In the 2026 job market, there is no eternal “iron rice bowl”—only irreplaceable core competitiveness.
1. Experiencing the Outsourcing Predicament Firsthand: From Quitting an Alibaba Outsourcing Job Cold to Observing Today’s Industry Reality
Recently, I took part in an interview for an Alibaba AI technical-support outsourcing position. I attended mainly because the current hiring environment is so bad—there’s almost nothing in the way of official headcount being hired on BOSS Zhipin, and everyone who contacted me proactively was an outsourcing company. So I wanted to get a sense of the current market—after all, it’s been several years since I sent out a résumé to look at opportunities.
Let me be clear: I have zero intention of joining an outsourcing company.
The interview result opened my eyes: the position paid only 15,000 yuan a month, and they said “let’s talk” could get it to 16,000. Even so, that’s less than half the salary of the official-headcount position I held before, and the benefits system is extremely inadequate—social insurance and housing fund are all paid at the lowest coefficient.
This reminds me of when I first came to Hangzhou in early 2014. At the time, because I was switching careers and lacked work experience, I briefly worked at an Alibaba outsourcing company for 8 months. That experience made me deeply aware of the limitations of outsourcing positions: utterly no future. So I chose to quit cold and leave.
Over the years since, I’ve kept observing: Alibaba is still bringing in outsourced staff on a large scale, and top internet companies like ByteDance, Baidu, and Tencent have also one after another expanded their outsourcing recruitment.
And this phenomenon isn’t unique to China. As the book Feeding AI: A Documentary of the Global Underclass Workers of the Artificial Intelligence Industry reveals, this employment predicament exists in the global tech industry—the TVC workers (Temps, Vendors, Contractors) at Google’s data centers are a typical example.
Based on this, I’ve come to deeply realize: the core of career choice is not finding a job to make a living, but avoiding the “replaceability trap” and building your own irreplaceable core competitiveness.
A Cruel Reality: Core Competitiveness Can Also “Expire”
Ironically, ever since I quit that outsourcing job cold in 2014, I’ve been desperately improving my own core competitiveness—learning programming, learning operations, learning networking, attending training and study through all sorts of channels, working as a lecturer, going deep into various front-end and back-end internet roles, working at companies large, medium, and small. I’ve also produced many impressive results and projects.
I thought that as long as I worked hard enough and was well-rounded enough, I could gain a firm footing in the industry.
But at the end of 2025, I was still laid off.
Not because my abilities were insufficient, but because AI’s rapid development made the company feel that a lot of work could be done “with fewer people + AI tools.” After they laid me off, the TL figured that finding someone of a lower rank to take over my work, combined with AI, would do just as well. In fact, after I left, my former colleagues privately reached out to me several times saying something had broken down and hoping I could help fix it. I politely declined each time. After leaving, the former company’s affairs are no longer my concern; over-involvement could blur boundaries and isn’t necessarily good for my own career development either.
The colleague who took over my work—before he left, during the handover, I found out he didn’t even know how to use Docker, and didn’t know how to type Linux commands. But none of that mattered; what mattered was that the TL thought I was too expensive. My previous TL met the same fate—when the company went downhill, he was laid off, and his TL was laid off too. It really gives the impression of “a new emperor, a new court.” Before that, he too had led teams to many impressive projects, won many awards, and made many contributions to the company. The new TL who succeeded him was clearly inferior in technology and team management, but in “managing upward” and “cost awareness” he obviously better fit the requirements of the high-P people. This precisely confirms a reality: from the company’s point of view, being cheap, easy to use, and obedient is often more important than technical depth. As for employees’ career development—that’s never a priority.
Economist Milton Friedman believed that digital technology makes the global labor market fairer—employers and employees can connect freely across borders for a win-win. But Friedman’s view has a fatal flaw: equal connectivity does not equal equal bargaining power. Many companies pursue flat management; it looks like there are no ranks and no hierarchy, but more often than not it’s still a one-man show, and the frontline employees below have no real right to raise opinions or suggestions, and no bargaining power. For example, if your view differs from your TL’s, they might think you’re hard to manage, not obedient. Once that impression forms, at a critical moment—like during layoffs—it becomes a reason to be “optimized.” At that point, the company won’t mention “flat management” or “equal dialogue” anymore; it’ll use another set of standards to measure your “value.” This say-one-thing-do-another phenomenon is not rare: externally they proclaim people-first, while in actual decisions it’s cost-first; they preach innovation and inclusiveness on their lips, but what they truly can’t tolerate is a different voice.
