文 | 产联社
机器人赛道，现在最贵的不是算力，是人才。
今年以来，具身智能行业持续升温，企业一边忙着融资、扩产，一边开始在人才市场上“抢人”。有公司为顶尖科学家开出上亿元年薪，也有企业把招聘范围延伸到应届生和在校博士，生怕错过下一个技术骨干。
今年春招期间，AI领域岗位量迎来爆发式增长，其中具身智能相关岗位的招聘需求更是暴增15倍，平均月薪达到6.2万元。
此前4月，优必选公开招募具身智能首席科学家，开出1500万元起步、最高1.24亿元的年薪总包。按照工作日折算，最高薪酬接近一天50万元。
这场重金求贤的背后，是整个行业在资本推动下的快速扩张。数据显示，国内具身智能赛道今年上半年融资总额已突破900亿元，同比增长近5倍。
然而，当资本蜂拥而至、百亿独角兽急于跑马圈地时，制约企业发展的最大瓶颈已经不再是资金或者算法，而是适配人才的匮乏。
机器人赛道，天价抢人才
具身智能行业有多缺人？招聘市场的数据已经给出了答案。
据脉脉高聘人才智库发布的《脉脉2026春招职场洞察报告》，2026年1月至4月，国内AI领域岗位量同比增长8.7倍。其中具身智能相关岗位更是增长15倍，平均月薪达到6.2万元。
从整个机器人行业来看，招聘需求同样在快速增长。
猎聘大数据研究院发布的《2026机器人领域人才供需趋势洞察报告》显示，2026年1月至7月，机器人领域新发职位同比增长60.17%，其中人形机器人细分赛道增长147.02%。
大量的人才缺口，直接推高了顶尖人才的薪酬上限。
2026年4月2日，港股“人形机器人第一股”优必选，公开面向全球招募具身智能首席科学家，开出的年薪总包最低1500万元，最高可达1.24亿元，薪酬水平直接对标硅谷顶尖科技巨头。
优必选在2025年财报中透露，其全尺寸具身智能人形机器人实现收入8.21亿元、销量1079台，在全年净亏损7.9亿元背景下，仍亮出49.2亿元货币资金储备，以此表明砸重金招募技术领军者的决心。
紧接着在2026年4月中旬，追觅旗下的魔法原子内部流出消息，放言要以2亿元年薪招募首席科学家，再次刷新行业对人才薪酬的想象。
在海外市场，竞争同样白热化：2025年5月，OpenAI以65亿美元的价格收购了一支仅有55人的AI硬件团队，相当于人均身价超1亿美元；波士顿动力从特斯拉挖走机器人项目负责人时，开出的年薪也超过200万美元。
顶尖人才的薪酬水涨船高，也进一步带动了校园招聘和实习市场。
在2026年秋招季，具身智能方向的优秀应届硕士、博士毕业生，年薪50万元的 offer 已被戏称为“白菜价”。
为了提前锁住名校苗子，部分初创公司甚至推出了“实习即享全职待遇”的抢人策略。只要候选人承诺毕业后入职，在博三、博四实习期间，就可以按照全职offer的标准发放薪资，月薪最高可达10万元。
高薪之外，企业之间的人才争夺也越来越激烈，甚至出现了围绕核心人才抬高公司估值的说法。
据钛媒体报道，有投资人透露，在具身智能行业内部，流传着一种“人头估值”的公式：只要招揽到清华、浙大等名校的顶尖教授担任顾问或团队成员，公司估值便可顺势抬高2亿至5亿元；挖来头部公司的核心高管，估值直接增加5000万元。
各家企业疯狂抢人，本质是为后续产业比拼储备力量。但这种不计成本高薪挖人的模式，也让行业潜藏的问题慢慢暴露出来。
算法之外，工程人才也吃香了
具身智能为什么这么缺人？核心原因在于，它是一个“跨专业”的赛道。
传统软件只需要在数字世界里运行，但机器人要进入现实环境，既要理解指令、识别物体，还要通过机械结构和电机完成动作。它不仅要“想得明白”，还得“做得出来”。
因此，具身智能并不是单一的软件算法或硬件制造，而是一个涉及计算机科学、自动控制、机械工程、人工智能和动力学等多个领域的交叉行业。
以往高校培养的人才大多局限于单一领域，懂算法的人未必了解真实机器人的物理限制，懂机械结构的人又未必熟悉大模型技术。
