Ten-Person Unicorn Landlords
Yohan@yohanlequere
−10.1%
since Oct 8
The idea
I believe the next two years will see billion-dollar companies run by fewer than ten people, leaning entirely on AI and cloud instead of headcount. Rather than guess which tiny team wins, this basket owns what they all have to rent: usage-priced developer platforms, the big cloud providers, and the AI compute layer underneath. It stays concentrated by design, keeps a cash reserve to add on dips, and the idea breaks if AI model makers start selling direct to small builders and squeeze out the middle layer.
What it owns
- Tiny teams pay these companies for what they use, not per employee, so their revenue grows with the work AI does instead of with headcount. · CloudflareIt charges by usage and makes it cheap for one developer to ship an app worldwide, replacing what a sub-ten-person company would otherwise hire an ops team to do., DigitalOceanIt is built for tiny teams and small businesses that pay by the hour for servers and AI computing, growing from small AI-heavy companies rather than flattening as they graduate to bigger clouds., MongoDBIts database is a common default for startups, and its cloud version bills by data and traffic usage, so a tiny team with a big product pays meaningful amounts., DatadogIt charges by how much a company monitors rather than headcount, so a ten-person company running a big service still pays real money as AI apps get busier., SnowflakeIt bills by the compute its customers burn on their data, so lean AI companies crunching large datasets pay for usage instead of staff as consumption grows from AI workloads., TwilioIt is the pipe for automated emails, texts and calls that AI handles in place of an employee's hours, charging per message as AI agents drive more volume through its platform., ElasticIts search engine is what many AI apps use to look up the right information before answering, and its cloud version is priced by usage as builders choose it for that job.42%
- Almost every tiny AI company ends up renting servers and AI models from these giants, which get paid no matter which startup wins. · AmazonAWS is the biggest place small companies rent servers and AI models, providing steady demand as startups choose it and build on its infrastructure., AlphabetIt rents both cloud and Gemini models that tiny AI companies build on, with cloud backlog supporting sustained growth as small builders adopt its services., MicrosoftAzure rents AI models and computing to companies of every size through its OpenAI partnership, with usage growth driving value as tiny teams build on its platform., OracleIt has become a big renter of AI computing capacity and the raw power behind models tiny teams use, with value held as long as that capacity stays full and paid for.27%
- The AI that replaces staff hours runs on chips and GPU clouds that small teams rent by the hour. · NVIDIANearly every AI model a small team rents runs on its chips, so more AI-built companies means more hardware bought by the clouds that serve them., CoreWeaveIt rents GPU computing by usage, the AI horsepower that lets a few people do the work of many, and stays in demand as long as capacity stays paid for., Nebius GroupIt is a smaller GPU cloud that serves AI builders and startups with rent-by-usage computing, staying relevant as long as it keeps signing customers and filling its capacity., Advanced Micro DevicesIt sells the main alternative to NVIDIA chips, helping keep rented AI computing cheaper for small teams as big clouds and AI labs buy its hardware in volume.19%
- Cash12%