The state of indie hacking in 2026

Indie hacking in 2026 is the same bet as in 2016, under worse conditions for the easy version of the bet. You own a product. Customers pay. A funding round is not the plan. What changed is the price of shipping, the price of attention, and the credibility of a screenshot. This page is that present tense. The eras that produced it are the timeline. The argument about extinction is whether it is dead.

The picture below stops in late September 2026, because that is when this page was written. A live feed will move. The structure will not move as fast: cheap builds, expensive attention, ugly base rates.

The base rate did not become a success story

More people start alone. Carta put solo founders at 36.3% of new U.S. startups on its books in the first half of 2025, from 23.7% in 2019. Stripe Atlas says 20% of Atlas companies reach a first paying customer within 30 days, up from 8% in 2020. Starting is faster. A first customer is not a company you can live on.

ChartMogul still shows a minority of SaaS cohorts reaching $1 million ARR within three years. The 2021 cohort’s 10.1% is the weak published mark they highlight. Easier editors did not buy better odds of scale.

At the indie-visible end, Steal What Works compiled a TrustMRR snapshot: 67.8% of businesses under $1,000 in lifetime revenue, 0.9% over $1 million. Self-selected, lifetime rather than MRR, and the board moves. The shape is the state. Most attempts are small. The feed photographs the tip. Benchmarks and can you still make money hold the caveats in one place. Nothing in 2026 repealed the pyramid.

Shipping is cheap

ChatGPT opened in November 2022. In February 2025 Andrej Karpathy named vibe coding: describe the thing, accept the generated code, keep moving. He was talking about throwaway projects. The culture applied the name to anything small.

On 30 July 2026 Pieter Levels wrote that indie hackers might be first to lose execution as a moat, and he put a building stack at about $20 a month. Customers, in that essay, vibe-code a private copy of a thin tool. In September he posted a cancel-list of his own vendors and a founder-reported savings near $25,000 a month. The stack story is the cost side. The killing question is the moat side. Both can be true. Cheap building helps the owner who already has demand, and it destroys the owner whose only demand was “you cannot build this.”

Marc Lou’s March 2026 month is the portfolio version. He posted $78,120, led by TrustMRR and DataFast, with ShipFast and CodeFast smaller, and he said AI had hurt the course and the boilerplate. The shortcut he sold in 2023 was less of a shortcut once agents could wire auth and payments. His record is adaptation across products, not loyalty to one SKU.

Attention is the product being sold

If everyone can ship, being seen is the scarce seat. The scene spent 2026 selling the seat.

Verification first. Fake MRR screenshots were common enough that Lou’s newsletter dates a Levels call-out to 28 October 2025 and TrustMRR to the next morning. Read-only Stripe, a directory, a place to list a startup for sale, and ads beside the proof. He wrote that sidebar ads made $20,378 in three days. That figure is his. The category stuck. By March 2026 TrustMRR was the top line in his month, which he tied partly to a 3% acquisition fee. Proof became a business. The ads did not fully leave.

Bidding next. Jonathan Wilke launched outbid.lol on 19 August 2026. Pay more than the person above you, sit at number one, no revenue proof required. A week-one recap reported about 1.36 million visitors, about $221,000, and hundreds of clones. Those are reported figures, not an industry audit. A tracker, afford.bid, was listing hundreds of boards within the month. Clones are the saturation tell. Ads versus bidding separates a sponsor slot on a site that already has readers from a rank that is only a price.

The stunt end of the same instinct is physical. Marc Lou auctioned temporary tattoo slots for a HYROX race and the closed board summed to $112,000 on his site. The body-ad story is a week, not a strategy guide. It rhymes with bidding because the inventory is attention.

Rob Hallam sold a different look: a pre-order for an X-growth course, price rising every 50 sales, videos dated 31 October 2026. A morning-after screenshot showed a $14,481.90 balance. The pre-sale story keeps that figure inside the dashboard’s own labels. SuperX, the SaaS, is a separate meter, around $20,000 MRR depending on which public number you pick that week. Attention is both his tool and his curriculum. Most products are not that.

Agents showed up as users

The other 2026 plot is a buyer that is software. Postiz, Nevo David’s scheduler, added a public API, a CLI, a skill file, and an MCP server. A stranger’s article about an OpenClaw agent using it traveled. Founder-reported MRR moved from the teens into the area of $143,000 by late July 2026, in his Starter Story telling. On 15 September 2026 TrustMRR showed about $222,851 MRR, Stripe-connected. The story does not invent how much of that came from agents. It does say the product’s pitch leaned toward being callable. The hub is the short version.

Anthropic’s Model Context Protocol, from November 2024, is the socket people point at. The state in 2026 is that a small SaaS can win by being the default tool an agent installs, and can still need a human to pay. That is a new channel, not a repeal of churn.

What the word covers now

The definition did not shrink to “person who posts MRR.” It still covers job boards, directories, info products, ads, and subscription software. The feed covers a narrower costume: solo, SaaS, public, often selling to other makers. Oversaturation is about that costume and about generic clones. It is a bad description of every niche where the buyer does not read founder Twitter.

Versus a startup still matters. Cheap shipping did not make venture math kinder. A small profitable shop remains a finished outcome in the indie plan and a disappointment in a fund plan. People keep using the words as if they were one career ladder. They are two finishes.

Courses and boilerplates are still for sale, and the honest 2026 note is Lou’s: AI hurt the ones that sold setup and syntax. Courses that sell judgment about distribution may outlast courses that sell steps a model can do. That is a prediction about a shelf, not a promise.

What a serious reading looks like

Believe specific, dated, sourced claims more than mood. Levels’ shops in a $200,000–$250,000 month band are a different sentence from a $10 million mix of revenue and gains on 23 September 2026. The story exists so the sentences stay apart. Lou’s $3 million is a decade. Wilke’s week is a week. Postiz’s TrustMRR print is Stripe for that account, not a category report.

Believe channels less than before. Informational SEO lost clicks where AI summaries appear. Pew and Ahrefs are the citations on the dead page. Build-in-public posts are a weaker cold start. Paid placement is the visible substitute, which is why the year looks like an auction.

Do not believe that the feeling of a crowded timeline is a measurement of all small software. It is a measurement of what is posted. Quiet products are still how a lot of owners get paid. They will not be the state of the feed. They are part of the state of the work.

The next era, as the homepage already says, starts when people get bored of paying to be seen, and someone finds another way to be seen anyway. Until then, 2026 is cheap software, expensive attention, and a living that was never as common as the screenshots.

FAQ