One thing I want to point out about software is that its perpetually commoditized. Any piece of valuable software will get a free version after some amount of time.
It happened to compilers, operating systems, and renderer. It's gradually happening to game engines.
That's why the largest software companies are really just network effect monopolies. You only use it because other people use it or it relies on some sort of proprietary format.
Many companies like Adobe, Microsoft and Google could sustain a dramatically lower headcount if they started from scratch and never went outside the confines of their network monopoly markets.
From that perspective a one person unicorn is very feasible if we think about it as building a network effect monopoly.
In fact a close example already exists -- Minecraft was very close to being a single person unicorn.
If we have a one person unicorn, I'd imagine it'd be something more like Minecraft than someones newsletter or Google!
> Any piece of valuable software will get a free version after some amount of time.
There are still plenty of areas where this isn't true. Or where the free versions are so far behind it may as well be true.
I work on CAD software that integrates with Catia, NX, and Creo, and the pro CAD software is light years ahead of the free alternatives. And even if the free stuff catches up, there's practically zero chance the biggest users will switch from what they're currently using. CAD lock-in is tricky. A lot of times the different CAD systems target different niches, so it's not just file format lock-in, but that a certain CAD system is better for making planes, or designing for CNC, or whatever. And most companies have automation that integrates with their particular CAD system. And the parts have to be read by newer software for decades and produce identical results, or they won't fit together right.
Some of the vendors release free versions, but they're not open source, and they're only there to steer people towards the paid versions later on.
> From that perspective a one person unicorn is very feasible if we think about it as building a network effect monopoly.
To build a one person unicorn, the main thing you need is an arbitrage opportunity for getting users that is unattractive to multi-person companies. Once established, this business may well turn into a network effect monopoly, but I don't think that is the crux of what is needed to get going.
Is this really true, though? Having a one-person unicorn doesn't preclude having multi-person company competitors. They might have found competitive axes where having multiple actual humans is an advantage.
I kind of wonder if something like this will come along for non-enshittified hardware and software.
I keep seeing most companies "getting on the bandwagon", such as printer companies locking their ink refills, hdtv companies collecting viewer and viewing data, and cellphones getting between you and the device you own in just about every way possible.
There are still counterexamples. Nobody's commoditized Gurobi or k/qdb, and nobody will. But these companies won't ever become household names because they're not driven to grow and build a monopoly at all costs, they're driven to build and sell very specific software to very specific organizations.
The J language actually provides a lot of the niceties of the core Q / kdb functionality. It even has Jd, which is an on-disk columnar database. I don’t know if it counts as being commoditized, as the price in terms of time and effort to learn the language is still quite high.
This, and I'd add that if you're fortunate enough to have the resources to pursue an idea, use that time to your advantage. Don't rush: build it at your own pace. Time can be a real competitive edge compared to startups that are under pressure to sprint toward unicorn status. Network and build.
Building a global network requires not just building a successful product, but exporting that product to different geographic markets & cultures. Core product team might be 1 AI enabled developer, but what about all the other dimensions of a successful global enterprise?
> Any piece of valuable software will get a free version after some amount of time
Technically. The reality is much closer to "Any piece of valuable software will get a crappy free version after some amount of time, and multiple paid copies"
Most free versions of things are not as good as paid.
But it doesn't matter for many areas. Sometimes the free stuff will get better than the common paid one (linux and blender come to mind). Sometimes it takes time (Blender), but if the value is there commoditization will happen
Minecraft? Not at all. Please stop rewriting history. Minecraft definitely had a small dev team but it wasn’t a single persons effort and wasn’t close to being one. You don’t create a world changing video game by yourself.
Minecraft swung past the $10 million in sales mark, and was well on it's way to the $100 million mark, when he hired his first employee. It's not impossible that it would have hit a valuation of $1 billion without hiring on additional employees.
Ilya Sutskever's "business" (really his skillset/personal brand) was worth $1B+ the minute he stepped away from OpenAI.
Joe Rogan could fire all his staff and there would still be companies lining up around the block to buy his podcast for $1B+.
There are many such examples of a person having something or being something that is worth a lot to someone else, to the tune of billions.
So by that metric a one-person billion dollar company has already been achieved. What remains are the technicalities (does it need to be incorporated, does it need to have revenue, does it need to operate as a traditional business).
I'm sure we're already there with musicians as well. If you take Taylor Swift's annual revenue and multiply it by 5-10, you'll easily cross a billion. But she does require tons of staff for the stadium shows that provide the bulk of the revenue, etc. Not literally one person, but it's a company with a bus factor of one.
