香港維多利亞工商管理學院 香港維多利亞工商管理學院
English
Chinese Traditional Chinese English
Home >> News dynamics

News and Information

Contact Us

Email: HK_VCM@outlook.com

Address: Victoria Institute of Business Administration, Hong Kong

Information Details

AI giants earn a monthly salary of 30000 to rob "tiankeng professional" graduates! Huang Renxun's prediction came true

Author: This station Release date: 2026-08-26
Professional heat, is a feng shui turn of the process.

Over the past few years,CSAt the peak of the day, top-level students squeeze their heads into big science and technology factories.，However, civil engineering, materials, architecture and other majors are frequently considered to be graduates who can only squat on the construction site, enter the factory, or even lose their jobs in situ.“Sinkhole”。

Who would have thought that in just a few years, the wind has quietly changed.

As/WithAIa big outbreak,CSThe myth of employment is beginning to fade, and the tradition of ridicule is being seen.“Sinkhole”On the contrary, it has become a hot topic for top big factories.

This scene, it happened in the global science and technology circle upstartDeepSeekin the recruitment brochure.

InDeepSeekThe latest recruitment.“Facilities/Design Young Engineer”In the job requirements, it is clearly writtenElectrical, HVAC, Automation, Energy, Communication, Environment, Civil EngineeringSuch as professional can be extra points.

Seeing this recruitment, someone half-jokingly said:AIThe end is not the code, but the construction site, and now it's the turn.“Tiankeng Professional”Talent to eatAIThis bowl of rice.

But more people are curious: why did the bad-mouthing profession suddenly turn red?

图片

图片
Big factory out3Ten thousand monthly salary, roll call“Tiankeng Professional”Graduate

Recently, Zhejiang University's bully Liang Wenfeng foundedDeepSeekPublishIDCData center team recruitment information.

Team across Hangzhou, Beijing and Inner Mongolia Wulanchabu three places, recruitmentInfrastructure operations, planning and design and on-site implementation of three types of talent.

图片
Figure said:DeepSeekRecruitment Requirements

DeepSeekThe demand for this talent is not recent。

This year4In January, the company recruited a senior data center operation and maintenance engineer and delivery manager in Wulanchabu, with a monthly salary1.5Ten thousand3More than ten thousand yuan, or14Salary.

Arrived6In January, the company released the largest recruitment since its inception, setting upIDCDesign and planning engineer post, the tentacles extended to all aspects of data center construction.

DeepSeekis not an individual case,Big factories and big model unicorns are competing for such talents who have been looked down.

Byte Bounce InfrastructureDCEThe team is recruiting.IDCCost Specialist for data center location, design, delivery and cost management,Civil engineering, project management, drainage, HVAC, etc.Professional priority,Annual salary30Ten thousand to60Ten thousand yuan,14salary even.15Salary.

图片
Figure: Byte Beat Recruitment Requirements

An interesting contrast has emerged: real estate and traditional infrastructure are cold, and civil, environmental, construction and even materials majors have long been posted.“Dissuading Sinkhole”The label, HVAC and other traditional engineering have not escaped the fate of being rejected.

But a few years later, they wereAIThe rise of the industry, back to the front row of talent demand.

图片
Huang Renxun had already predicted that the over-gas professional collective turned red.

DeepSeek, Byte Beat and other technology companies call names.“Over-gas professional”The core reason is that they are building data centers on a large scale.

This year3Moon Huang Renxun proposed“Five-layer cake theory”gives the image of the answer。He putAIThe industry is compared to a five-tier cake, which, in turn, isEnergy, chips, infrastructure, models, applications.

图片
Source: NVIDIA official website

In the beginning, capital mania bet on the model layer,ChatGPT、DeepSeekSuch as large models take turns on the stage; now the volume to the application layer, letAIRoots in various industries.

But no matter how dazzling these two top floors are, whether they can run depends on whether the three layers of energy, chips and infrastructure can hold up. It was these three floors that gave“Over-gas professional”opportunities for re-pricing.

图片
Source: NVIDIA official website

First, the energy layer.

AIYes“electric tiger”，Large models consume a huge amount of power every time they are trained and inferred. According to the International Energy Agency, the power consumption of a single server rack is more than 10 times that of five years ago.

SoAIIn the end, the competition is not just algorithms, but also more sufficient, more stable, and cheaper electricity. This is why technology companies move their data centers to areas with sufficient energy and low costs.

DeepSeekWent to Wulanchabu, byte beat to Ningxia Zhongwei. Although the two places are five-tier cities, they are at the national level.“green electricity”Demonstration area, land and electricity costs are low.

