> ## Content Index > Fetch the complete content index at: https://www.groktop.us/llms.txt > Use this file to discover other available public pages before exploring further. # Fiverr’s AI Pivot Is Unproven: Who Gets to Move Up? - URL: https://www.groktop.us/fiverr-ai-pivot/ - Published: 2026-10-03T18:48:51.000Z - Updated: 2026-10-03T18:48:51.000Z - Description: Fiverr is chasing bigger projects as buyers leave. AI is shifting freelance demand, but its gains don’t reach every worker. - Author: Magnus Hedemark - Tags: Enterprise AI, AI Strategy, Future of Work [Fiverr’s latest results show a marketplace losing buyers and marketplace revenue while remaining buyers spend more](https://investors.fiverr.com/news-releases/news-release-details/fiverr-announces-second-quarter-2026-results?ref=groktop.us). That’s a serious business problem. It doesn’t prove corporate collapse, and it says little by itself about what freelancers took home. The harder question is what buyers stop outsourcing and who gets access to the work that grows. A [peer-reviewed study of one anonymous freelance platform](https://discovery.ucl.ac.uk/id/eprint/10204198/?ref=groktop.us) found lower demand for some tasks that artificial intelligence (AI) could substitute for, and mixed results for work the technology could complement. A separate [study of highly AI-exposed freelance occupations](https://profiles.wustl.edu/en/publications/the-short-term-effects-of-generative-artificial-intelligence-on-e/?ref=groktop.us) reported short-term employment and earnings losses. Those findings make this harder to describe as either “AI killed gigs” or “AI created better gigs.” ## Fiverr is chasing bigger projects as buyers leave In the second quarter of 2026, Fiverr reported that marketplace revenue fell 15.5% year over year and annual active buyers fell 21.9%. Annual spend per buyer rose 15.6%, services revenue rose 2%, and the company said clients completing projects of at least $1,000 grew 13% on a trailing twelve-month basis, according to [Fiverr’s Q2 results](https://investors.fiverr.com/news-releases/news-release-details/fiverr-announces-second-quarter-2026-results?ref=groktop.us). ![Vintage engraved two-panel data plate on parchment. Title: Fiverr Q2 2026: Mixed Signals. Left heading, Marketplace: active buyers down 21.9% and marketplace revenue down 15.5%, both with down arrows. Right heading, Remaining Business: spend per buyer up 15.6% and services revenue up 2.0%, both with up arrows. Footer: Year over year. Small modern freelancers work at a laptop and tablet beneath the figures. Arrows indicate direction only, not relative magnitude. The comparison shows falling marketplace activity alongside higher average spend and a small services increase.](https://storage.ghost.io/c/f1/0e/f10e80f4-9285-43fc-acd4-35910a12c5f0/content/images/2026/10/inline-01-fiverr-metrics.png) Fiverr had fewer marketplace buyers and less marketplace revenue, alongside higher spend per buyer and services revenue. Those numbers describe a marketplace under strain, but they don’t say what happened to freelancer income. In the same release, Fiverr [defines annual spend per buyer as marketplace gross merchandise value divided by active buyers](https://investors.fiverr.com/news-releases/news-release-details/fiverr-announces-second-quarter-2026-results?ref=groktop.us). When lower-spending buyers leave, the average for those who remain can rise even if no individual freelancer receives a raise. The release also says total company revenue fell 10% year over year and attributes revised guidance to AI-related demand and traffic headwinds. That is management’s explanation, not a measurement that isolates AI’s share of the decline. The buyer count was already falling before Fiverr’s Q2 2026 outlook cited AI-related demand and traffic headwinds. For full-year 2025, total revenue grew 10.1% to $430.9 million, while marketplace revenue slipped 1.8% and annual buyers fell 13.6%. Gross merchandise value from transactions above $1,000 grew 22.8%. A year earlier, buyers were down 10% even as total revenue grew 8.3%, according to [the 2025 results](https://investors.fiverr.com/news-releases/news-release-details/fiverr-announces-fourth-quarter-and-full-year-2025-results?ref=groktop.us) and [the 2024 report](https://investors.fiverr.com/news-releases/news-release-details/fiverr-announces-fourth-quarter-and-full-year-2024-results?ref=groktop.us). Those declines came after