目前，高校已经开始调整培养方向。2026年，教育部在本科专业目录中首次增设“具身智能”专业，并支持哈尔滨工业大学、北京航空航天大学等9所高校增设相关专业。
不过，新专业刚刚起步，人才培养需要时间。首批本科生最快也要到2030年才能进入市场，远水解不了近渴。短期内，企业还是得从计算机、自动化、机械工程等相邻专业，以及自动驾驶、大模型等相关行业寻找合适的人选。
与此同时，企业对人才的需求也在发生变化。
前两年大家争抢的主要是算法工程师，重点打磨机器人的“大脑”；但随着人形机器人从实验室演示走向工厂落地，企业的关注点从“算法能不能跑通”，转向“产品能不能稳定量产、批量交付”。
优必选就是一个例子。2025年，优必选的Walker系列进入汽车制造、智能制造等场景实训，全年工业人形机器人交付量超过500台。2026年，公司定下目标，要把工业人形机器人年产能提升到一万台。
产能扩大之后，企业面对的就不只是算法问题了：机器人能不能长时间稳定运行？零部件能不能按时供应？生产工艺能不能标准化？设备出了问题，能不能及时调试和维护？这些都离不开工程、制造和交付方向的人才。
智元机器人也在加快推进规模化生产。2026年3月，智元宣布第1万台通用具身机器人正式下线。从5000台到1万台，公司只用了约三个月。机器人数量快速增加，背后离不开供应链管理、生产制造和质量控制等环节的配合。
招聘数据也印证了这一变化。猎聘数据显示，2026年1月至7月，具身智能企业新发职位中，算法工程师岗位同比增长43.73%；工业工程师岗位增长102.33%，电气工程师增长98.94%，解决方案工程师增长85.98%，自动化工程师增长85.51%。
不难看出，企业不只需要能让机器人跑起来的算法人才，更需要能支撑量产、落地工厂、解决现场问题的工程人才。
资本持续加码，进一步放大了这场抢人热潮。
IT桔子数据显示，2026年上半年，国内具身智能赛道融资总金额突破935亿元，同比增长近5倍，融资事件超过300笔。截至2026年年中，国内估值超过百亿元的具身智能独角兽企业已剧增至22家。
大量资金涌入，给企业扩团队、搞研发、建产线提供了底气。但拿到钱，不等于商业化就能顺利跑通。
机器人研发、硬件制造、测试、场景部署都要持续烧钱，如果企业一味砸高薪抢人，迟迟拿不出稳定产品和真实订单，高昂的人力成本最终也会成为经营负担。
说到底，具身智能行业现在最缺的，不只是能把模型做得更聪明的人，更是能把模型装进机器人、让机器人在真实环境中稳定干活的人。
上亿元年薪可以吸引顶尖科学家，但机器人能不能卖出去、能不能在工厂里持续稳定运行，靠的不是某一个人的技术突破，而是整个团队把技术一步步变为成熟产品。
参考资料：
《核心岗位年薪轻松破百万，最高1.24亿：具身智能“抢人大战”到底有多狠？》，财闻；
《具身智能本科招生火爆：当AI“长出身体”，大学开始培养第一批“造身体的人”》，羊城晚报；
《“人形机器人第一股”，最高1.24亿元年薪招首席！》，证券时报网；
《巨亏之下，优必选1.24亿年薪挖“完人”》，融中财经；
《具身智能抢人大战：毕业一年，年薪300万元》，36氪；
《应届生涌向机器人秋招：年薪50万元成“白菜价”，正式offer不好拿》，第一财经；
《企业蹲点高校宿舍楼，具身人才短缺引发抢人大战》，澎湃新闻；
《实习生也能月薪10万了》，投中网；
《具身智能融资半年破900亿 投资逻辑从“后空翻”转向“搬砖头”》，中国证券报；
《65 亿美金抢 55 人、日薪 50 万！全球机器人抢人，彻底失控》，钛媒体；
《具身智能上半年：百亿独角兽增加近20家，港股IPO排队火热》，第一财经。
完整产业数据报告、市场趋势分析，移步「产联社」客户端
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
United Nations Population Fund
The robot track, it's not the math, it's the talent.