The deeper problem is: so-called “core competitiveness” is actually a relative concept.
From the employee’s point of view, everyone has a certain autonomy over their own work tasks—you can decide what tools to use and what method to use to complete the need at hand. But when it comes to important company-level decisions like resource allocation and strategic direction, the say lies entirely with investors and management. Whether you’re a data annotator or a machine-learning engineer, all employees fail to gain the formal power to participate in company decisions, and there’s an obvious power asymmetry between them and the employer.
Even the most capable, best-paid engineer, once disagreeing with the boss, has almost no bargaining chips if relying on their own strength alone. Employees occupy a structurally weak position in the enterprise, because from the company’s level, everyone in every position is replaceable—an individual’s skills, training, and labor are not hard to find substitutes for in the labor market.
This forces me to reflect: in the AI era, what is truly “irreplaceable”?
A cruel fact is that the purpose for which many companies introduce AI and automation was never to “empower employees,” but to lower labor costs—from “replaceable” to “fully replaced.”
Amazon’s supply-chain optimization system (SCOT) is a typical case. This logistics-management system, called “the most advanced of the 21st century,” controls the whole process from demand forecasting to delivery planning. Through billions of data points (page views, order numbers, warehouse barcode scans, etc.) it optimizes delivery routes in real time and generates work metrics. Internal documents from 2019 show that the system automatically tracks less efficient employees, and if they consistently fail to meet targets, the system automatically fires them—with no supervisor involvement.
What’s more terrifying is that this AI system also concentrates decision-making power entirely in the hands of executives, while at the same time degrading employees’ skills through a job-rotation system. In 2020 Amazon announced it would use a machine-learning algorithm to set work schedules based on employees’ muscle usage, rotating workers among jobs that use different muscle groups, euphemistically called “reducing fatigue,” but actually to maximize output. Here AI is not an assistant, but a profit-squeezing machine—it depends on labor power, yet through fine-grained control it extracts human value to the extreme.
This reveals a truth that has long been whitewashed: the progress of productive forces has not liberated workers from work; on the contrary, the main beneficiaries of innovation are often investors and company owners. Chaplin’s line in The Great Dictator is still deafening today: “Machinery has brought material abundance, yet it has left us in poverty.”
When technology is applied to the workplace, it first becomes a means for bosses to squeeze profits and control workers—from Amazon’s automatic-firing system, to algorithm-driven performance monitoring, to rotation scheduling based on muscle usage. Workers gradually realize that the first beneficiaries of technological development are capitalists, not workers. Dissatisfaction with management often shows up directly as resistance to technological change, because the two are tightly linked and mutually causal.
We can conceive of more ways to liberate productive forces, but the premise is a completely different social formation. Under the current business logic, optimizing AI systems is about raising efficiency and lowering production costs, not about employee well-being. The company’s profit and the customer’s needs come first—as for employees’ career development, work dignity, or even basic livelihood, those are never target variables in system optimization.
More importantly, this technological control is spreading from blue-collar to white-collar positions. For many industries whose productivity is already low, the main value of AI systems lies precisely in their ability to extract more value from employees.
The book Feeding AI reveals a broader global picture: in Nairobi, Kenya, data annotators work 10 hours a day, and every 55 seconds must process one piece of violating content. They sit in offices decorated in “Silicon Valley style,” with a slogan on the wall reading “accelerating human development,” yet in reality they are closely monitored like laborers in the plantation era—from biometric clock-ins to omnipresent cameras, recorded every second. One annotator must process 500–1,000 pieces of violating content a day, and many admit that this job has left an indelible mark on their lives.
This mode of work is nothing new. As the book points out, its history can be traced back to the colonial-era plantations: overseers recorded each slave’s daily picking amount, and one pound short earned a whip lash. Today’s data-annotation centers are merely the digital upgrade of this management science in the AI era.
Digital monitoring technology was first applied in call centers—an area that has long recorded all employees’ calls, euphemistically called “training and monitoring.” Today’s AI management technology is merely the iterative upgrade of this traditional approach. It isn’t some brand-new technological revolution, but an extension of the existing management-control model to new groups of laborers.
The widespread application of AI across industries, especially in areas traditionally seen as “white-collar safe harbors,” is having far-reaching effects. Since 2020, workplace monitoring has surged, with many companies starting to collect all information about employees’ work—keystroke frequency, email response times, even mouse-movement trajectories—sometimes without the employees’ knowledge. Companies use machine algorithms to analyze this data, predict employees’ future behavioral patterns, assess “attrition risk,” and adjust management strategy accordingly.
