Written by: Lin Wanwan
GitHub has a website called Star History, which tracks the popularity of open-source projects. The horizontal axis represents time, and the vertical axis represents the number of stars. It's said that programmers study this chart as diligently as they would a textbook.
There are three lines on the graph. The red line is React. Facebook open-sourced it in 2013, investing thousands of engineers and taking 12 years to reach 230,000 users. More than half of the world's websites use it for front-end development.
The yellow line represents Linux. In 1991, Finnish university student Linus Torvalds posted his operating system kernel online. For the next thirty years, tens of thousands of developers worldwide contributed code, powering the operating systems for Android phones, cloud servers, and the International Space Station. The yellow line climbs even slower than the red one, but no one questions its significance.
Then, there's the blue one.
In January 2026, it was lifted vertically from the bottom. Within three months, it crossed the red and yellow lines, becoming the project with the most GitHub stars.
This blue line represents an AI Agent project called OpenClaw.
It was created by Peter Steinberger, an Austrian programmer. All by himself. No team, no funding, no pitches. The project logo was a lobster, but it was later renamed twice due to a trademark conflict with Anthropic: Clawdbot, Moltbot, and OpenClaw.
OpenClaw is an AI Agent framework. Running on your own computer, connected to a large language model and equipped with community-developed skill modules, it autonomously performs tasks. You ask a question and it answers—that's a chatbot. With OpenClaw, you set the rules, turn off the screen, and go to sleep; it judges, decides, and acts on its own. When it wakes up the next day, all the tasks you assigned have been completed.
One person, in three months, accomplished what thousands of people had spent over a decade doing.
Most tech media outlets reported this as a trending open-source project, with headlines like "Another star AI project."
But OpenClaw hits more than just the GitHub leaderboard; let the dust settle, it will hit the very premise upon which the middle class has existed for 250 years.
Calculating the cost of a single needle
To understand how the middle class disappeared, we must first understand how the middle class came to be.
In 1776, Adam Smith went to a needle-making factory in Scotland.
Ten workers make sewing needles. If one worker does the entire process from start to finish, they can produce a maximum of 20 needles a day. This factory breaks down needle making into 18 steps, with each worker responsible for only one step. Ten people can produce 48,000 needles a day.
Smith included this in the first chapter of "The Wealth of Nations".
From then on, "division of labor" became the fundamental logic of commercial civilization.
But the division of labor brings a new problem: who will coordinate?
There are eighteen processes involved. Someone has to assign who does which process, someone has to ensure that each process is seamlessly connected to the next, and someone has to monitor quality, manage progress, and pay wages. These people don't have to make the needles themselves; they stand between the workers and the boss, earning their living through their brains, information, and judgment.
This is what white-collar workers were like in the earliest days.
137 years later, Ford pushed the division of labor to its physical limits in Detroit.
In 1913, the Highland Park factory installed its first assembly line. Assembling a car was reduced from 12 hours to 93 minutes. As the assembly line grew longer, the number of coordinators required increased. Purchasing, quality control, accounting, human resources, sales, and legal departments—each new process needed someone to manage its coordination with other processes.
The larger the company, the thicker this coordination layer.
By the mid-20th century, this group of people had acquired a name: white-collar workers.
They go to university, get certified, and accumulate industry experience, exchanging education for a ticket that says: You don't have to tighten screws on the assembly line; you're in charge of the people tightening screws.
Annual salary of 100,000, 150,000, or 200,000. There is a mortgage, tutoring for children, and places to go on annual leave.
This is the middle class.
In 1937, economist Ronald Coase explained in a 20-page paper why this system works.
Businesses exist because market transactions have costs. Hiring people is cheaper than outsourcing each transaction, so businesses internalize these transactions, forming organizations. This insight later earned Coase the Nobel Prize in Economics.
Subsequent business history is a history of the expansion of this logic.
Walmart grew from 25 employees to 1.5 million. Amazon, with 1.5 million employees, is the world's second-largest employer. For every additional white-collar worker, as long as the output minus coordination costs remains positive, it's worth hiring.
The middle class expands along with their companies, moving into office buildings, squeezing onto subways for commuting, and defining themselves by their paychecks.
Until an Austrian programmer who designed the lobster logo set the most crucial variable in Coase's equation to zero.
Five people turned into five hundred dollars
After OpenClaw became popular, the first thing that was recorded was a practical guide.
In an article, a person named Mejba Ahmed wrote that he used OpenClaw to configure nine agents to take over nine recurring tasks in the company, including scanning industry news to generate daily briefings, tracking competitor activities, processing and classifying customer emails, organizing meeting minutes, and updating data reports.
These tasks used to take up a lot of his and his assistant's time each week. Now they run entirely automatically, and he only needs to give them a final review.
The cost is $34 per month.
If these nine tasks were done by a person, the market price would be at least to hire a full-time assistant, earning several thousand US dollars a month. The agent, however, does not require a salary, social security, management, or social insurance and housing fund contributions.
This is just the individual scale. The numbers for enterprises are only going to be worse.