And I'm willing to bet all that staff is contracted. Her personal employees (people she pays a regular fixed salary to) is probably in the single digits, or possibly even 0.
I'd predict a one person unicorn company potentionally possible, but this will be an inflated valuation company backed by someone, who is willing to prove a point. Not a company that delivers a great product and can be sustainable in the long run.
Purely from the operational perspective, you need to handle support, you need to handle billing, to handle lawsuits, to handle infrastructure, to handle incidents... and still have a good idea of what's going on.
I agree with the article that it won't happen because it's just natural to hire your way out of some pain.
You'd need someone who cares enough to make a point to a very niche community but doesn't care about the product enough to hire, and I think the two circles don't overlap.
... but if it did happen I expect it'd be a NSFW AI chatbot/companion, not VC-backable (or at least not openly)
There are some people building those that don't want or care about building the best product: they've realized if you get the cheapest most heavily quanitzed model around to produce NSFW content and tell a lonely person they're loved, they'll give you increasing amounts of money.
Oh wow, that's simply not true. Many investors have spent more than that to prove a point. Sometimes in one place, sometimes over many places and years.
Cryptocurrency like Bitcoin can probably be made by LLM so if you have a good idea for cryptocurrency or some crypto token and you can vibe code it with LLM, perhaps it can reach a $1bn market cap.
The only purpose of this piece is the gratuitous sideslap at Richardson and her individual political stance. First Law of Modern Politics: nothing is about what it says it is about.
yeah this is some insanely out of touch nonsense.. I have no idea who she is but the idea that any of the current crop of models could replace a copy editor is laughable. Just ask it how many times the letter b appears in blueberry.
The models can replace humans. They can't do it well, but it might be good enough. If you are tiny and just barely making money at all in your job you should evaluate if this cost cutting is good enough or not.
i'm getting the idea though the whoever she is - she is making good amount of money: she should demand a higher bar that I don't think llm's can meet today just because she can afford it and the risk that it matters is too high.
LLMs are good at copy editing. They recognize when words are misspelled or grammar is poor very easily, despite not being able to see individual letters.
For understanding the guy's point it helps if you know anything about Heather Cox Richardson. Merely saying her copyeditors could be replaced by AI is a gross minimization, probably most of her own work could be too.
HCRs newsletter is popular because she delivers much purer, harder Trump Derangement Syndrome than anyone else on the market. That's it, that's the entire secret to her success. She delivers an ordinary and easily LLMable service (summarizing news), but ramps the Trump hate up orders of magnitude. Her competitive edge is she delivers to a market of liberals who find themselves disgusted with the mainstream journalism because, as they see it, it tries much too hard to be neutral and unbiased. It's sort of like how the Guardian makes millions off of donations by being a very openly left wing paper, except 1000x more extreme.
One might think US journalism is not particularly neutral, but they claim they are, so if you reject the whole notion that journalism should be unbiased as idiotic appeasement the market doesn't have much for you. Into this gap steps Heather Cox Richardson, a humanities academic who for a low low price will send you a daily news report that, every day, tells you how evil and bad Trump is and how good a person you are for realizing that.
Go read some of her reports and then tell us an LLM couldn't write them cold, I dare you. It probably wouldn't even need to be a big LLM. Summarization and rewrites are things they're really good at. She never even cites sources and hallucinations aren't the sort of thing her customer base are going to notice.
I would say that there has already been one, Notch and Minecraft. Though he did hire people and step down as dev lead, he was pretty solo and already on the 1 billion dollar trajectory.
Has ConcernedApe ever talked about how much he's made from Stardew Valley before? I've seen estimates from $50M to $300M. Certainly shy of the $1B mark, but, especially if his next project is another solo coded one and does well, he's well on his way.
This seems the closest you are going to get, but Mojang had dozens of employees at the time of the sale. One Reddit post claims there were 40 people there.
Yes, but I think the point the parent comment was making is that the value was created by one person. He didn’t HAVE to hire those people and it still would have eventually made a billion dollars.
This isn’t to say that the other people who worked on Minecraft provided no value, they definitely accelerated the game’s success.
A single founder and a one-person company aren't the same thing. Oculus had about 100 employees when they were acquired; they had about 10 at the time of their Kickstarter campaign.