Since the beginning of the year, the intelligent industry has been on the rise, and enterprises have begun to “snatch people” in the talent market while busy with financing and expanding their production. There are companies that offer hundreds of millions of dollars a year for top scientists, and companies that extend recruitment to current students and school doctors who fear missing the next technocrat.
During this spring's recruitment, the number of jobs in the field of AI rose sharply, with recruitment demand for smart-related posts increasing 15 times, with an average monthly salary of $62,000.
Earlier, in April, Eugève had publicly recruited leading scientists with intelligence skills, with an annual lump sum of up to $124 million starting with $15 million. On the basis of the working day, the maximum remuneration is close to $500,000 a day.
Behind this heavy financial commitment is the rapid expansion of the entire industry, driven by capital. The data show that domestic smart track financing has exceeded $90 billion in the first half of the year, almost five times the same level.
However, when capital flows and billions of unicorns rush to run in circles, the biggest bottleneck in enterprise development is no longer finance or algorithms, but the shortage of suitably qualified people.
The robot track. It's priceless.
How many people are missing in the smart business? The data on the recruitment market have given the answer.
In January-April 2026, the number of jobs in the field of AI in the country increased by 8.7 times over the same period. Of these, the number of smart-related jobs has increased 15 times, with an average monthly salary of $62,000.
Demand for recruitment is also growing rapidly across the robotic industry.
The Insight Report on Human Supply and Demand Trends in Robotics, 2026, by the Grand Data Hunting Institute, publish, shows that between January and July 2026, new jobs in robotics increased by 60.17 per cent over the same period, with a 147.02 per cent increase in the segment of human robots.
There is a large talent gap that directly pushes up the pay ceiling for top talent.
On 2 April 2026, the Hong Kong Unit's “first unit of human robots” was chosen with the open recruitment of leading scientists with intelligence skills worldwide, with a minimum annual gross salary of $15 million and a maximum of $124 million and a direct pay level to the top technocrats of Silicon Valley.
In the 2025 financial paper, Euphoria revealed that its full-sized intelligent human robots earned $821 million and sold 1,079 units. Against a net loss of $790 million throughout the year, the reserves of $4.92 billion in monetary money were rekindled, thus demonstrating the determination to pay heavily to recruit technology leaders.
Immediately after that, in mid-April 2026, news came out of the search for the magic atoms under the flag, saying that the chief scientists would be recruited at 200 million yuan a year, and that the industry's image of talent would be refreshed.
In overseas markets, competition is equally hot: in May 2025, Open AI purchased an AI hardware team of only 55 people at $6.5 billion, equivalent to over $100 million per person; and Boston was able to extract robots from Tesla with more than $2 million per year.
The high remuneration of top talent has also further boosted the market for school recruitment and internships.
In the fall season of 2026, an excellent master's degree and doctorate with an intelligent orientation, with an annual salary of $500,000, was dubbed the “beverage price”.
Some start-ups even introduced a strategy of “internship or full-time treatment” in order to lock down famous school children in advance. As long as the candidate is committed to taking up his or her post after graduation, a full-time offer ' s salary of up to $100,000 per month may be paid during the course of the Boulder and Bourbon internships.
In addition to high wages, competition for talent among enterprises has increased, and even claims that corporate valuations around core talent have emerged.
According to titanium media reports, investors have revealed that a “head valuation” formula is being circulated within the smart industry: once top professors from the top schools of Zhejianghua and Zhe Da have been recruited to serve as consultants or team members, the company's valuation will rise by between $200 million and $500 million; and the core executive of the head company will be dug and the valuation will increase directly by $50 million.
Businesses are crazy to rob people, essentially to build reserves for the successor industry. However, this cost-neutral pattern of digging people has slowly exposed the potential of the industry.
In addition to algorithms, engineers are enjoying themselves.
Why are you so smart? The central reason is that it is a “cross-professional” track.
Traditional software only needs to operate in the digital world, but robots need to get into the real environment, understanding commands, identifying objects, and completing their actions through mechanical structures and electric power. It must not only “think well” but also “do it”.
Therefore, smartness is not a single software algorithm or hardware manufacturing, but rather a cross-cutting industry involving a variety of fields, including computer science, automated control, mechanical engineering, artificial intelligence and kinetics.
In the past, the skills developed by higher education institutions were mostly limited to a single field, and those who understood algorithms were not necessarily aware of the physical limitations of real robotics and those who understood mechanical structures were not necessarily familiar with LLM technology.
