Two of the most followed voices in indie software looked at the same market this week and reached opposite conclusions. Pieter Levels says revenue and traffic are falling across indie hackers, including his own projects. Greg Isenberg says more companies are being formed, faster, by fewer people, than at any point in history. One of them must be wrong. Except they are not.
Quick answer: AI is not killing software founders. It is killing one specific setup: cheap tools sold to other builders, with Google SEO as the only growth channel. Stripe Atlas data shows the middle of the indie market shrinking while the top pulls away. As Greg puts it, this is a channel shift wearing a collapse costume.
The debate started with a chart. Jon Yongfook posted that his Google referrals have roughly halved this year. Levels replied that he sees the same decline everywhere and blamed BigAI for eating what used to be apps. Greg quoted him and answered with Stripe numbers. We watch this from both sides at Cipher Projects. Our solo-founder content now gets real referral traffic from AI answers, and the Frontier Club events we sponsor in Vietnam exist because distribution, network, and access to capital now matter more than another weekend app.
What Levelsio Is Actually Saying
Levels is not saying nobody can make money with software. He is describing a squeeze on the classic indie product: a small paid app that people found through search, Product Hunt, or builder Twitter, in a market where competitors used to need weeks of engineering to catch up.
I'm seeing a trend here of declining revenue and traffic with indiehackers
On my own projects too
To me it seems clear BigAI is cannibalizing everything that used to be apps
Not bad btw, just times are changing and we have to adapt
He has been building to this point for months. Customers can now vibecode a private clone of your app and stop paying. AI raises everyone's product quality at the same time, so making your app ten times better no longer means ten times the revenue, because every competitor improved too. The scarce resource is no longer code. It is distribution.
Yongfook's chart is the upstream signal. When Google referrals halve, everything built on top of them weakens. His conclusion was calm and practical: explore YouTube, newsletters, and creators, because the old version of just writing content is losing its power.
What Greg Isenberg Is Actually Saying
Greg is not saying Levels is wrong about traffic. He is saying do not confuse a dying channel with a dying category.
There's panic right now that AI is eating indie software. I see it differently.
For 10+ years you could spin up a few hundred auto-generated pages, rank them, and let Google send you signups for free. Now the answer people used to click your site for sits right inside ChatGPT and Google's AI Overviews.
If your whole business ran on that one faucet, the water's turning off, and it feels like the end of the world.
Then he brings the data. New companies on Stripe Atlas are up sharply year over year. Companies are hitting $10M ARR within three months of launch at double last year's rate. One in five Atlas companies charges its first customer inside the first month, up from 8% in 2020. And solopreneurs, the exact group saying they are getting crushed, are forming the most companies and earning the most when their products are AI-native.
His punchline is the cleanest frame in the whole debate: it is a channel shift wearing a collapse costume. The most exposed business is a cheap tool sold to other indie hackers, with all its traffic coming from SEO. Two cannons pointed at one little boat.
The Stripe Data Both Sides Need
Skip the vibes and go to the primary sources. Greg's formation numbers match Stripe's public reporting. Stripe's solo-founder research adds the part optimists skip: the middle is getting harder even as the top accelerates.
| Signal | Number | Source | What it means |
|---|---|---|---|
| Atlas incorporations, Q1 2026 | Up ~130% YoY, an all-time high | Stripe Atlas / Patrick Collison (May 2026) | More people are starting companies, not fewer |
| First customer within 30 days | 20% of Atlas startups in 2025, up from 8% in 2020 | Stripe 2025 update | Time to revenue is shrinking |
| $10M ARR in 3 months | Double the prior year (2025 cohort) | Stripe 2025 update / TechCrunch | Breakout speed is real, for a thin tip |
| Solo-founded share of Atlas C-corps | 63% in Q2 2026, an all-time high | Stripe: top solo founder traits | Solo founding is now the default |
| AI-native solo revenue by year 2 | Roughly 2x other solo startups | Stripe solo-founder analysis | AI as the product core beats AI as a feature sticker |
| Median solo 6-month revenue (2025) | Down 23% YoY | Stripe solo-founder analysis | This is what Levels is feeling, in numbers |
| Top-decile solo 6-month revenue (2025) | Up 19% YoY | Stripe solo-founder analysis | The winners are winning harder |
| Top-decile vs median gap | 61x in the first 6 months, up from 34x four years ago | Stripe solo-founder analysis | The market is becoming a barbell |
| Top solo founder traits | AI-native, global from day one, B2B, early retention, recurring billing | Stripe solo-founder analysis | A playbook, not a pep talk |
Read only Greg and you miss the median decline. Read only Levels and you miss the formation boom. The honest picture is uncomfortable and useful: more founders, faster companies, and a much wider gap between the middle and the top.
Why Both Can Be Right: The Solo Founder Barbell
Stop thinking of the indie market as one average. Think of it as a barbell.