What does this mean? It means that even if you’re sitting in an office, your every move is being watched by algorithms. Your “work efficiency” isn’t defined by you—it’s determined by a data model.
In the past we believed that technical depth, project experience, and being a generalist were core competitiveness. But when AI can write code, troubleshoot faults, and generate test cases, many of the skills we once took pride in are depreciating rapidly.
Even more alarming: AI’s threat isn’t limited to “blue-collar-ized” technical positions—even creative workers can’t escape. Feeding AI records the discovery of an Irish voice actor: her voice was synthesized with an algorithm by the company without permission, creating a “soulless AI voice actor,” quoted at only one-tenth of a human’s price. “It’s as if someone cloned me and made a Frankenstein’s monster.”
Screenwriters, illustrators, translators, musicians… these professions once thought to require “human creativity” are facing the same threat. AI companies are selling what doesn’t belong to them—the value of image-generation tools is built on human original works, and the ability of language models comes from the collective intelligence of human creators. In the book’s words, they’ve twisted the tech industry’s motto “move fast and break things” into “move fast and steal things.”
This isn’t to say that improving core competitiveness is useless—it’s to say that the definition of core competitiveness is changing.
Ten years after quitting outsourcing cold, I still couldn’t escape the fate of being laid off. This perhaps shows that relying solely on individual effort to improve skills is not enough to withstand systemic risk. We need a clearer understanding of industry trends, and to lay out earlier in the areas AI can’t easily replace in the short term.
2. The Spread of Outsourcing: Full Penetration from Basic Posts to Technical Posts
As a practitioner who has long worked deep in the internet industry, my previous company’s employment adjustments made me feel the spread of outsourcing even more directly:
The company once laid off basic posts like customer service on a large scale, with a layoff ratio approaching half, and then brought in outsourced temp workers to take over the related work.
Not only that: in the department I was in, frontline technical-support posts also gradually began bringing in outsourced staff, and even some front-end and back-end development posts saw outsourced staff appear.
Through observation it’s easy to see that the basic work of such basic posts and some development posts mostly lacks a core technical barrier, being largely standardized, process-based repetitive operations that can be done without deep professional accumulation.
This post characteristic also determines that they are extremely easy to outsource, becoming the main entry point for enterprises to cut costs and raise efficiency.
3. The Essence of Outsourcing: The “Second-Class Citizen” Predicament Under Capital’s Cost-Cutting and Efficiency-Raising
In fact, when top internet companies bring in outsourced staff on a large scale, the essence is the inevitable choice of cost-cutting and efficiency-raising under an equal-work-unequal-pay model.
Whether Alibaba, ByteDance, or Baidu, Tencent, their outsourcing posts are mainly concentrated in supporting, process-based posts. Such posts:
- Do not directly participate in the company’s core business
- Do not create core revenue
- Only carry auxiliary support functions
Therefore they become the priority choice for companies to compress labor costs.
And practitioners in such outsourcing posts often face:
- ❌ Low pay
- ❌ Inadequate benefits
- ❌ No path to promotion
- ❌ Poor employment stability
Reduced to “second-class citizens” in the industry’s development.
4. A 2026 Career-Choice Guide: Avoid Traps, Build Irreplaceable Competitiveness
Combining my own experience and industry observation, I believe that in 2026, against the backdrop of increasingly fierce industry competition and the continuously spreading trend of outsourcing, career choices need to hold to a core logic:
Avoid posts with strong replaceability, and focus on building your own core competitiveness.
Specifically, one can start from the following aspects:
(1) Avoid Highly Standardized Process-Based Posts
The work content of such posts can be broken down into fixed operational norms, with results easily quantifiable and no personalized value.
Typical posts:
- Basic customer service
- Frontline technical support
- Junior data entry
- Simple code writing (CRUD grunt work)
Risk warning:
Such posts can be mass-replicated; to compress costs, companies prioritize the outsourcing model. Practitioners not only get low pay and benefits, but also face the risk of being replaced by AI technology—as generative AI becomes widespread, basic consulting, fault troubleshooting, and simple code generation can be done automatically by AI, and the survival space of such posts will shrink further.
(2) Stay Away from Supporting Posts Weakly Coupled to the Core Business
Supporting posts are an auxiliary link in a company’s operation and don’t directly create core value.