Microsoft, Google,MetaS. technology giants have also moved their data centers to Arizona, Georgia, Virginia and other states.

Large model training needs24Hours of uninterrupted and stable electricity, wind and solar energy“rely on the sky to eat”So the big factories are also eyeing nuclear energy.

Subsequently, energy demand is transmitted to the talent layer.。Electrical engineers tube power distribution, energy engineers tube energy efficiency, environmental engineers tube energy consumption assessment and carbon emissions.

图片
Source: NVIDIA official website

The next layer is the chip.

AIMore and more computing power is needed. In addition to integrated circuits, electronic engineering, chemistry, and physics, the chip is also inseparable from materials science.

图片
The cost of materials, energy, equipment and high-end engineering talent in the process of chip manufacturing and packaging has risen, leading to a sharp rise in the ex-factory price of semiconductors in the United States.

Next is the infrastructure layer.

Huang believes that in the past, data centers were for storing information; nowAIThe original intention of the factory is to make intelligence, the raw materials are data and electricity, and the products areToken。

It sounds high-tech, but it solves specific problems.。How to choose land, where electricity comes from, how to build computer rooms, how to dissipate heat from servers, and how to stably operate hundreds of thousands of chips are almost all within the capabilities of traditional specialties such as civil engineering, architecture, HVAC, environment, and electricity.

AndAIThe data center of the era, giant transformers, large batteries, high-power generators, advanced liquid cooling and air.The tuning system is readily available and the mechanical, electrical and plumbing systems are updated approximately every four to six years.

According to McKinsey estimates, the cost of mechanical and electrical equipment for ordinary factories accounts for the total construction cost.15%To25%, largeAIData Center Directly40%To50%。

A buildingAIFrom ground breaking to stable operation of the data center, not only programmers but also a large team of engineers are standing behind it.

As the Nvidia website writes:“AIThe factory needs electricians, plumbers, pipe fittings, steel workers, network technicians, installers and operators, etc. These positions are highly technical and well-paid, and are still in short supply.”

图片
Figure: Huang Renxun said that three types of facilities are currently being built on a large scale around the world: chip manufacturing plants, supercomputer factories,AIFactory. This is becoming the largest infrastructure in human history, and the process is just beginning.

图片
Chinese and American Universities Act Collectively, Old Majors Race to Change Racetracks

The education side also smells high demand: Chinese and American universities are performing“Old Professional Change Track”, had been dissuaded from the professional not only did not disappear, but took the initiative andAINew energy, intelligent manufacturing and other new industries.

According to the Ministry of Education,“The 14th Five-Year Plan”During the period, the domestic colleges and universities to cancel the suspension1.22Ten thousand traditional foreign language, management and pure liberal arts majors, and new1.02Ten thousand counterpart trend majors, new this year alone38One.

图片
Figure: Ministry of Education2026New professions approved for inclusion in the catalogue in

This38In this major, there are three types of past“Tiankeng Professional”There was a big upgrade.

The first category is“Bucket running”the hardest hit areasCivil, transportation, landscape.

They turnWisdom landscape construction、Engineering InternetThe major of urban renewal has shifted from staying up late to drawing pictures and eating soil and moving bricks at construction sites to digital construction, smart cities and smart infrastructure.

Tongji University officially enrolled engineering Internet majors this year, integrating computer, civil engineering, architecture, transportation, surveying and mapping and other disciplines, allowing undergraduates to use Internet management thinking to solve major engineering problems.

American universities are also taking action.

University of Michigan this year6Monthly LaunchAIWith the Smart City major, located under the civil and environmental engineering system, research and use.AIautomation systems and sensor data to solve problems such as smart buildings and urban construction.

Northeastern University offers civil engineering andCSProfessional, training smart infrastructure talents;UCBThere is an intelligent infrastructure center that uses robotics to automate site maintenance.AIThe model predicts the problem of urban hydropower network.

图片
Source: University of Michigan official website

The second category is labeled“High pollution, burning boilers, pulling wires”of the labelEnergy and mining majors.

They step on the new energy andAIThe explosion of computing power has evolved new majors such as transportation energy integration engineering, energy science and engineering, and deep-ground science and engineering, specializing inAIThe power failure and heat dissipation problems of the times.

Xi'an Jiaotong University Enrollment Expansion This Year360People, focus on the layout of intelligent manufacturing, energy power, energy storage science, electrical engineering and other hard-core engineering, energy storage science and engineering is also included in the strong base plan.

Xi'an Jiaotong University is the first school in the country to open energy storage science and engineering. This is to train talents for data centers and new energy industries with trump cards.

图片

The third category was once regarded as a sunset industry.Mechanical and material majors.