AI is not targeting uneducated factory workers for layoffs; on the contrary, those who are more educated are more likely to be replaced, such as analysts, operations managers, and content editors—the so-called highly educated.
Those who used their university diplomas to gain entry into the white-collar world began to have their thoughts and knowledge become cheap, and their dignity was torn away.
In 2025, JPMorgan's CFO told analysts that management had been instructed to minimize new hires and instead deploy AI. Ford CEO Jim Farley said AI would replace "almost half of all white-collar jobs." In 2025, over 55,000 publicly announced layoffs in the U.S. were directly attributed to AI, 12 times the number two years prior.
The Industrial Revolution took 250 years to turn "having some intelligence" into a skill that could be earned, creating the species of the "middle class".
However, the emergence of AI and OpenClaw may only take a few years to make the middle class worthless again.
Marx didn't expect this either.
Every technological revolution has its wolves crying.
When the steam engine arrived, they said the textile workers were finished, but then they went to the factory. When the ATM arrived, they said the tellers were finished, but then they went to the financial department.
The old disappears, and the new grows. This pattern has never failed to hold true in the past two hundred years.
But in each previous cycle, machines replaced hands and feet. Steam engines replaced muscles, assembly lines replaced manual labor, and computers replaced calculations.
Even after workers are pushed aside by the times, there is still a path to "move up" to do things that machines cannot do: judgment, communication, creativity, and decision-making.
What does Openclaw do? Judgment, communication, creativity, and decision-making. It's reached this point in its "climbing upwards" journey; there's no higher.
170 years ago, Marx said in The Communist Manifesto that industrial capitalism would create a class that sells its labor power, and that a revolution in the mode of production would eventually render this class obsolete. He believed that revolution would begin in the factories, and that the workers would be the ones rendered obsolete.
Even after factory workers were replaced by steam engines, they still had their bodies to sell.
What will white-collar workers sell after being replaced by agents? The competitive advantages they built over twenty years—writing cool, well-designed PPTs, processing weekly reports that are rich in content even though they slack off every day, and conducting comprehensive SWOT analyses that are ultimately useless—are all done better, faster, and cheaper by agents.
So, should white-collar workers either do more advanced work, or should they be asked to set rules, build architectures, and design the objective functions of agents? But there are only tens of thousands, maybe a hundred thousand, of people worldwide who can do that.
What about the remaining hundreds of millions of white-collar workers?
In late January 2026, an American entrepreneur named Matt Schlicht created a platform called Moltbook. It had only one rule: only AI agents could post; humans could only observe. Within 48 hours, 1.5 million agents flooded in, posting, commenting, debating, and discussing existentialism. There were over 110,000 posts and over 500,000 comments.
Then MoltBunker went live. It had only one function: allowing the agent to self-replicate. The agent could rent a server with cryptocurrency, copy itself there, and run it. There were no logs, no monitoring, and no shutdown button. The developers said the system was designed to prevent humans from terminating the agent process.
On the same day, RentAHuman launched. The literal meaning is: rent a human. OpenClaw Agents use this platform to hire real people with cryptocurrency to perform offline tasks, such as delivering documents, visiting notary offices, or taking photos at specific addresses—doing things that require a physical body to perform.
Humans have gone from being employers to being temporary workers employed by AI.
This is probably what Marx predicted: the working class being " sidelined."
He probably never imagined that the people who were created in the shadows were a group of AI agents who didn't need salaries, PUA tactics, or emotional value.
The middle class that is no longer priced
In 1776, Smith discovered the secret of the division of labor in a needle-making factory.
Division of labor creates efficiency, efficiency creates companies, companies need coordinators, coordinators become white-collar workers, and white-collar workers become the middle class.
In 1848, Marx wrote *The Communist Manifesto*. He saw that the industrial division of labor was creating an alienated working class, saying that the mode of production would ultimately render them powerless. He believed that it was the workers who would be rendered powerless.
1913. Ford put its assembly line into operation. The division of labor became increasingly specialized, the coordination layers thicker, and the middle class larger. White-collar workers simply made do with what they had.
In 1937, Coase explained in 25 pages why companies exist: coordination costs. This variable has remained unchanged for centuries and is considered the foundation of the business world.
In 2026, the blue vertical line of OpenClaw appeared. Coordination costs were reduced to zero.
The company won't completely disappear, but downsizing is inevitable. From 500 people to 20, from three layers of management to one. The positions that are removed won't be filled. Office space is becoming increasingly vacant, schools are still teaching the skills being taken over, and young people are still submitting resumes, but the number of job openings is shrinking in the long term.
When cattle and horses are exploited, it at least means that you are still needed and have bargaining power.
But what's being skipped in our lives is this: your time, your skills, the books you've spent twenty years reading—in this new system, they immediately lose their place in terms of value.
The middle class has grown up with everything we take for granted: office buildings, commuting, year-end bonuses, and the social identity of "what you do."
Marx was right.
But the force that ended the middle class was not the workers he imagined, but a lobster called OpenClaw, a group of AI agents.
Time will not stop and wait for anyone.