Typical posts:
- Non-core administration
- Basic logistics
- General financial auditing
- Basic operations and maintenance
Real case:
I once built a financial-audit RPA for my company’s finance team. Once this RPA went live, it replaced the equivalent of one finance headcount. Afterward, finance posts were almost entirely replaced by interns and outsourced staff.
Bringing in RPA on a large scale to achieve various process automations—such posts have no direct impact on the company’s core business, and are a disaster zone for outsourcing. Practitioners not only find it hard to get promotion opportunities, but may also be the first cut during industry adjustments, lacking employment stability.
(3) Focus on Core Competitiveness, Choose Posts with Strong Irreplaceability
The core of career development is building your own barrier—long-term competitiveness matters more than short-term salary.
In 2026, core posts in core tracks like AI, cloud-native, and new energy have the following characteristics:
- ✅ High barrier
- ✅ High complexity
- ✅ Non-standardized
Recommended directions:
| Field | Core posts |
|---|---|
| AI | AI Agent development, RAG knowledge-base construction, model fine-tuning engineer, algorithm engineer |
| Cloud-native | Advanced K8s operations, SRE, platform engineer |
| Data | Data architect, real-time computing engineer |
| Security | Application security engineer, security architect |
But one must clearly recognize: these posts’ current “irreplaceability” is only relative and temporary. As AI technology keeps advancing, model fine-tuning may be simplified by automation tools, and the algorithm engineer’s work may also be assisted or even partially replaced by smarter AI. Today’s high barrier may be flattened by a technological breakthrough tomorrow.
Core competitiveness also has an expiration date—there’s no once-and-for-all safe zone. Only by maintaining the ability to keep learning and continuously updating your skill set can you maintain a relative advantage amid technological iteration.
(4) Rational Advice on the “Outsourcing Springboard”
For job-seekers preparing to switch careers or with insufficient experience, can an outsourcing post serve as a “springboard”?
My advice is: yes, but with strict preconditions.
First, we need to clarify a cruel reality: as mentioned above, the Alibaba outsourcing post pays only 15,000 yuan a month, with social insurance and housing fund paid at the lowest coefficient and a highly inadequate benefits system—but this doesn’t mean outsourcing is the “worst choice.” As far as I know, many small companies in Hangzhou with dozens or hundreds of people currently offer pay and benefits even worse than outsourcing, and in serious cases there’s even wage arrears, some dragged out for years.
In this employment environment, outsourcing has instead become a “relatively not-bad” option—at least big-company outsourcing generally pays on time, social insurance and housing fund may be paid at the minimum but are at least there, and the work environment is relatively formal.
If you’re in the following situations, outsourcing can serve as a transitional choice:
- Switching careers from zero and urgently needing a related work record
- A long gap period, needing to re-enter the workplace
- Temporarily having no better choice, or the available small and medium companies can’t even pay wages
But you must clearly recognize the following points:
1. The time red line: don’t exceed 1 year
The longer an outsourcing stint stays on your résumé, the greater the negative impact. An outsourcing stint of more than 1 year will make HR and interviewers doubt you:
- “Why didn’t you become a regular employee after so long?”
- “Is it because your ability is insufficient and you can only do outsourcing?”
- “Once used to the pace of outsourcing work, can you adapt to a formal position?”
2. The career brand: an outsourcing stint gets “labeled”
Many companies openly need outsourced staff to work for them, but deep down discriminate against outsourced staff:
- Outsourced staff are often excluded from core meetings
- They can’t access core business and key data
- Even if they perform excellently, it’s hard to gain equal recognition
3. The conversion illusion: the probability is less than 0.1%
Many outsourced workers get “pie in the sky”: “Work hard, and you’ll have a chance to become a regular employee.” But the reality is:
- Big-company conversion quotas are extremely scarce
- Even if there’s a vacancy, campus recruitment or social recruitment gets priority
- Outsourcing conversion requires layer upon layer of approvals—nearly impossible
4. Negative impact on future job-seeking
An outsourcing stint leaves a mark on your professional record:
- When the next company’s HR sees the outsourcing stint, they’ll doubt your professional ability
- Under equal conditions, employers lean toward candidates without outsourcing experience
- During salary negotiation, outsourcing experience gets used to push your price down
Summary: outsourcing can serve as an “emergency plan,” but must never be a “long-term strategy.” If you have no choice but to take outsourcing, be sure to do the following well:
- Set a clear time limit (at most 6–12 months)
- Use this time to study like crazy and accumulate project experience
- Actively apply for official-headcount positions, and don’t get comfortable with the status quo
- When leaving, package your record well and downplay the outsourcing nature
5. Conclusion: Is There Really No Way Out for Wage Labor?
Writing to this point, I must state the cruelest conclusion—wage labor will forever have no way out.
