、传输速率高的特点,主要应用于智能手机、个人计算机、服务器等大规模标准化电子设备。利基型DRAM依赖成熟技术,注重产品的定制化、更优的性价比、长生命周期支持和可靠性,满足汽车、通讯、工业应用、医疗设备等多样化市场需求。

目前市场上的主流DRAM产品为容量8Gb以上的DDR4以及后代DRAM,容量8Gb及以下的DDR4和前代DRAM则被视为利基DRAM。

在商业模式上,力积存储延续了ZentelJapan的Fabless模式,而晶圆代工的重任,自然落到了力晶创投旗下的力积电肩上。

力积电不仅是力积存储的唯一第三方晶圆代工厂,更在产能紧张时给予优先保障。ZentelJapan历史上开发的所有制程产品,基本都由力积电制造,这种深度的技术与产能绑定,构成了力积存储最核心的护城河。

在ZentelJapan技术的加持和力积电的产能输血下,力积存储的营收规模迅速扩大。2023年,公司营收达到5.8亿元。到了2025年,随着存储大市回暖,公司年营收飙升至11.05亿元,同比大增70.9%,并成功实现扭亏为盈。

但应伟的野心不止于吃老本。他很清楚,要让ZentelJapan的资产真正为中国本土的半导体突围服务,就必须留住那些掌握核心技术的“大脑”。

在这个极度依赖人才密度的行业里,如何用股权绑定外籍顶尖专家,是一个极具挑战性的命题。应伟在2025年4月专门设立一个名为“EaglestreamPartners”的离岸架构。

这个量身定制的平台,正是力积存储专为非中国籍雇员及顾问设立的持股平台。

通过EaglestreamPartners,像曾任职SK海力士约30年的韩国籍专家金峻虎、拥有超25年DRAM研发经验的日本籍专家久保贵志等外籍核心骨干,能够合法合规地分享到力积存储IPO的资本红利。

这不仅是对他们带领团队向19纳米先进制程和WoW 3D堆叠技术冲锋的物质激励,更是力积存储在激烈的中日韩半导体人才争夺战中,打出的一张诚意的底牌。

兄弟一路支持,杭州国资压阵

从2020年成立至今,力积存储共完成四轮融资。

最早下注,也是支持力度最大的产业投资方,当属“龙芯系”。2020年12月,力积存储刚成立不久,龙芯立积与龙芯创信便在天使轮以4.40元/股的成本,合计投资6500万元。此后,龙芯系几乎参与了公司从天使轮到Pre-IPO轮的全过程,累计投资约2.535亿元。IPO前,龙芯系合计持有力积存储超过10.13%的股份。

除了龙芯系,另一家重要的产业背景股东是力晶创投,持有力积存储10.1%的股份。

在财务投资机构中,最引人注目的莫过于应伟的“老东家”鼎晖投资。2025年4月至5月的Pre-IPO轮中,鼎晖投资通过天津鼎晖等主体,以13.29元/股的成本,合计斥资约1.19亿元突击入股。

此外,在早期支持者中,由独立投资人么贵师控制的林芝鼎孚和天津灏鑫扮演了重要角色。作为联合创始股东,么贵师在公司成立之初出资950万元,目前通过林芝鼎孚和天津灏鑫合计持有公司8.5%的股份。

国资平台的加入,特别是浙江本土国资的接力,为力积存储的半导体国产化故事增添了重要注脚。2022年5月的A轮融资中,具有深圳国资委背景的盈富泰克投资7500万元。而随着力积存储的战略重心向浙江倾斜,浙江国资开始发挥关键作用。

2020年,力积存储在浙江金华起步,金华市国资委曾短暂入股。2025年6月,力积存储将注册地从金华迁至杭州,并在同年9月将总部正式落地杭州钱塘区。

杭州国资迅速跟进。2025年5月,杭州钱塘新区管委会最终控制的杭州和达,在Pre-IPO轮以1亿元入股,成为该轮最大单笔投资。同期,浙江本土头部券商浙商证券旗下的浙商佳富也投资2500万元。