At present, higher education institutions have begun to reorient their education. In 2026, for the first time, the Ministry of Education introduced the subject of “smartness” into the subject's catalogue and supported nine higher education institutions, including Harbin Industrial University and Beijing Aerospace University.
However, new professions are just beginning and talent development takes time. The first undergraduates will not be able to enter the market until 2030 and will not be free of thirst. In the short term, enterprises still have to find suitable candidates from neighbouring professions such as computers, automation, mechanical engineering and related industries such as autopilot and LLM.
At the same time, business demand for talent is changing.
The first two years were dominated by algorithm engineers, who focused on grinding the robot's “brain”; but as human robots moved from laboratory demonstrations to plant landings, the focus of business shifted from “calculations running” to “products stabilizing production and bulk delivery”.
Precedence is one example. In 2025, the preferred Walker series was trained in auto manufacturing, intelligent manufacturing, etc., with over 500 industrial human robots delivered throughout the year. In 2026, the company set a target of increasing the annual production capacity of industrial human robots to 10,000.
With the expansion of production capacity, businesses are faced with more than a problem with algorithms: can robots operate steadily over time? Can spare parts be supplied on time? Can production processes be standardized? Is there a problem with the equipment, can it be debugged and maintained in time? These are talent that cannot be separated from engineering, manufacturing and delivery.
Brain robots are also accelerating scale production. In March 2026, Ji-won officially announced the release of 10,000 universal body robots. From 5,000 to 10,000, the company spent only about three months. The rapid increase in the number of robots is linked to supply chain management, manufacturing and quality control.
Recruitment data also confirm this change. The hiring data show that between January and July 2026, the number of new jobs in smart enterprises increased by 43.73 per cent in comparison with the same period; the number of posts in industrial engineers increased by 102.33 per cent, the number of electrical engineers increased by 98.94 per cent, the number of solutions engineers increased by 85.98 per cent and the number of automated engineers increased by 85.51 per cent.
It is easy to see that firms need not only algorithms that allow robots to run, but also engineers that support mass production, landing plants and solving problems on the ground.
The increase in capital has further amplified the wave of human trafficking.
According to IT orange data, in the first half of 2026, the total amount of smart track financing in the country exceeded $93.5 billion, almost five times the same rate, with over 300 financing incidents. By mid-2026, the number of intelligent unicorn enterprises valued domestically at over $10 billion had increased dramatically to 22.
Large inflows of funds have provided the basis for expanding the enterprise team, undertaking research and development, and constructing production lines. But getting the money doesn't mean commercialization can run smoothly.
Robust development, hardware manufacturing, testing, scene deployments are subject to constant burning of money, and high human costs can eventually become a business burden if enterprises fail to provide stable products and real orders.
In the end, what is most lacking in the smart industry is not just people who can make models smarter, but people who can put models into robots and make them work in real conditions.
Hundreds of millions of dollars per year can attract top scientists, but robots can sell, they can operate steadily in factories, not by technological breakthroughs of an individual, but by the team as a whole turning technology into a mature product.
References:
" The annual salary of the core job is one million easy, up to 124 million people: how bad is the smart “House of Man”? ", financial news;
"Accessed with intelligence undergraduates: When AI "breaths the body", the university began to train the first "body-builders" in the Sheep City Night Post;
“First share of human-shaped robots”, top of the 124 million-year wage call!
"By great loss, you will choose 124 million years' salary to dig "Fair" and settle your fortunes;
The Smart Man's Fight: One Year of Graduation, 3 million yuan a year, 36 krypton;
"Autumn moves to robots: half a million dollars a year for a "bucket price", official offer bad. First financial account;
(b) The " Company Crouched High School Hostel Building ", a shortage of skilled personnel that led to a war on human rights, and news;
" Interns can also earn 100,000 a month " and go to the network;
"By six months, smart finance broke 90 billion dollars ", investment logic shifted from "back-to-back flip" to "drawing bricks", China Securities Journal;
"The $6.5 billion grab, 55 people, half a million daily! The world's robots are out of control," titanium media;
" The first half of the year with intelligence: 10 billion unicorns increased by almost 20, Hong Kong ' s IPO lined up hot ", first fortune.
Full industry data reporting, analysis of market trends, moving to "Father Associations" clients End
Auto-fetched and translated by AI Atlas. All rights belong to the original author.