Ahmed Shaami
MBO Partners counted 5.6 million American independent workers earning more than 100,000 dollars a year in 2025, up 19 percent from 4.7 million the year before. Over the same period the total independent workforce moved from 72.7 million to 72.9 million, which is flat in any practical sense. Earnings at the top are climbing quickly while the number of people doing independent work stands still. That gap, rather than the headline workforce number, is the finding worth acting on if you sell services.
Flat headcount, a fast-growing top tier
The figures come from the fifteenth annual State of Independence in America study, which surveys the US independent workforce each year. A 19 percent jump in six-figure earners against a workforce that grew by roughly 200,000 people means the increase is not coming from new entrants arriving and immediately earning well. It is coming from people already in the market moving up.
Put differently, the same population is splitting into two groups that are drifting apart. Reporting on the study by Staffing Industry Analysts noted that part-time independents fell for a third consecutive year, down to 7.9 million, while full-time independents held roughly steady at 27.6 million. The shrinking part-time tier and the swelling six-figure tier are two views of the same movement.
Where the growth actually came from
Almost all of the headcount growth landed in one category: occasional independents, people who earn irregularly and periodically, now totalling 37.4 million. That group is more than half the entire independent workforce, and it is the part least likely to be building a business.
This matters because the raw “72.9 million freelancers” statistic gets quoted constantly as evidence of a booming market. Most of that number is occasional and part-time activity. The portion running independent work as a primary income stayed flat. Anyone deciding whether to go full-time on the strength of workforce growth is reading a number that does not describe their situation.
The AI adoption figure underneath it
The study also found 74 percent of independent workers using generative AI, up from 65 percent the previous year. Adoption at that level in a single year is unusual, and it changes what the tools are worth competitively. When a minority used AI, using it well was an advantage. At three quarters adoption, it is the baseline, and the advantage shifts to what you do with the time it frees.
The platform data points the same direction. Upwork reported that demand for skills explicitly referencing AI grew 109 percent year over year in its In-Demand Skills 2026 release, against 23 percent for other in-demand skills, with AI video generation and editing up 329 percent. Clients are not buying “knows how to use AI” as a service. They are buying outcomes that AI happens to make feasible, which is a different pitch. We have written before about how Upwork clients now search for AI roles rather than AI tasks, and the skills data supports that reading.
Why does platform work sit on the wrong side of this split?
Upwork’s own results show a marketplace consolidating rather than expanding. In its second quarter 2026 results, active clients fell 4 percent year over year to 763,000, while gross services volume per active client hit a record 5,230 dollars, an eighth consecutive quarter of sequential growth. Fewer buyers, each spending more.
An Irish Times analysis published on 18 September 2026 argued that white-collar platform work is unusually exposed to automation precisely because of how it is packaged: jobs broken into discrete remote tasks, stripped of institutional knowledge and personal relationships. A task defined clearly enough to be posted, priced and delivered by a stranger is also defined clearly enough to be handed to a model. The packaging that made the work easy to sell is what makes it easy to replace.
That explains why the middle is thinning while the top grows. The work most resistant to substitution is the work that requires knowing a client’s business, their constraints and their history, and none of that fits in a job post.
What the high earners are actually selling
Three characteristics show up consistently in the work that is holding its price.
Accountability for a result. Charging for a deliverable invites comparison against whatever produces that deliverable fastest. Charging for an outcome, with responsibility for whether it worked, does not compare cleanly against a tool. Our guide to which AI-adjacent work pays more and which is getting cheaper breaks that split down by service type.
Accumulated client context. The second month with a client is more profitable than the first because you stop rediscovering things. Context compounds, and it does not transfer to a competitor bidding on a fresh brief.
Access to larger buyers. Upwork’s Business Plus tier grew its client count 219 percent year over year off a small base, which is where the spending concentration is showing up. We looked at that shift in detail in our piece on where the new clients are arriving on Upwork in 2026.
Moving from task work to retained work
The practical version of all this is unglamorous. Stop optimising for the volume of jobs won and start optimising for the depth of a small number of relationships.
Pick three past clients and write down what you know about their business that a new freelancer would take two months to learn. That list is your actual product. Then look at what you currently charge and whether it reflects the result you are responsible for or the hours you spent producing something. For anyone pricing this kind of work from Pakistan, our breakdown of AI automation freelance rates and what to charge gives concrete ranges to anchor against.
The workforce is not growing. The money inside it is redistributing, and it is moving toward people who are difficult to swap out. Being difficult to swap out is a thing you build deliberately, over a small number of clients, across a longer period than most people are willing to spend.
