文 | 医药研究社
近期，多家医药上市公司发布参设、参投或认购基金进展公告。从资金投向来看，部分医药及大健康类上市公司从“投药”部分转向“投芯片、投AI”。
亿帆医药全资子公司合肥欣竹生物科技有限公司出资1亿元做有限合伙人，参与设立天津高榕长赢股权投资基金合伙企业（有限合伙），公告对投向的表述是“主要投向新兴产业领域”；特宝生物以1亿元认购天津砺思星深股权投资合伙企业（有限合伙）份额，该基金聚焦人工智能，这笔投资同样因与关联人共同投资构成关联交易。时间再往前一点，保健食品企业汤臣倍健以5000万元投资原粒（北京）半导体技术有限公司，持股0.97%，公司定性为财务性投资，标的仍在亏损，因董事长梁允超配偶间接持股，构成关联交易。
说“药企不投药”当然夸张，但钱在分流是真的。比“投了什么”更值得看的，是这钱为什么此时、以此种方式挪过去。
钱变紧了，药也就没那么好投了
药企把钱挪向芯片和AI，不是因为突然对芯片有了兴趣，而是主业这头的资本水位在下降。集采推进，仿制药利润空间被压薄，靠成熟产品维持增长的路径越走越窄；创新药虽有政策支持，但周期长、不确定性高，短期补不上腾出的份额。
官方数据已经写明这种压力。据国家统计局数据，2025年全国规模以上医药制造业营业收入24870亿元，同比下降1.2%；利润总额3490亿元，同比增长2.7%；营业成本下降1.3%，降幅略大于收入。利润的这轮改善出在成本端，不在收入端。
更能说明问题的是固定资产投资：2025年医药制造业固定资产投资同比下降13.5%，同期全国制造业投资增长0.6%。新建车间、上产线的钱在缩，这是全行业口径的硬收缩，不是某家公司的选择。
盘子在收，压力落到单个公司就是业绩换挡：亿帆医药2026年上半年营业收入24.68亿元，同比下降6.36%，归母净利润2.05亿元，同比下降32.48%，正处在这个节点。此时把部分自有资金配置到高景气赛道，既补财务回报，也分散单一主业的波动风险。
另一头，人工智能与半导体的资本市场热度摆在那里。据官方公开数据显示，AI制药市场规模预计2026年达29.94亿美元。药企对AI并不陌生，药物研发数据密集与AI能力边界有交集，这让“投科技”比纯粹跨界并购更容易被董事会和投资者接受。战略转型、资本周期与政策红利确实同时在场，但最直接的推力还是主业端的资本收缩。
钱挪出去，走法不一样
不同的投法，决定了这笔钱离主业有多远。
最常见的是参投私募基金，间接持有敞口。亿帆医药、特宝生物、鱼跃医疗多走这条路，单笔多在数千万元至一亿元之间。钱交到管理人手里，参与有限、下注分散，不改变主业结构，更像资产配置而非战略转向。
也有企业以自有资金直接参股特定芯片公司。汤臣倍健投资原粒半导体即属此类，持股0.97%，标的2025年净亏损6295.94万元、2026年一季度续亏1091.52万元。这种投法有了直接敞口，但持股低、话语权有限，本质仍是“用一点钱买一个观察窗口”。
最激进的走法，是通过并购把半导体资产装进上市公司。向日葵推进得最深：2025年12月26日，深交所就重组预案下发关注函，要求说明标的兮璞材料的产能分布、业务实质、关联交易及实控人所涉诉讼；2026年1月14日重组终止，同日收到证监会《立案告知书》；2026年4月8日，公司收到浙江证监局《行政处罚决定书》，对公司及两名时任高管罚款合计510万元。当“投科技”上升到控制权层面，信息披露颗粒度和标的成色就成了硬约束。
三条路径与主业的距离差别很大。基金参投和少数股权投资，钱出去了，主业还在；并购意味着战略重心实质偏移，整合难度与监管审视不在一个量级。另一条线索并不喧哗：百普赛斯出资1.5亿元参设基金，投向明确落在生命科学工具与服务、生物医药、医疗器械、生物制造。
最能说明分野的是同一家公司的两笔钱。鱼跃医疗全资子公司鱼跃（香港）控股有限公司出资9490万港元，认购CFTC Summit Fund SPC旗下基金份额，投向明确写着人工智能、半导体芯片与集成电路；另一边，公司早在2023年就参与设立南京盛宇黑科新智造创业投资合伙企业（有限合伙），目前认缴6000万元，方向为“以医疗器械产业链为核心”。一个往远走，一个往近补。
药企的钱并非一致地“离开医药”，而是在重新权衡回报周期与确定性——有的挪得远，有的挪得近，算盘也不一样。
挪出去的钱，能不能挪回能力
短期看，药企以LP身份进入AI与芯片基金，对行业格局的直接改变有限。几千万元到一亿元，在半导体和算力赛道算不上决定性力量，对营收结构的贡献短期也难量化。信号意义大于财务意义：传统药企的现金流在寻找主业之外的对冲工具，科技资产的高估值与活跃的交易叙事提供了这个选项——尽管私募基金份额流动性并不高。
但有两重边界，不该被热闹盖过去。
AI对医药主业的“反哺”不会自动发生。公开可查的行业观点普遍认为，AI在药物发现早期能缩短临床前研究时间；但进入Ⅱ期临床之后，它是否真的优于传统研发路径，尚无足够规模的公开临床样本给出结论。换句话说，AI能够更快筛选出候选分子，但目前尚未被证明能把更多分子送上市场。投资芯片和算力，与提升研发效率之间，并不存在一条笔直的因果链。钱挪出去了，能力能不能挪回来，是另一回事。
监管对“跨界叙事”的态度已经用罚单给出答案。向日葵被罚，是因为重组预案称标的兮璞材料采取“定制化代工+自主生产”相结合的模式，监管查明截至披露日该标的自有工厂仍在建、不具备自主生产能力，主要产品实为标准化产品——构成误导性陈述，公司及时任董事长、董秘合计被罚510万元。亚辉龙逻辑相似：2026年3月17日，深圳证监局下发《行政处罚决定书》，认定公司三次公告披露，均未准确完整说明合作方脑机星链的技术路径与产品开发阶段，对公司及时任董事长、董秘作出合计750万元的罚款处罚。两起案例指向同一取向：新赛道的故事要经得起产能、团队、资金和标的真实价值的逐项追问，以财务投资之名行市值管理之实的空间正在收窄。
对药企来说，把部分资本配置到高景气赛道无可厚非，关键是分清“配置”与“转型”的界限。用基金份额换取认知窗口和财务回报，与把上市公司命运押到自身缺乏产业积累的领域，是性质不同的决策。钱往那边挪可以，挪出去之后能不能挪回、挪回对研发效率和生产数字化有无提升，才是值不值的问题。AI和芯片值得投，但药企最终的立足点仍然是药。
本文由 AI宝库自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Society for Medical Research
Recently, publish, a pharmaceutical company listed on the market, has announced the progress of the establishment, investment or subscription fund. In terms of financial investment, some of the pharmaceutical and health listed companies have shifted from “input” to “chip and AI”.