If one person can make a billion dollar company, it likely means that the product itself isn't too complex or big - unless said person has been stealth developing a product for years. But in this case, let us assume it is something founded recently, and aided with AI because, well...AI. That's the pitch right?
So what moat could such a company have? If you can be one man building a billion dollar company in a very short time, hundreds, probably thousands of competitors will spawn in an instant. And everyone will be racing to the bottom to acquire customers.
Ok, so maybe one of those - the most popular, will receive massive VC funding, so that they can simply outspend their competitors into market dominance. But still, that doesn't change the fundamentals: That the product is likely trivial enough to be solo-coded by a bunch of devs out there.
I think it is mostly AI founders hyping up their AI products. "Now you too can become a billionaire overnight, if you just completely integrate these AI tools into your workflow!"
My take is that as the big AI players advance, and the models become better, we'll get some sort of platform where we can just generate whatever tools we need on the spot. That will, to some extent, kill off the LLM wrapper business. In the end, why pay $10 a month for some subscription if you can get the same product directly from the company that makes the machinery. Right now this isn't the case, as they don't have any such tools - but, I'd guess that in a couple of years that is possible.
> I think it is mostly AI founders hyping up their AI products. "Now you too can become a billionaire overnight, if you just completely integrate these AI tools into your workflow!"
yep, one-person billion dollar company is less than 10x of per-employee valuation of NVDA, OpenAI, PLTR, etc. - a bunch of companies where per-employee valuation has crossed $100M. That is tens of thousands of people, not just one individual.
The $1b per employee large corporation is much easier to imagine though, per the point of the OP - it's hard to imagine someone generating so much revenue that their company is valued at $1b not choosing to hire at least one person to do stuff they haven't got time for or don't like doing.
Funny how people forget this simple fact and are still arguing here if a 1-person unicorn is even possible in their imagination. Well, step out of the trees in the forest...
Readit News
It happened to compilers, operating systems, and renderer. It's gradually happening to game engines.
That's why the largest software companies are really just network effect monopolies. You only use it because other people use it or it relies on some sort of proprietary format.
Many companies like Adobe, Microsoft and Google could sustain a dramatically lower headcount if they started from scratch and never went outside the confines of their network monopoly markets.
From that perspective a one person unicorn is very feasible if we think about it as building a network effect monopoly.
In fact a close example already exists -- Minecraft was very close to being a single person unicorn.
If we have a one person unicorn, I'd imagine it'd be something more like Minecraft than someones newsletter or Google!
There are still plenty of areas where this isn't true. Or where the free versions are so far behind it may as well be true.
I work on CAD software that integrates with Catia, NX, and Creo, and the pro CAD software is light years ahead of the free alternatives. And even if the free stuff catches up, there's practically zero chance the biggest users will switch from what they're currently using. CAD lock-in is tricky. A lot of times the different CAD systems target different niches, so it's not just file format lock-in, but that a certain CAD system is better for making planes, or designing for CNC, or whatever. And most companies have automation that integrates with their particular CAD system. And the parts have to be read by newer software for decades and produce identical results, or they won't fit together right.
Some of the vendors release free versions, but they're not open source, and they're only there to steer people towards the paid versions later on.
To build a one person unicorn, the main thing you need is an arbitrage opportunity for getting users that is unattractive to multi-person companies. Once established, this business may well turn into a network effect monopoly, but I don't think that is the crux of what is needed to get going.
I kind of wonder if something like this will come along for non-enshittified hardware and software.
I keep seeing most companies "getting on the bandwagon", such as printer companies locking their ink refills, hdtv companies collecting viewer and viewing data, and cellphones getting between you and the device you own in just about every way possible.
I can hope.
Deleted Comment
Technically. The reality is much closer to "Any piece of valuable software will get a crappy free version after some amount of time, and multiple paid copies"
But maybe other things matter more to you, that is possible of course.
But it doesn't matter for many areas. Sometimes the free stuff will get better than the common paid one (linux and blender come to mind). Sometimes it takes time (Blender), but if the value is there commoditization will happen
Deleted Comment
[1] Conveniently ignoring the contributions to graphics and music of Simon Foster and Allister Brimble, but still...
Ilya Sutskever's "business" (really his skillset/personal brand) was worth $1B+ the minute he stepped away from OpenAI.
Joe Rogan could fire all his staff and there would still be companies lining up around the block to buy his podcast for $1B+.
There are many such examples of a person having something or being something that is worth a lot to someone else, to the tune of billions.
