Something large is happening in the United States, even if it is not always easy to see at first glance. Companies with no employees, often called one-person companies, are growing quickly.
The Numbers Show the Shift
According to Carta's 2025 report on solo founders, 35% of new U.S. startups formed in 2024 were started by a solo founder. In 2017, that figure was 17%. In seven years, the share roughly doubled. The traditional startup model of three to five co-founders building a team together is moving in the opposite direction.
The broader numbers are even larger. According to recent U.S. Census Bureau data, there are now 29.8 million solopreneurs in the United States, contributing $1.7 trillion to the economy. That is 6.8% of total U.S. economic activity. The U.S. Small Business Administration's data points in the same direction: more than 80% of small businesses in the United States operate without employees, with only the owner running the business. The old assumption that you need to hire people before you can start a business no longer holds.
Forbes reported in March 2025 that the number of one-person businesses generating more than $1 million in revenue doubled in a single year. This is not just a story about tiny lifestyle businesses. One-person companies are becoming capable of producing meaningful revenue.
Two forces sit behind this shift. The first is generative AI. Code, design, marketing copy, customer support, and accounting tasks that once required several people can now be handled, at least in part, with AI tools. The second is a change in how people think about work. According to Gusto's 2025 report, 54% of solopreneurs started their businesses to be their own boss, and 53% did so for schedule flexibility. Among Gen Z respondents, 62% said they either plan to start a business or are likely to do so.
Why AI Makes One-Person Companies Possible
There is a clear reason one-person companies are growing so quickly: AI is making previously impossible workflows possible.
Ten years ago, starting a startup usually required at least one person to build the product, one person to validate the market, and one person to run the business. When Instagram co-founder Mike Krieger added a photo filter in 2010, the work could take weeks. Resource constraints were tightly tied to headcount.
That situation has changed. Krieger, now back at Anthropic as CPO, recently used Claude to build a prototype in 25 minutes. The same kind of work used to take him six hours. Inside Anthropic, Claude is now used for a substantial share of coding work, and usage of Claude's code agent has reportedly grown by 40%.
The range of work AI can support is broad. GitHub Copilot research found that developers using AI completed tasks 55% faster, with a completion rate of 78% compared with 70% for the non-AI group. One solo retail founder saved more than 15 hours a week by combining ChatGPT and Canva. Companies adopting AI tax automation tools reported an average 15% reduction in tax burden and a 90% reduction in classification errors.
Another technical shift is accelerating this change. Anthropic's Model Context Protocol (MCP) lets AI connect directly with tools such as GitHub, Notion, Stripe, and Webflow. A user can give instructions in natural language, and the AI can write code, deploy changes, manage customer support, update documents, and track metrics. AI is no longer only a conversational chatbot. It is becoming a collaborator embedded in digital workflows.
In the past, doing more usually meant hiring more people. AI breaks that formula. One person can now handle work that used to require five to ten people, and the cost can fall by orders of magnitude.
Layoffs and AI Automation
On the other side of the one-person-company boom are existing companies. They are also in the middle of the same shift, and their direction is clear: reduce headcount and replace work with AI.
The World Economic Forum's Future of Jobs Report 2025, published in January 2025, gives a sense of the scale. The survey included more than 1,000 major employers representing over 14 million workers. Respondents identified technological change as the biggest force reshaping labor markets from 2025 to 2030, and the report expects large-scale job restructuring during that period.
Anthropic is a symbolic example of this shift. Claude generates a substantial share of Anthropic's internal code, and human developers are shifting from writing code directly to reviewing AI-generated code and setting direction. Amazon has also started restricting junior and mid-level engineers from submitting AI-generated code without approval from senior engineers.
Existing companies have an obvious economic reason to reduce headcount. The same work can now be done by fewer people and done faster. But this strategy has a structural limit that many companies can miss. Costs may go down, while the speed of decision-making remains unchanged.
Costs Fall, but Decisions Stay Slow
AI tools have clearly increased the execution speed inside companies. Code is written faster. Marketing content is produced faster. Customer support gets processed faster. Accounting work can move faster. Across many functions, AI shortens the time required for repetitive work.
But there is an important question: if execution became faster, did decision-making become faster too?
The short answer is no. If AI can write code in one day, but the company needs three weeks to decide what should be built, the total cycle is still three weeks.
Research from Harvard Business Review and Clayton Christensen's Innovator's Dilemma have long highlighted this problem. Large-company decision-making comes with meetings, approvals, cross-functional coordination, and risk avoidance. As the number of stakeholders grows, the number and duration of meetings can grow exponentially. As approval chains lengthen, the time between decision and execution stretches out.
In The Innovator's Dilemma, Christensen argued that the very success patterns of large companies can become structural barriers to innovation. Processes optimized around existing customers, decision structures built around existing products, and the desire to protect stable revenue in existing markets can all delay decisions about new markets, new technologies, and new ways of working.
