#90: The Problem With a 1-Person, $1B Company
How do you exit?
Back in April, The New York Times ran a story about how Medvi, a one-person company, was worth over $1B. The exposé detailed how the founder was able to scale the telehealth company so quickly thanks to AI.
Take that valuation with a grain of salt, as reports quickly surfaced following the Times publication alleging that Medvi may have engaged in fraudulent behavior. So maybe we don’t have the first credible one-person billion-dollar company yet.
But the concept is starting to emerge, given the step-function development of AI. Those closest to AI seem to believe this type of company structure is imminent, where one person manages teams of AI agents (in lieu of people). Check out this clip of Sam Altman (CEO of OpenAI) mentioning the near-term possibility of ten-person billion-dollar companies, as well as the glorified one-person billion-dollar company.
The idea is mesmerizing. It turns conventional wisdom about building a company on its head. For most businesses with ambitions of being valued at over $1B, conventional wisdom says you need massive teams, capital, and infrastructure.
I’d love for one of my business ideas to take off and for me to build it by myself all the way to $1B.
But what happens if I want to sell the business and get out? Who would be a buyer of a one-person, AI-native company?
At the billion-dollar level, potential suitors tend to fall into two main buckets: strategic acquirers or financial buyers.
Quick aside: a “strategic” is a company that buys another company for strategic reasons (in its purest form). For instance, Unilever, consumer goods holding company, acquired Gruns in April. Gruns fits in well with Unilever’s portfolio and corporate strategy, thus it was a “strategic acquisition”. On the other hand, a “financial” buyer looks to purchase a company, make operational improvements, and then sell the company. Strategics are typically corporations, and financial buyers are usually private equity firms.
For a strategic buyer, a lean AI-native company is attractive because they don’t have to deal with cultural integration risk. You don’t have to worry about the team you’re acquiring being gun-ho about assimilating into the culture of your company. But you do deal with key-person risk. Sure, AI agents aren’t going to jump to your competitor (or maybe they will if AGI is achieved), but that one person at the ten-person start-up could. Each employee becomes mission critical, so retention packages effectively become embedded into the valuation. Although management retention packages are commonplace in mergers, the stakes are higher if there are only ten people at the target (versus thousands).
Enter, financial buyer. If AI allows companies to operate hyper-efficient workflows and each employee managing teams of AI agents that complete various tasks, how does an outside investor come in and justify a valuation premium if there are limited operational improvements left to be made? Synergies will need to be on the revenue side, as the cost of AI agents will likely be as efficient as possible. What operational efficiencies can private equity drive if there aren’t any more operational efficiencies to create? Read: laying off workers. But wait, those don’t exist!
Financial buyers often rely on margin expansion, and financial engineering to drive returns. Venture-backed technology companies typically rely more heavily on equity financing than debt financing.
Quick tangent: I see both sides of the AI agent efficiency coin on my LinkedIn feed. Some people are bragging about how many thousands of dollars they were charged for Claude API usage the prior month as if this is a signal for AI productivity. I’ve also seen reports that entire companies, including Uber, are blowing through their full-year AI budgets within months. This inherently challenges the thesis that AI agents are cheaper than human employees.
On the flip side, I see others on LinkedIn offering Claude Skills they built to optimize token usage. I imagine a new layer of companies will emerge to help AI agents become as efficient as possible with token usage. In general, as the race to build more data centers continues and the cost of compute decreases, per-token usage costs should decrease. Over time, competition and infrastructure buildout should continue pushing token costs lower, even as total AI usage rises dramatically.
AI-native companies are lean in nature. Highly AI-fluent people are managing teams of AI agents instead of teams of employees to get work done. This means a company’s operating leverage could become extraordinarily high, prior to any potential acquirer coming in. Essentially, AI expands a company’s operating margin to the point where it becomes as lean as realistically possible. If you’re looking to make a purchase, you need a clear revenue-synergy story to justify the acquisition price, as the cost-synergy story is unlikely to be there.
Say you decide to acquire a ten-person AI-native company for $1B. You almost need to do it in all-stock, to keep the operating team incentivized to stay (assuming the target company would even accept the deal). Otherwise, those ten people just became insanely rich, and it’s unlikely they’ll be as motivated as they were pre-acquisition to work at the combined company. Because AI systems are still imperfect, you need someone to stick around and remain accountable for the outputs those agents produce.
Look, if you’re able to build a billion-dollar company by yourself, you are crushing it in life. You’ll eventually figure out how to sell the company. But it does pose a tricky question for an acquirer. Not much meat on the bone in the way of incremental operating efficiencies. Combine that with significant key-person risk, and you’re not looking at a great acquisition story.
If you’ve made it this far, let me know how you think AI will impact valuations and exit potential for the super-lean, AI-native businesses being built today. It may ultimately force potential acquirers to ask themselves: what exactly can we do better? Because the list of answers is likely to be shorter.