This isn’t an emotional vent, but the inevitable result of the earlier arguments:
- You have no say at all in company decisions (structural weakness)
- All your behavior is monitored and evaluated by algorithms (technological control)
- Your skills may be replaced at any time by AI or cheaper people (replaceability)
- Your “core competitiveness” will expire with technological progress (limited shelf life)
No matter how hard you work, how outstanding you are, as long as you’re still inside the “wage labor” framework, you’re forever a cost item, forever an object that can be optimized. You might avoid outsourcing today, but face “AI-ization” tomorrow; today you might be official headcount at a big company, tomorrow the entire business line might be cut. The company isn’t your home, the boss won’t take responsibility for your career, and HR’s KPI is pushing costs down rather than protecting employees.
So where is the way out?
The only way out is to jump out of the structural trap of “wage labor.” This doesn’t mean quitting immediately to start a business, but rather building something of your own while still working:
- A personal brand and influence: let yourself be seen and remembered, not merely a name on a résumé
- An independent source of income: a side business, consulting, knowledge monetization—even a few thousand yuan more a month is the confidence to leave the platform
- Transferable abilities and resources: a client network, industry connections, exclusive experience—things you can’t carry away when you leave, but the connections remain
- A mindset to confront uncertainty: accept that there’s no such thing as a “stable job,” and build your sense of security on yourself rather than the company
This isn’t encouraging reckless entrepreneurship—data show nine in ten startups fail. It’s saying that in this algorithm-dominated, cost-first era, relying purely on one salary to survive carries far higher risk than diversifying your investment.
I feel this deeply. In earlier years, I used my spare time to lecture for platforms, earning around ten thousand a month from side work; I also ran a WeChat public account, earning some income by writing articles. But later, as work got busier and busier, my energy was taken up by meetings, projects, and overtime, and these side businesses gradually got shelved. At the time, I felt that since my full-time salary had risen, focusing on the “main job” was more worthwhile.
Thinking about it now, that trade-off really wasn’t worth it. The salary did rise somewhat, but the price was losing the possibility of exploring side businesses—the content that could have settled into personal assets, the fans accumulated, the industry connections built, all dissipated as I stopped updating. A job can give you a monthly salary of 30,000, but as long as you’re in that position, your time will forever belong to the company; while a side business’s value may be only a few thousand yuan at first, but it belongs to you, and it has the compounding of time.
Having worked in the industry for many years, I’ve come to deeply feel: the wave of outsourcing in the industry doesn’t eliminate posts—it eliminates replaceable practitioners.
In the 2026 job market, there is no eternal “iron rice bowl”—only irreplaceable core competitiveness.
For practitioners, in the short term you can:
- Avoid posts in outsourcing disaster zones
- Go deep in core tracks
- Continuously improve your professional ability
- Build an irreplaceable competitive barrier
But in the long run, the real “irreplaceability” is that you no longer depend on any company to decide your value. Only by owning your own business, your own clients, your own brand can you truly gain a firm footing amid industry change, and completely break free from the “outsourcing fate”—even break free from the fate of “being employed” itself.
Postscript: Cold Reflections on “Collective Action”
At its end, the book Feeding AI offers several suggestions for improving the situation of laborers: building transnational unions, promoting shared decision-making systems, exploring the worker-cooperative model, strengthening government regulation, and so on. These suggestions sound grand, but as a practitioner who’s struggled in the workplace for years, I think they’re nearly impossible to implement.
The reality is: when unemployment is high and workers can’t find formal jobs, no one will give up a job in hand for the sake of “improving working conditions”—because you know there are countless people outside the door queuing for your position. Transnational unions? In an age when AI can move work to any corner of the globe at any time, the interests of laborers are often in conflict, not aligned. Shared decision-making? Let labor representatives onto the board? Look at those companies claiming “flat management”—the real decision-making power was never in the hands of those nominal “representatives.”
History tells us: only when the pressure of survival exceeds the limit of endurance will the system be forced to change. Before unemployment exceeds half the population and most people “can’t survive,” “rising up” will only be the tragedy of a few brave individuals. Until then, as individuals, the most pragmatic thing we can do is to lay out our own Plan B as early as possible, while we still can.
This isn’t pessimism—it’s clarity.
Related reading:
评论
0 条评论