在IPO前,公司最终控制人应伟及其一致行动人合计控制公司约44.52%的股份。主要外部投资者持股比例如下:龙芯系(10.13%)、力晶创投(10.1%)、天津灏鑫(么贵师控制,他和应伟一同投资了龙芯中科,8.45%)、杭州和达(杭州钱塘新区管委会,4.06%)、鼎晖投资(3.32%)。

等来存储超级周期

2023年,全球半导体行业正处于DRAM两年下行周期的末期,需求不振。高价库存与低价甩卖的“剪刀差”,让力积存储陷入亏损。2023年至2024年,公司分别实现营收5.8亿元、6.5亿元,毛利率经历了从3.7%到9.3%的艰难爬坡,同期净亏损分别为2.44亿元、1.09亿元。

但市场的风向在2025年迎来了反转。随着AI需求爆发和行业库存去化,存储芯片迎来超级周期。力积存储在2025年实现营收11.05亿元,同比大增70.9%;毛利率更是飙升至16.1%。公司在这一年虽然依然净亏损1054万元,但实现了经调整净利润转正,达到2609万元。

业绩狂飙的背后,是力积存储在业务结构上的调整。其内存芯片业务收入占比从2023年的83.5%一路下滑至2025年的55.0%;而2022年才启动的内存模组业务,收入占比从2023年的8.1%跃升至2025年的38.8%。

力积存储的供应商和客户正逐渐多元化。

在供应链端,2023年至2025年,力积存储向前五大供应商的采购额占比分别为88.6%、76.2%及68.7%,逐步降低。力积电是其唯一的第三方晶圆代工商,且保持为第一大供应商,但向力积电的采购额占比从2023年的49.1%降至2025年的25.8%。随着力积存储收入结构的调整,对内存模块组装公司威刚集团的采购额占比不断上升,2025年达到25.6%,威刚集团成为第二大供应商。此外,台湾的福懋科技及南茂科技等厂商把控着其封装测试环节。

在客户端,报告期内,前五大客户收入占比分别为66.8%、52%和43.4%。2024年,位于深圳和香港的蓝海集团跃升为第一大客户,销售额从2023年的3610万元激增至8751万元。2025年,电子元件贸易商创威电子则以1.7亿元的销售额成为第一大客户。

要在巨头林立的存储赛道活下去,力积存储选择了一条“避其锋芒”的道路——深耕利基型DRAM市场。

在这个规模约占全球DRAM市场7.0%(按2025年销售收入计)的细分赛道里,竞争格局正在发生剧变。三星、SK海力士和美光三大原厂正加速将产能向HBM和DDR5倾斜,逐步退出利基市场。这为国产厂商留下了巨大的替代空间。

根据弗若斯特沙利文数据,按2025年利基DRAM收入计,力积存储在中国无晶圆厂公司中排名第四,在全球所有参与者中位列第十二(市占率1.0%)。排在其前面的,不仅有台湾地区的华邦电子、南亚科,还有国内全品类存储龙头兆易创新和北京君正等强劲对手。

为了打破“旧技术、小市场”的刻板印象,力积存储正在押注未来,已取得高带宽堆叠DRAM(WoW 3D异构集成)技术的研发突破。该技术特别适合人工智能及高性能计算应用,通过最大限度地减少内存与处理单元之间的数据移动,增强了实时推理和训练性能,使其成为数据中心和延迟敏感型工作负载的理想选择。

此次冲刺港股IPO,力积存储明确表示,募集资金将主要用于提升研发及创新能力(扩充高带宽存储产品及内存产品研发团队)、提升生产和测试设备及能力、提升全球销售及营销能力,以及战略性投资与收购。

注:文/薛皓皓,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。

文章来源:创业邦

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