BecauseAIWith the arrival of industries such as robots, semiconductors and new energy, they have been pushed to the forefront of high-end manufacturing, turning red into rare earth science and engineering, fiber science and intelligent manufacturing, semiconductor technology and equipment.

This year,“Cradle of Steel”The University of Science and Technology Beijing has newly introduced integrated circuit design and integrated systems, and intellectual property majors. The latter is oriented to new materials, green and low-carbon and other fields to escort high-end manufacturing.

The professional names of Chinese and American universities are different, but the direction is the same.。Old majors are moving to new industries, civil engineering. AIInfrastructure, Energy+Digitalization, Environment/Materials AISuch combinations also frequently appear in new professional introductions.

图片
The professional tuyere will change, and the ability to cross the cycle will not

The return of the past professional to the golden age can make childrenSelectionFirst, be clear:Professional is not eternal.“Sinkhole”and“Chosen by Heaven”What has really changed is the rise and fall of the industrial cycle.

DeepSeekInIDCThe Data Center Team Mission states:“Every generation has its own infrastructure revolution. It used to be railways, power grids and the Internet, but today it is supported.AIThe development of ultra-large-scale computing clusters.”

The fate of a profession is closely related to the industry it connects.。

When the industry expands, companies are hiring like crazy, wages are rising, and unpopular majors are turning red; when the industry shrinks, demand is decreasing, supply is still increasing, and popular majors may cool down rapidly.

Recently, the New York Fed counted the U. S.70Employment in multiple professions。CSUnemployment rate7%, Computer Engineering7.8%, at the same level as anthropology, performing arts, fine arts and other humanities and social sciences majors.“Employment King Fried”It also began to experience cyclical cooling.

图片
Source: Federal Reserve Bank of New York.

The same profession, different times, two fates. No major can stand on the air forever.

Therefore, choosing a major is the most taboo. What kind of fire is rushing in. What I really want to see is，When a child graduates four or six years later, the industry corresponding to the major is expanding, or has entered a period of talent saturation.

There are also families that will say，The truth is clear, but ordinary people simply cannot judge the future of the industry. If they can predict accurately, there will not be so many people who flock in when their major is at its peak, and find that the tuyere has passed when they graduate.

Therefore, for most families, instead of being obsessed with predicting the next outlet, it is better to let children have the ability to adapt to the future.


First, the ability of the bottom layer to migrate and lay the foundation.



The ability of the ground floor is like the foundation of the building.，If the foundation is strong enough, the superstructure will not be easily crooked.。The underlying capabilities are solid, more malleable, and migrate faster.

These underlying abilities may be logical thinking, mathematical ability, expression ability, or the ability to solve problems, obtain information, and collaborate with others.

They don't expire quickly as industries rise and fall。Children who study civil engineering have solid engineering thinking and practical ability, and can enterAIThe core base of the big factory; students who study the environment can also enter the new energy or green industry.

When the bottom-level ability is set up, when the industry changes, even if the major does not match, the child can find the intersection of the original ability and the new industry.

图片
Source unsplash


Second, keep the enthusiasm of continuous learning and dare to start again.



Gone are the good times when you can eat for decades with one skill.

Now the technology iteration is getting faster and faster, and the talents needed in the industry change every few years,“professional counterpart”The standards are increasingly blurred.

The real danger in the future is not to learn the wrong major, but to stop learning after graduation. Have the ability to continue learning, learn new tools and new knowledge from the original skills, and even take the initiative to enter a new field, you can make a comeback.


Third, never lose yourself and choose what you are really interested in.



This may sound like chicken soup, but the real interest from the heart will make a person willing to continue to invest in the industry when it is not so glamorous, and then find a breakthrough.

The market is always changing. What a person is interested in and willing to invest time is the starting point for long-term accumulation.

No one can accurately predict the future, but being able to turn around at any time, re-learn, know what you have, and where you want to go is the real confidence of a child.

Reference link
https://app.mokahr.com/social-recruitment/high-flyer/140576#/job/1c7e328d-3da9-4129-a938-21edfd77d38c
https://www.mckinsey.com/industries/private-capital/our-insights/scaling-bigger-faster-cheaper-data-centers-with-smarter-designs
https://blogs.nvidia.cn/blog/ai-5-layer-cake/
https://edition.cnn.com/2026/07/28/economy/ai-boosting-inflation-us
https://wallstreetcn.com/articles/3773162
https://www.newyorkfed.org/research/college-labor-market#--:explore:outcomes-by-major

Victoria Institute of Business Administration, Hong Kong Registration No. 35509134-003-04-23-2

©Copyright 2023 Victoria Institute of Business Administration, Hong Kong

↑