The "one-person company" is booming: Some earn millions a year, while others see their income shrink by 90%.

The "One-Person Company", namely the model (OPC) of an individual leveraging tools to start an independent business in the AI era, has become popular.

The year 2026 is known as the "Year of the One-Person Company". With the support of AI, ordinary individuals have entered the golden age of entrepreneurship. One person plus AI can achieve entrepreneurship at the lowest cost, writing code with large models, generating content with GPT, and monetizing with AI. AI is no longer just a technological concept; it has begun to truly create commercial value for individuals.

The wealth-creation stories of "One-Person Companies" are emerging one after another. Dozens of cities across the country have introduced policies to encourage and support entrepreneurship in "One-Person Companies". According to public data, as of mid-2025, the stock of "One-Person Companies" nationwide has exceeded 16 million, accounting for 27.4% of the total number of enterprises in the country.

Of course, there have also been many controversies behind this upsurge. Some reports claim that "the first batch of bosses of one-person companies have exited the market". Some people complain that the "One-Person Company" has no essential difference from the previous entrepreneurial models of "self-employed individuals" and "brand managers". Even statistical data shows that the truth behind the frenzy of OPC one-person companies is that 52.7% of them earn less than 7,000 yuan per month.

In fact, there are risks in entrepreneurship at any time, and only a small number of people can succeed in business. AI has lowered the threshold for entrepreneurship, reduced costs and increased efficiency, but it does not guarantee the success rate. What is the real survival situation of "One-Person Companies", and who has reaped the benefits? We talked to several entrepreneurs.

Developed 6 bullet-hell games and earned 1 million yuan a year

Entrepreneur: Mr. Zhang, in the game development field

Since I left my job to start a business in March 2025, I have developed a total of 6 bullet-hell games in one year. The total turnover is close to 20 million yuan, and my personal take-home income is about 800,000 - 1 million yuan. Among them, the most successful game has accumulated a turnover of more than 8 million yuan.

Platforms such as Douyin have a relatively high commission rate. The platform takes 50% of the commission, the anchor gets 42%, and the developer only gets about 8% of the turnover in the end.

The game industry generally suffers from the pain points of "high cost and long cycle", and the "One-Person Company" perfectly solves these problems. The production cost of a single game can be controlled at 1,000 - 1,500 yuan, mainly covering the subscription fees for AI tools and the purchase of asset packages. Compared with the R & D cost of hundreds of thousands of yuan for traditional teams, it has achieved extreme cost compression.

Compared with traditional mobile games or Steam games, bullet-hell interactive games rely on platform traffic and hardly require additional publicity and promotion investment. The advantage of zero publicity and promotion has lowered the threshold for entrepreneurship.

AI tools have replaced traditional art and programming teams, achieving efficient production. During the game development process, AI has taken on about 70% of the art workload for me. A complete chain of "Doubao (keywords) -> Banana (original painting) -> Triple3D/Member 3D (modeling) -> Blender (face reduction and optimization) -> Mixamo (skeletal binding)" has been established.

AI has also enabled auxiliary code development. Using AI to write Java code, the accuracy rate can reach 99%, greatly lowering the threshold for programming development. Apart from AI, I am mainly responsible for the final assembly, UI interface adjustment, and engine adaptation work.

In terms of the game development cycle, AI has effectively compressed the production cycle. It takes a traditional team (about 10 people) 2 - 4 months to develop a similar game, while using the AI toolchain, the single-person development cycle is shortened to about 15 days. Moreover, the quality of the final product can almost match that of large companies, and is no different from the output of traditional game company teams. Ordinary players and anchors can hardly tell whether it is made by AI.

Of course, game development entrepreneurship in a one-person company requires certain core capabilities from entrepreneurs: the one-person company model is suitable for practitioners who "have channels (client resources)" or "understand the process and have ideas". AI cannot replace creativity and project management ability. Experience is the best barrier for entrepreneurship. Although AI has lowered the technical threshold, it is difficult for "zero-based" people without industry experience to handle it, because they cannot judge the correctness of the content generated by AI and make effective corrections.

Currently, there are serious layoffs in the game industry, and it is difficult to find a job in traditional positions. The application of AI tools has enabled individuals to have the productivity to compete with teams, providing a window period for independent entrepreneurship for experienced practitioners.

