March 26, 2026
In a 2024 chat with Alexis Ohanian, Sam Altman predicted that we’ll soon see the first one-person billion-dollar company.
It seemed far-fetched but by last summer the idea was everywhere. Y Combinator’s Request for Startups featured an item for the first “10-person, $100B Company”. It doesn’t specify a category or industry so it reads more like the “make no mistakes” meme than an investment thesis.
Sam was talking about a company with a $1b valuation. That isn’t controversial because valuations are completely disconnected from revenue, from having a product, or even having a plan for a product.
How about a “real” $1b business, something like $100m in ARR? Could that be run by one person? Somebody already beat me to this analysis and the short answer is we’re not even close.
In February, Lenny Rachitsky interviewed OpenAI’s head of platform engineering. They covered all the things AI agents are enabling. Lenny enthusiastically endorses everything (“amazing”) but when the conversation turns to the $1b 1-person business, he is immediately skeptical. As an entrepreneur, he can’t imagine keeping up with the number of support tickets his small business requires, let alone running the entire thing solo.
Anyone who has run a business before will have the same reaction as Lenny. I don’t believe we’ll ever see a $100m ARR one person operation regardless of how much AI automation improves. At some point it would be dumb not to hire some help with all that money. (The counter argument is that AI will be so good you’d prefer more AI to making that hire. I don’t buy it.)
But it’s still worth thinking about: how big of a product could you build and operate alone?
First, let’s define some terms: I’m specifically talking about SaaS/software businesses, not content. For example, Nathan Barry could sell his “App Design Handbook” solo but Kit could never be a $40m+ ARR company with one person.
And “big” is defined along two distinct dimensions: ARR and number of users supported.
Single digit million solo software businesses definitely exist. Can you go higher?
The Web 2.0 benchmark is PlentyOfFish. Markus Frind scaled to $10M/year solo before hiring a second employee. PlentyOfFish was eventually acquired by Match Group for $575m.
What about the number of active users? I know you can support 140k users solo because I’ve done it. How about more? The kind of business you’re running definitely affects this. It’s easier to support 100k users on a fantasy football platform than 100k customers of vertical SaaS for hair salons. The largest solo-supported number of users I could find was vibe coding platform Base44, which had 400,000 users at the time of its acquisition. The founder didn’t make a hire until just before the sale.
These numbers should be eclipsed soon. We already know AI can automate coding. But that’s just one small part of building a software business.
Many customer support tickets are repetitive and can be resolved with a playbook so it should be possible to automate. But it’s unclear whether any company has done it successfully at scale. Like everything in the AI hype cycle, it’s hard to know what’s real and what’s just a good headline.
Another interesting area where AI can help is incident response. Today, when production breaks, a groggy on-call engineer scans a few dashboards and application logs. They move down the list of possible culprits (is a deploy in progress, what was the last commit, was a configuration change made, is the database under heavy load, etc). AI can check all of those things simultaneously and then tell you where to look first. Many engineers are already doing this manually (e.g., by connecting an LLM to DataDog via MCP). There will be a product that makes it easier to do this.
Maybe $100m is not realistic but your leverage is definitely growing. I’m excited to see how far it can go.