ChatGPT went public in November 2022, as [the independent study’s event timeline notes](https://discovery.ucl.ac.uk/id/eprint/10204198/?ref=groktop.us). The dates make AI a plausible part of the story, but they don’t show how much of the buyer decline it explains. Fiverr didn’t wait until the latest slump to change direction. Its [2025 results describe a move toward larger, more complex projects](https://investors.fiverr.com/news-releases/news-release-details/fiverr-announces-fourth-quarter-and-full-year-2025-results?ref=groktop.us). In February 2025, it launched Fiverr Go, which it described as a personal AI creation model trained on a freelancer’s work, plus an assistant for client communication and routine tasks. At launch, access was limited to vetted creators with established track records, as [Fiverr’s announcement said](https://investors.fiverr.com/news-releases/news-release-details/fiverr-unveils-fiverr-go-visionary-ai-platform-puts-creators?ref=groktop.us). Fiverr’s product announcement documents the strategy. Its latest reported marketplace figures still show fewer buyers and lower marketplace revenue, but they don’t measure freelancer income. ## AI’s effect on freelance work is uneven Evidence from other online freelance markets shows why Fiverr’s results don’t tell the whole story. A 2025 study of several million daily job postings on one anonymous global freelance platform, collected from January 2021 through September 2023, compared skills that large language models might replace, complement, or leave largely unaffected. The researchers found a roughly 25% relative decline in postings for substitutable skill clusters after ChatGPT’s release. The decline was concentrated in short-term projects expected to last one to three weeks. The platform wasn’t named, and the study measured postings rather than filled jobs or wages, as [the authors’ University College London (UCL) record and accepted paper explain](https://discovery.ucl.ac.uk/id/eprint/10204198/?ref=groktop.us). ![Square data plate titled Freelance Tasks After ChatGPT. The first section is headed Substitutable Skills and reads: About 25% fewer postings vs unaffected; Short-term jobs fell most. The second section is headed Complementary Skills and reads: No significant aggregate effect; Some AI clusters grew; Novice demand fell in some clusters. Footer: One anonymous platform; postings, not wages. A down arrow marks substitution; a neutral fork marks mixed outcomes. This is one platform’s posting evidence, not Fiverr or the whole freelance market.](https://storage.ghost.io/c/f1/0e/f10e80f4-9285-43fc-acd4-35910a12c5f0/content/images/2026/10/inline-02-labor-sorting-1.png) The study found a relative decline in substitutable postings, mixed complementary results, and weaker novice demand in some complementary skills. The complementary results were more modest. The [study found no significant aggregate effect](https://discovery.ucl.ac.uk/id/eprint/10204198/?ref=groktop.us) across those skill clusters, even though some AI-related categories grew. It also found lower demand for novice workers in complementary skill clusters. Someone whose short-term gigs disappear doesn’t automatically get a path into those newer AI-related jobs. A separate [2024 study in *Organization Science*](https://profiles.wustl.edu/en/publications/the-short-term-effects-of-generative-artificial-intelligence-on-e/?ref=groktop.us) looked at employment and earnings, rather than postings. Its abstract reports short-term reductions for freelancers in highly AI-affected occupations. Past performance did not appear to protect workers from adverse employment effects, and the researchers describe only suggestive evidence that top freelancers were disproportionately affected. The study’s institutional record doesn’t name the platform, so it can’t be presented as Fiverr evidence. Still, it complicates the comfortable story that beginners lose while experienced specialists always win. Postings, filled contracts, and earnings answer different questions. The [postings study](https://discovery.ucl.ac.uk/id/eprint/10204198/?ref=groktop.us) and the [employment study](https://profiles.wustl.edu/en/publications/the-short-term-effects-of-generative-artificial-intelligence-on-e/?ref=groktop.us) don’t establish that the entire gig economy is shrinking. They do show that AI exposure can hit some categories and workers directly. A listing is not a completed contract, and a result