For me personally, in addition to self-developed games, I also use the mature AI production process to undertake outsourcing projects. The quotation for a single order is 30,000 - 50,000 yuan, and the construction period is about one week, mainly to achieve a stable cash flow.

Can't work without AI all the time, income is 3 times that of white-collar workers of the same age

Entrepreneur: Shiyue, in the material R & D field

I am an entrepreneur who started a one-person company in Japan. I have permanent residency in Japan. I used to work in the traditional material R & D industry. I established a one-person company in July last year, engaging in deep processing + retail (customized services). Currently, the company has been in operation for less than a year.

Currently, my monthly income is about 3 times the average level of Japanese white-collar workers of the same age, and my annual income is about 15 million yen, which is much more than what I used to earn at work.

Regarding the opportunity for entrepreneurship, I decided to start a business after seeing the emergence of generative AI. With the support of AI, ordinary individuals have the ability to start an independent business. 90% of the problems can be solved by querying AI tools. I can hardly work without AI tools every day: customer service translation, chatting, drawing, order processing, programming, legal advice, etc.

Currently, my cost of using AI tools is not high. I only need to pay a monthly fee of about 3,400 yen. My work does not involve high-cost API calls. The main usage scenarios are:

1. Language localization: Use ChatGPT to correct, polish, and localize professional vocabulary, and solve the problem of language differences between Chinese and Japanese.

2. Knowledge acquisition and learning: For new knowledge in the field encountered in entrepreneurship (such as taxation, regulations, processing knowledge, and business judgment), quickly query and summarize through AI to assist in decision-making.

Compared with the trend of "One-Person Companies" in China, there are not many people starting one-person companies in Japan, but the number has increased compared with before. The number of Japanese people starting one-person companies is doubling. The threshold for foreigners to start a business in Japan is relatively high: they need to apply for a business management visa, which requires an initial investment of at least 30 million yen; without permanent residency or a stable work visa, the threshold for entrepreneurship is extremely high. The qualification requirements for different industries in Japan vary greatly. Catering, medical aesthetics, second-hand trading, etc. all require specific business licenses. The qualification requirements for my industry are relatively loose.

The industry barriers in Japan are very high. In the Japanese market, establishing the credit rating of an individual or a company is a key prerequisite for obtaining supplier resources. The Japanese market is extremely exclusive. To enter an industry, you need at least an "insider" to introduce you. Otherwise, it is difficult to establish a trust relationship. Obtaining supplier resources requires a strict credit review process. In the early stage of my entrepreneurship, the business progress was very slow. In the past six months, through word-of-mouth accumulation, it has achieved exponential growth, and the business has now become stable.

Compared with China, the Japanese civilian market has a slower acceptance of new things, such as intelligent tools and technology. Scanning codes to place orders only became popular in recent years, which provides opportunities for Chinese entrepreneurs to fill the market gap. There are also certain differences in the competitive environment between the two markets. The Japanese market is not as "intensely competitive" as the Chinese market. The population base is small and the industry competition is relatively mild, so there is a certain living space for one-person companies. Low-price competition doesn't work well in Japan. Japanese consumers are skeptical of low-price products. Adopting a high customer unit price strategy is more likely to build trust.

AI undertakes 80% of repetitive work, income doubles

Entrepreneur: Xiaotao, in the biotechnology field

I started my business in August last year. A one-person company does not create customers. I was introduced by a former colleague to the customer needs in the biotechnology field. I had customers first and then decided to establish a one-person company to undertake the business.

The industry has a certain threshold. I mainly provide biological cell culture and development services for B2B customers. The core is to use algorithms to optimize the formula of biological culture media and improve prediction efficiency. I had accumulated some experience in AI algorithms in a biological company before. That company focuses on the field of biological culture media development, which only a few institutions in the world can do. Using AI to predict the formula has formed a relatively high technical threshold.

Compared with a full-time job, my one-person company model has doubled my income (about twice), and the freedom of time has been significantly improved, achieving a higher input-output ratio.

The core of my daily work is to use the DeepSeek model and combine it with the Claude model to build an AI Agent. The development environment is based on VS Code, and automated operations are achieved through the Agent.