- The exposed end. Utilities priced at $9 to $49 a month, sold to other builders, found through SEO. Products a customer can rebuild with Cursor in a weekend. These businesses feel the Levels chart first.
- The winning end. AI-native products where the core value sits in models, proprietary workflows, brand, community, or distribution that a customer cannot paste into a private repo. These are Greg's Stripe outliers: global, B2B, recurring revenue, strong early retention.
- The empty middle. A pretty good SaaS with a blog. That used to be a living. In 2026 it is the danger zone: easy to clone, too generic to earn AI citations, too boring to get shared.
We see the same barbell on the ground. Cipher's solo-founder guides still pull readers, and a growing share of that discovery now arrives through AI answers instead of blue links. That does not prove SEO is fine. It proves first-hand, decision-shaped content can still win while the old auto-page faucet cannot.
The Indie Hacker Echo Chamber Is Not the Whole Economy
There is one more thing the doom takes miss, and it needs saying carefully, because it is a structural point, not a criticism of any builder.
The loudest indie voices earned their audiences by building in public. That audience is mostly other people who want to build in public. So the natural product for that audience is a tool, a course, or a template for becoming an indie hacker. The loop feeds itself:
- Build a following by shipping and sharing revenue numbers.
- One success becomes the story: here is how I did it.
- The next product serves the fans: use my tool to do what I did.
- The fans build tools for the same crowd, and sell to each other.
Indie hackers creating indie hackers, selling to indie hackers. The demand is real, but the buyer pool is narrow, self-referential, and full of people who can clone whatever they buy. When AI cut the cost of building to near zero, this niche was always going to feel it first and hardest. A traffic chart from inside that loop tells you about the loop, not about software.
Meanwhile the rest of the economy is barely touched. Niche B2B workflows, ecommerce operations, logistics, clinics, compliance, trades, agencies. These buyers do not read indie Twitter and will never vibecode their own tools. They pay for outcomes, they stay when the product works, and Stripe's data says B2B solo founders out-earn B2C ones by four times at the median. The market is not too crowded. One small corner of it is.
The Two Cannons Aimed at One Boat
Greg's metaphor deserves a table, because most founders are patching the wrong side of the hull.
| Cannon | What changed | Who gets hit first | What to do |
|---|---|---|---|
| Product cloneability | Customers and competitors can vibecode an 80% clone | Cheap builder tools, thin wrappers, generic utilities | Sell outcomes, workflow lock-in, data, and trust that AI cannot paste |
| Traffic replacement | ChatGPT and AI Overviews answer the click that used to reach your blog | Businesses whose only growth lever was informational SEO | Build citation-worthy assets, an owned audience, community, and creator channels |
If both cannons are pointed at you, shipping more features will not save you. You have a distribution and positioning problem dressed up as a product problem.
SEO Is Not Dead. Thin SEO Is.
Cipher has a dog in this fight. We still invest in content for solo founders, and it still pays. What died is the 2018 to 2024 factory: hundreds of thin pages, keyword stuffing, and the hope that Google sends free signups forever.
What still works, for Google and for AI citation:
- Decision content, not definition content. Buyers ask who, which, and should I. AI answers definitions without sending anyone to you.
- First-hand numbers and named sources. Stripe tables, event recaps, real build details. Generic roundups get summarized and skipped.
- Question-shaped structure. Direct answers under each heading, and FAQs written the way people actually ask ChatGPT.
- Authority that is visible in text. Who you are, what you ship, where you operate, and why your answer is earned.
Yongfook is right that just writing content is fading. The answer is not to stop writing. It is to write assets that models want to cite and humans want to share, then back them with channels Google does not own.
What Wins Instead: Distribution, Network, AI-Native Moats
Stripe's top-decile pattern is blunt: build AI-native, sell globally early, prefer B2B, win retention in month one, and use recurring billing. Levels adds the other half of the lesson: audience, creators, and communities compound while search clicks shrink.
This is why Frontier Club matters to us as much as any blog post. The meetups we sponsor in Da Nang and Saigon help founders grow three assets the AI era cannot clone:
- Social following and public proof. Builders who show up, ship in public, and get seen by peers.
- A real founder network. Operators across Vietnam and APAC who trade intros, stack advice, and customer leads.
- Paths to capital. Warm connections to investors when a product actually has pull, not fundraising theater.
Spend an afternoon at Frontier Club Meetup #2 or AgentForge Saigon and the debate stops being abstract. Founders are not failing because software died. They are failing because they brought a 2019 acquisition plan to 2026.
Cipher Projects works on the other side of that gap: fractional CTO and launch-stack work for solo founders who need production systems, SEO and AI-search foundations, and agents on infrastructure they own. Tools are cheap now. Taste, shipping judgment, and distribution are not.
A Practical Playbook for Solo Founders in 2026
If you are staring at a soft month and wondering whether to quit, run this filter before you rewrite your landing page again.