in highly exposed occupations is not a market-wide total. Buyer counts alone miss whether workers lose access to paid assignments. ## Who gets the value in AI-enabled freelance work? Upwork’s 2026 index offers one sign of expansion in skilled freelance work, though the company sells services in this market. It reports that skilled freelancers represented 38% of surveyed knowledge workers in the United States, up from 28% in the prior-year index. The survey covered 2,400 skilled workers based in the United States. That’s an encouraging participation signal, not a census of gig work or proof that AI caused more people to freelance, according to [Upwork’s report and methodology](https://investors.upwork.com/news-releases/news-release-details/upworks-future-workforce-index-2026-how-ai-redefining-value-work?ref=groktop.us). ![Vintage engraved data plate titled AI Work: Volume and Pay. Row 1: Complex AI Work; up arrow; +45%; Earnings. Row 2: Generative AI Creative Production; up arrow; +90%; Contract Starts. Row 3: Generative AI Creative Production; down arrow; −13%; Earnings per Contract. Footer: Year Over Year · Upwork Marketplace Data. Arrows mark direction only, not relative size. Upwork-reported platform metrics, not causal evidence or outcomes for all freelancers. The three rows represent separate categories, not one combined total.](https://storage.ghost.io/c/f1/0e/f10e80f4-9285-43fc-acd4-35910a12c5f0/content/images/2026/10/inline-03-ai-work-value-v3.png) Upwork reports 45% higher earnings for complex AI work; generative-AI creative-production starts rose 90%, while per-contract earnings fell 13%, year over year. Upwork’s marketplace data complicates that participation signal. The company says freelancers doing more complex work with AI saw earnings rise 45% year over year, while AI-augmented professional services grew in volume and earnings. Yet generative-AI and creative-production contract starts rose 90% as earnings per contract fell 13%. These are proprietary platform measures and associations, not proof that AI caused the earnings changes. More work can appear while the price of each piece falls. So what counts as empowerment? [Upwork’s report on AI-augmented professional services](https://investors.upwork.com/news-releases/news-release-details/upworks-future-workforce-index-2026-how-ai-redefining-value-work?ref=groktop.us) and [Fiverr Go’s creator-tool announcement](https://investors.fiverr.com/news-releases/news-release-details/fiverr-unveils-fiverr-go-visionary-ai-platform-puts-creators?ref=groktop.us) point toward ways AI may help freelancers deliver larger services or handle routine work. Neither establishes that workers broadly kept the gains. The [freelance-posting study](https://discovery.ucl.ac.uk/id/eprint/10204198/?ref=groktop.us) and [employment study](https://profiles.wustl.edu/en/publications/the-short-term-effects-of-generative-artificial-intelligence-on-e/?ref=groktop.us) show why a tool that expands what someone can produce does not guarantee more paid work. AI access alone is not empowerment; workers need access to valuable assignments, a way to retain part of the added value, and a route into the market. For a deeper look at how AI can change the shape of a job rather than erase an entire role, see [AI Is Recombining Jobs](https://www.groktop.us/recombining-jobs/). And for the question of what human judgment contributes when AI handles more execution, [The Best AI Companies Are Hiring for More Judgment](https://www.groktop.us/hiring-for-judgment/) offers a related lens. [Fiverr’s results make it a useful warning, not a proxy for every freelancer](https://investors.fiverr.com/news-releases/news-release-details/fiverr-announces-second-quarter-2026-results?ref=groktop.us). Its marketplace is contracting while the company tries to move toward larger projects and creator-focused AI. Whether that move can offset the buyer and marketplace revenue declines remains open. The [UCL postings study](https://discovery.ucl.ac.uk/id/eprint/10204198/?ref=groktop.us) and the [employment study](https://profiles.wustl.edu/en/publications/the-short-term-effects-of-generative-artificial-intelligence-on-e/?ref=groktop.us) offer a more specific picture than a gig-economy collapse: some tasks lose demand, some new work appears, and workers do not share those outcomes evenly. The real measure is who can reach valuable assignments and how much of the resulting value freelancers keep.