The announcement of a partnership (limited partnership) in which a billion-dollar pharmaceutical company, the All-Fail Pharmaceuticals Co. Ltd., contributed $100 million as a limited partner and participated in the establishment of a win-and-win investment fund (limited partnership) in Tianjin. The announcement stated that the investment was “mainly in new and emerging industries”; the specialty creature subscribed to the 100 million yuan share of Tianjin Sing Sing Deep-Sing (limited partnership) in a partnership (limited partnership) that focused on artificial intelligence, an investment that was also linked to joint investment with related persons. A further step forward, the health-care food company Tom Tentseng has invested 50 million yuan in the original (Beijing) semiconductor technology company, Ltd., with 0.97 per cent in shares, which is characterized as a financial investment and is still subject to losses, and constitutes a related transaction because the director of the board of directors, Liang Yoon Spouse, holds indirect shares.
It's an exaggeration to say, "Doctors don't throw drugs," but the money's been diverted. What is more worthwhile than “what” is why this money is being moved at this time and in this way.
The money's getting tight. The medicine's not so good.
The money was moved to chips and AI not because of sudden interest in chips, but because the capital level of the industry was declining. Pooling, where the profit space for imitating pharmaceuticals is being squeezed, and the path to growth sustained by mature products is becoming narrower; while policy-supported, innovative medicines are long-term, uncertain and short-term offsetting.
Official data already indicate this pressure. According to data from the National Statistical Office, in 2025, the national income from the pharmaceutical manufacturing industry and above was 24.8 billion yuan, a 1.2 per cent decrease over the same period; total profits of 349 billion yuan, a 2.7 per cent increase over the same period; and operating costs decreased by 1.3 per cent, slightly more than income. This round of profit improvements is at the cost and not at the income end.
More indicative of the problem is fixed-asset investment: in 2025, fixed-asset investment in the pharmaceutical manufacturing sector declined by 13.5 per cent over the same period, while national investment in manufacturing increased by 0.6 per cent over the same period. The cost of new workshops and production lines is shrinking, and this is a hard industry-wide contraction, not an option for a company.
The pressure on individual companies is a trade-off: $2.468 billion in the first half of 2026, or 6.36 per cent in the same year, and $205 million in the same year, or 32.48 per cent in the same year. At this point, part of the self-financing is allocated to the high-scenes track, both to compensate for financial returns and to spread the risk of fluctuations in a single major industry.
On the other side, the capital market heat of artificial intelligence and semiconductor is there. According to official public data, the size of the AI pharmaceutical market is projected to reach $2.994 billion in 2026. Drug companies are not new to AI, and there is an intersection of data-intensive drug research and development and AI capabilities, making “input technology” more acceptable to boards and investors than purely cross-border M&As. While it is true that strategic transformation, capital cycles and policy dividends are present at the same time, the most immediate thrust is capital contraction at the end of the industry.