1. Name which cannon is hitting you
Traffic down but conversion holding? Channel problem. Traffic fine but paid conversions dying? Cloneability or wrong buyer. Both? You are the little boat. Admit it early and act.
2. Escape the indie-to-indie $20 trap
Selling cheap software to other indie hackers is the most exposed market on earth. Your buyers can rebuild your product, and they read the same AI answers you do. Stay only if you have an audience or community moat strong enough that clones do not matter. Otherwise point the same product at a boring vertical that pays for outcomes.
3. Become AI-native in the Stripe sense
Not a chatbot bolted on. Core functionality that depends on models, agents, or workflows that improve with use. Pair it with B2B if you can. Stripe's median B2B solo founder earned more than four times the median B2C solo founder by month 24.
4. Rebuild acquisition as a portfolio
Keep SEO, but only for citation-grade assets. Add one owned channel like a newsletter or community, one creator or social loop, and one real-world network. Frontier Club rooms count. So do vertical Slack groups and operator dinners. Cold Google is no longer a strategy. It is one pipe.
5. Put your AI time savings into distribution
If agents cut your build time in half and you spend the saved hours polishing the same features, you are accelerating into a wall. Spend the surplus on outbound, primary-data content, demos, partnerships, and events.
6. Ship the small paid wedge, then harden
Founders are charging sooner than ever. Match that pace. A concierge or AI-assisted version with paying customers beats a perfect agent factory with none. When the wedge works and production hardening is the blocker, bring in a fractional partner instead of hiring a team. See our Founder Launch Stack and the idea filters in Mark Pincus's open-water playbook.
Shareable Takes from This Debate
Copy these if they help the signal travel. Attribution optional.
1. Collapse costume
AI isn't killing indie software.
It's a channel shift wearing a collapse costume.
The businesses dying share one setup: cheap tools for other builders + SEO as the only faucet.
2. Two cannons
Two cannons. One little boat.
1) Your $20 SaaS can be vibecoded in a weekend
2) The Google traffic that fed you now answers inside ChatGPT
If that's your whole model, of course it feels like the end.
3. Barbell
Stripe Atlas solo-founder data in one line:
Median early revenue down 23%.
Top-decile revenue up 19%.
Gap now 61x.
Levels feels the median.
Greg watches the tip.
Same market.
4. Echo chamber
The indie loop:
Build a following by shipping in public.
Sell your fans the tools to do the same.
They sell tools to each other.
Real demand, tiny pond. When building got free, this corner flooded first. The rest of the economy barely noticed.
5. Thin SEO
SEO isn't dead.
Thin SEO is.
Auto-pages ranking for answers ChatGPT now owns? Over.
First-hand, decision-shaped assets that models cite and humans share? Still a weapon.
6. Formation boom
While indie Twitter mourns traffic charts:
Atlas formations up ~130% YoY
63% of new Atlas C-corps are solo-founded
AI-native solos ~2x revenue by year two
Software is a bigger game than ever. The old faucet just closed.
7. Network moat
AI can clone your CRUD app.
It cannot clone your founder network, your public proof, or the room where operators introduce you to customers and capital.
Distribution and community are the new code.
8. Reallocate
If AI halved your build time and you spent the surplus polishing features, you accelerated into the wall.
Put the saved hours into distribution.
Your calendar tells the truth. Your agent factory does not.
FAQ
Is AI killing indie hackers in 2026?
No. AI is killing the low-moat, SEO-only, indie-to-indie micro-SaaS playbook. Company formation and AI-native solo revenue are rising at the same time many classic indie products report declines.
Who is right, Levelsio or Greg Isenberg?
Both, about different slices of the market. Levels describes the exposed median and SEO-dependent products. Greg describes the formation boom and the AI-native top of the barbell. Stripe's data, median down 23% while the top decile is up 19%, is the bridge between them.
Is SEO still worth it for solo founders?
Yes, if you build decision content with first-hand insight, clear answers, and proof that AI engines can cite. No, if your plan is thin keyword pages waiting for free Google signups. Treat SEO as one channel next to an owned audience, creators, and real-world networks.
What does Stripe say solo founders should optimize for?
Build AI-native products, sell globally from day one, prefer B2B, win retention in the first month, and use recurring billing. Top-decile solo founders also earn far more of their revenue internationally than the median.
What should a solo founder do this month if Google traffic is down?
Check whether your product is cloneable and whether your acquisition depends on one channel. Ship one citation-worthy asset, start one owned audience loop, and join one founder network where buyers and operators already gather. If production hardening is blocking you while you chase distribution, use a fractional engineering partner instead of freezing.
How does Cipher Projects fit this shift?
Cipher Projects helps solo founders ship launch systems, content and SEO foundations built for AI search, and production agents on stacks they own. We also help run and sponsor Frontier Club in Vietnam so founders can grow network and distribution, not just code. Details on the solo founders page.