Most people are learning to build websites with AI.
The ones making money learned to sell them.
Watch me find a local business with no site, build them one, put it live, then break down exactly what to charge.
my read: 2025 was about does this protocol work. 2026 is about can a CISO sign off on it. AWS GA answers the first question. the march 9 MCP roadmap is openly aimed at the second. MCP is no longer a bet, it is the assumption.
i think MCP just hit react-scale adoption faster than react did. SDKs are pulling 110 million downloads a month. that took 16 months. react took three years to hit similar volume. and the AWS MCP Server going GA this week is the proof, not the hype.
what it means for indie builders: the month of backend work you were going to do to wrap AWS for an AI feature is now an IAM role and a config line. the moat moves from plumbing to product.
the pattern across every verified win:
1. prior shipping experience or a prior exit
2. distribution before the product
3. high-margin software or subscriptions
4. timelines vary by 100x. 6 months for Base44. 10+ years for Levels.
if you're starting from zero followers and zero reps, the product is not the bottleneck.
I dug into every solo $1M+ AI founder story from 2024 to 2026.
the actual pattern is uglier than the Twitter version.
here's what's verified, what's hype, and what nobody is quoting.
the macro number that's actually defensible: Carta says 36.3% of new US startups in H1 2025 were solo-founded, up from 23.7% in 2019.
Zoom + US Chamber: 29.8M solopreneurs, $1.7T contribution.
the widely-quoted 41.8M / $1.3T figure has thinner sourcing. use the Carta and Zoom numbers.