So by that metric a one-person billion dollar company has already been achieved. What remains are the technicalities (does it need to be incorporated, does it need to have revenue, does it need to operate as a traditional business).
I think the premise is that it is only ever one person.
> Meta "bought" Alex Wang for $14.3B.
> Ilya Sutskever's "business" (really his skillset/personal brand) was worth $1B+ the minute he stepped away from OpenAI.
And that we're not talking about a personal brand, but a real company with real ongoing revenue.
Basically it's a shortcut for the belief that all those other people that founders hire could be replaced with AI.
Purely from the operational perspective, you need to handle support, you need to handle billing, to handle lawsuits, to handle infrastructure, to handle incidents... and still have a good idea of what's going on.
So no, not sure it's happening.
You'd need someone who cares enough to make a point to a very niche community but doesn't care about the product enough to hire, and I think the two circles don't overlap.
... but if it did happen I expect it'd be a NSFW AI chatbot/companion, not VC-backable (or at least not openly)
There are some people building those that don't want or care about building the best product: they've realized if you get the cheapest most heavily quanitzed model around to produce NSFW content and tell a lonely person they're loved, they'll give you increasing amounts of money.
People don't generally spend $1B to prove a point.
They'll spend it on a network of users that is genuinely that valuable though. That's what Zuck did in dropping $1B on Insta.
And it seems entirely possible that if Insta were built today, it could be done with a single person.
Software itself will not make people use it.
I'd be surprised if she felt the same way.
i'm getting the idea though the whoever she is - she is making good amount of money: she should demand a higher bar that I don't think llm's can meet today just because she can afford it and the risk that it matters is too high.
For understanding the guy's point it helps if you know anything about Heather Cox Richardson. Merely saying her copyeditors could be replaced by AI is a gross minimization, probably most of her own work could be too.
HCRs newsletter is popular because she delivers much purer, harder Trump Derangement Syndrome than anyone else on the market. That's it, that's the entire secret to her success. She delivers an ordinary and easily LLMable service (summarizing news), but ramps the Trump hate up orders of magnitude. Her competitive edge is she delivers to a market of liberals who find themselves disgusted with the mainstream journalism because, as they see it, it tries much too hard to be neutral and unbiased. It's sort of like how the Guardian makes millions off of donations by being a very openly left wing paper, except 1000x more extreme.
One might think US journalism is not particularly neutral, but they claim they are, so if you reject the whole notion that journalism should be unbiased as idiotic appeasement the market doesn't have much for you. Into this gap steps Heather Cox Richardson, a humanities academic who for a low low price will send you a daily news report that, every day, tells you how evil and bad Trump is and how good a person you are for realizing that.
Go read some of her reports and then tell us an LLM couldn't write them cold, I dare you. It probably wouldn't even need to be a big LLM. Summarization and rewrites are things they're really good at. She never even cites sources and hallucinations aren't the sort of thing her customer base are going to notice.
https://www.reddit.com/r/Minecraft/comments/18mcpme/does_any...
This isn’t to say that the other people who worked on Minecraft provided no value, they definitely accelerated the game’s success.
Will there be a shovel that uncovers a billion dollar gold nugget? Why sell the shovel then?
If one person can make a billion dollar company, it likely means that the product itself isn't too complex or big - unless said person has been stealth developing a product for years. But in this case, let us assume it is something founded recently, and aided with AI because, well...AI. That's the pitch right?
So what moat could such a company have? If you can be one man building a billion dollar company in a very short time, hundreds, probably thousands of competitors will spawn in an instant. And everyone will be racing to the bottom to acquire customers.
Ok, so maybe one of those - the most popular, will receive massive VC funding, so that they can simply outspend their competitors into market dominance. But still, that doesn't change the fundamentals: That the product is likely trivial enough to be solo-coded by a bunch of devs out there.
I think it is mostly AI founders hyping up their AI products. "Now you too can become a billionaire overnight, if you just completely integrate these AI tools into your workflow!"
My take is that as the big AI players advance, and the models become better, we'll get some sort of platform where we can just generate whatever tools we need on the spot. That will, to some extent, kill off the LLM wrapper business. In the end, why pay $10 a month for some subscription if you can get the same product directly from the company that makes the machinery. Right now this isn't the case, as they don't have any such tools - but, I'd guess that in a couple of years that is possible.
> I think it is mostly AI founders hyping up their AI products. "Now you too can become a billionaire overnight, if you just completely integrate these AI tools into your workflow!"
Yep, and it gives me Ponzi vibes.