This is not just a matter of slow decisions. It is a matter of decision structures that are designed to resist change.
Small Teams Decide Faster and Run More Experiments
While existing companies reduce costs without necessarily increasing decision speed, another kind of competitor is moving differently: one-person companies and small teams.
A one-person company has no meetings. There is no approval chain. There is no cross-functional coordination. The person making the decision is often the same person executing it. In the time an existing company takes to make one decision, a one-person company may have already made and acted on ten.
The hypothesis-test-learn-revise cycle Eric Ries proposed in The Lean Startup can run its course in days inside a small team. In the past, experiments were expensive. A single experiment could require tens or hundreds of thousands of dollars, limiting how many experiments a team could afford. AI fundamentally lowers the cost of experimentation.
The Indie Hackers community has many public examples of one-person businesses generating millions of dollars in annual revenue. These founders often raise no outside funding and hire no employees. Darli, an AI chatbot for African farmers, serves 110,000 users with a small team.
One way to describe this shift is the democratization of experimentation. Experiments that once required major capital and large teams are now possible for one person with AI tools. As capital and execution gaps shrink, only the quality of ideas and the speed of decisions remain.
Same AI, Different Results
AI tools are available to both existing companies and one-person companies. Claude is used by Anthropic's internal developers and by solo founders. GitHub Copilot is used by enterprise engineers and freelance developers alike.
But the outcomes are different. Why does the same AI produce different results?
The tool itself is neutral. The environment around the tool is not. That difference in environment changes the outcome.
In a one-person company, the moment a decision is made, AI can begin writing code. In an existing company, AI often waits until the decision is approved. Execution may be faster, but if decision speed does not change, the total speed gain is limited.
A one-person company hears customer feedback directly, analyzes it directly, and applies it directly. In an existing company, feedback is collected, analyzed, reported, discussed in meetings, prioritized, and then handed to the development team. The loop from learning to execution is much longer.
A one-person company has less to lose when an experiment fails. In an existing company, a failed project can waste people, time, and budget. That makes companies more conservative about experimentation.
The AI may be the same, but the structure produces different results.
The Market Share One-Person Companies Can Take
The markets where one-person companies have an advantage are becoming clearer. AI-native niches are markets that might not have existed without AI, where technical barriers are lower and fast iteration matters. In professional services, AI tools can absorb repetitive work, allowing one expert to handle the amount of work that once required a small team. In local or language-specific markets, one-person companies can find opportunities in places that are too small or too specific for large companies to prioritize.
There are also areas that existing companies struggle to defend: markets that require rapid improvement, markets with low barriers to entry, and niche markets too small to attract serious attention from incumbents. Indie Hackers includes many examples of one-person SaaS companies reaching millions of dollars in annual revenue.
The next generation's view of work will accelerate the shift. With 62% of Gen Z expressing interest in starting a business, the rise of one-person companies is likely to continue.
How Existing Companies Can Survive
Existing companies can still adapt. But the direction of that adaptation needs to be different from what many are doing now.
They need to redesign decision-making. Approval chains should be shorter, and people close to the work should have the authority to decide and execute. Leadership should focus more on direction and strategic judgment. Meetings, presentations, and reporting should be reduced where they do not improve decisions.
They need to allow more experiments. If AI lowers the practical cost of experimentation, organizational culture needs to lower the psychological cost of failure. A company that cannot tolerate failed experiments cannot move at the speed AI makes possible.
They need to use AI for decision support, not only task automation. Most corporate AI adoption still focuses on automating execution. But AI can also support decisions through data analysis, market forecasting, customer behavior analysis, and risk assessment. Used well, it can improve both the speed and quality of decisions.
They also need to use the advantages of scale. Large investments, existing customer relationships, brand awareness, specialized talent, and trust are hard for one-person companies to build quickly. Those assets still matter, but they need to be deployed in a structure that can move at startup speed.
Which Path Would You Choose?
There is the path of the existing company: use AI to cut costs, increase productivity, and get more out of the remaining workforce. It is the more stable path. But if decision-making does not become faster, more and more market share will leak to smaller competitors.
There is the path of the one-person company: use AI to build a product alone or with a small team, find a market, and generate revenue. It is the freer path. But it also means carrying every responsibility yourself, including the full risk of failure.
There is no single correct answer. The right path depends on a person's situation, temperament, and goals.
But one thing is clear: this change is not going away. In the WEF survey of more than 1,000 major employers, technological change was identified as the biggest force reshaping labor markets from 2025 to 2030. This is not just a prediction. It is already happening.
If you are an employee, you can start by using AI tools to improve your own work. If you are thinking about starting a company, this may be the best timing you have ever had. If you run a company, now is the time to redesign the structure.
Which path would you choose?