The money's out. It's different.
Different voting methods determine how far this money goes from the main business.
The most common is participation in private fund-raising, with indirect exposure. Billions of sails, scavengers, fish jumps and medical treatments go this way, in the range of tens to 100 million dollars. Money is given to managers, participation is limited, the bets are decentralized, and the main business structure is not changed, more like asset allocation than a strategic shift.
There are also enterprises that directly participate in specific chip companies with their own funds. This is the case with the investment in the raw semiconductor, with a share of 0.97 per cent, with a net loss of $6.2959.4 million in 2025 and a continued loss of $1,0.0152 million in the first quarter of 2026. This method of voting has a direct opening, but it is low-share, limited-talking and essentially “buying an observation window with a little money”.
The most radical approach is to load semiconductor assets into listed companies through mergers and acquisitions. The most advanced was the sunflower: on 26 December 2025, a letter of concern was sent to the Institute regarding the proposed reorganization, requesting information on the distribution of capacity, the substance of the business, the related transaction and the proceedings involving the accused; on 14 January 2026, the reorganization was terminated and on the same day the CSRC filed its notice of case; on 8 April 2026, the company received the Zhejiang Censorship Authority ' s Decision on Administrative Penalties, which fined the company and the two senior managers at the time for a total of $5.1 million. When “technologies” rise to the control level, disclosure of information becomes hard bound by the size of particles and the colour of the target.
The distance between the three paths and the main industry varies considerably. The Fund ' s participation and a few equity investments, the money went out and the business was still in place; M&As implied a shift in strategic focus in substance, and the difficulty of integration was not at the same level as regulatory review. The other leads are not loud: 100 million yuan is spent on a fund for life science tools and services, biomedicine, medical equipment, and biomanufacturing.
The best illustration is two pieces of money from the same company. In 2023, the company was involved in the establishment of the New Smart Enterprise Investment Partnership (Little Partnership) in Jinyu, Nanjing, China, with a current contribution of $60 million, which is “at the heart of the medical equipment industry chain”. One goes far, one goes close.
The money of the pharmacist is not a consistent “leave the medicine”, but a rebalancing of the return cycle with certainty — some moving far, some moving closer, others not the same.
The money that's moved out. Can you get it back?
In the short term, pharmaceutical companies enter AI and chip funds as LPs, with limited direct changes in industry patterns. Tens to 100 million won are not a decisive force in semiconductor and calculator tracks, and the contribution to the revenue structure is short term difficult for quantization. Signals mean more than financial: traditional pharmacist cash flows are looking for hedge tools outside the main industry, an option offered by high valuations of technical and technological assets and active transaction narratives — albeit with a low private equity share.
But there is a two-fold border that should not be overridden.
It's not automatic for AI to “refrain” the pharmaceutical industry. There is general agreement in publicly available industry views that AI can shorten pre-clinical research in the early stages of drug discovery; however, after the second stage of clinical work, it is really better than traditional R & D paths, and there is not yet a sufficiently large open clinical sample to draw conclusions. In other words, AI was able to screen candidates more quickly, but has not yet proved to be able to put more molecules on the market. There is no straight causal chain between investing in chips and algorithms and improving R & D efficiency. It's a different thing to move the money out, to move the power back.
Regulatory attitudes to “cross-border narrative” have been answered by a ticket. The penalty for sunflower is due to the fact that the reorganisation of the pre-programmed material is based on a combination of “specified chemical sub-workers + self-production”, which regulates the identification of the target's own plant as of the date of disclosure, which is still under construction and does not have an autonomous production capacity, with the main product being a standardized product - - It constitutes a misleading statement that the company was fined a sum of $5.1 million for the timely appointment of its director and secretary. The Ahuilong logic is similar: on 17 March 2026, the Shenzhen Censorship Commission issued a Decision on Administrative Penalties, which concluded that the company ' s three public announcements disclosed that it had not provided an accurate and complete description of the technical path and product development phase of the cooperative brain chain, and imposed a fine of $7.5 million on the company ' s director and secretary in time. Two cases point to the same direction: the story of the new track is subject to cross-examination of the true value of capacity, team, funds and targets, and the real space managed in the name of financial investment is shrinking.
For pharmacists, it is important to allocate part of the capital to the high-scenes track, and the key is to distinguish between “allocation” and “transition”. The use of fund shares in exchange for cognitive windows and financial returns is a decision of a different nature than to place the fate of listed companies in areas where they lack the resources to accumulate. The question of whether the money could be moved back there, and whether it could be moved back, and whether it would increase the efficiency of R & D and the digitalization of production, would be worthless. AI and chips are worth the investment, but the ultimate foothold of the drug is still medicine.
Auto-fetched and translated by AI Gem Box. All rights belong to the original author.