Actually, the computing power cost is not high. Currently, the Token cost of using DeepSeek is less than 30 yuan, and the hardware investment is only a personal computer. The overall startup cost is extremely low. The AI Agent undertakes about 80% of the basic repetitive work (such as data analysis and report generation), and the remaining 20% of core judgment, framework design, and customer communication are mainly handled manually.

20% of the work cannot be replaced by AI. Although AI undertakes a large amount of work, entrepreneurs need to monitor the work progress and output direction of the Agent in real time to prevent it from deviating from the goal and ensure the delivery quality.

Personally, I think that positions that rely on electronic repetitive labor, such as data analysts, outsourcing product managers, and administrative positions, are at a relatively high risk of being directly replaced by AI Agents. At the same time, these people are also suitable for entrepreneurship: professionals with a technical background, business understanding ability, and customer resources are the best candidates for transforming into one-person companies. If they can master AI technology and have customer resources, they can try one-person entrepreneurship.

Currently, I have been in business for less than a year and have encountered many challenges. There are certain operational shortcomings. Entrepreneurs need to make up for non-technical shortcomings, such as interpreting government policies, company registration procedures, and handling fiscal and tax compliance, which usually need to be solved with the help of external resources. It is recommended not to blindly engage in "One-Person Company" entrepreneurship. Introverted people (I people) have natural disadvantages in customer acquisition and financing roadshows. They must ensure that they have stable customer resources before starting a business.

Income has shrunk by 90% compared with working in a large company. It is not recommended to blindly follow the trend of entrepreneurship

Entrepreneur: A Yuan's Free Navigation, in the cross-border e-commerce field

I used to work as an operator in a large Internet company. As the industry growth slowed down and I faced an invisible career ceiling, I decided to leave my job and start a business. The direction of my entrepreneurship is cross-border e-commerce in the Latin American market.

I was sent to Latin America for work before, so I am quite familiar with the Latin American market. Moreover, the trade between us and the Latin American market is growing rapidly and has great potential, so I chose to focus on the Latin American market. Then, due to lack of experience in the early stage of entrepreneurship, I first joined a startup team. I officially became independent in April this year and adopted the "One-Person Company" light-asset model, controlling all links from product selection to after-sales service in the whole process.

For me personally, the threshold for starting capital of a one-person company is not high. Registering a company is free, and the cost of bookkeeping services is 2,000 yuan per year. It is basically free to enter the e-commerce platform. Depending on the selected business model, you need to purchase a Mexican local tax number for less than 10,000 yuan.

Currently, I am mainly targeting the Mexican market and adopting the "overseas warehouse" model, that is, stocking goods in advance in the official warehouse of the platform to reduce logistics costs and improve fulfillment efficiency.

AI has replaced about 60% of the manual workload. For light-weight cross-border e-commerce entrepreneurship, product selection and optimized display of product pages are the most important. And this part of the work can almost all be completed by AI. AI tools cover aspects such as market product selection, generation of graphic/video content, optimization of product detail pages, and financial data analysis.

I mainly use GPT, Gemini, and Doubao video generation tools in my personal work habits. The average monthly subscription cost is more than 200 yuan, which can be said to have significantly reduced the operating cost compared with traditional manual design. Generally, the cost of manually designing pictures for a single product is as high as 500 yuan.

Of course, it is not recommended to blindly quit your job and start a business. There is an over - hyped phenomenon of "One-Person Companies" on the Internet, and you can't blindly follow the trend. Working in a large company still has a high cost - performance ratio. In the past, my annual salary in a large company was about 600,000 yuan. Currently, after two months of entrepreneurship, the sales volume is more than 10,000 yuan, and the profit is more than 1,000 yuan. Even with part - time agency operations, there is still a huge gap in income, which has directly shrunk by 90%.

However, the growth value brought by entrepreneurship is indeed far greater than the work experience of the past few years. There is a famous Western saying: "Get your hands dirty". Entrepreneurship also allows oneself to truly learn and understand the business deeply, shift from the perspective of the platform to front - line practice, deeply experience the hardships of entrepreneurship, and have a more down - to - earth understanding of the essence of business. The whole person has also become more down - to - earth.

This article is from the WeChat official account "Tech Planet" (ID: tech618), author: Zhai Yuanyuan, published by 36Kr with authorization.