Years ago, me and a couple of buddies, we had this crazy idea. We wanted to start an ISP, a clean one, no ads, just pure internet. But our so-called leader, he got greedy. He wanted to raise five million pounds just to print some damn CDs. Guess what? The whole thing went up in smoke. Not long after, some punks came up with the same idea and called it Facebook.
Then, I put my money in this startup, thought it was gold. They were supposed to sell customers to taxis, easy money, right? But those clowns running the show were all talk, no action. I pulled out before it went south. And guess what? Uber and Grab showed up and made a killing.
Another idea, another failure. The problem with these fools was they thought they needed big investors before making a move. But all the real successes? They started in garages, in bedrooms, with nothing but a dream and grit.
Yesterday, I hit up a SaaS company. Told them straight up, they needed to fix their software. Last paragraph? I threw it in their face, “I know it can be done because I built an app to do it myself last week.” Yeah, I’m technical, but before ChatGPT, I couldn’t code a damn thing.
Now, I use AI for everything. Google Ads? I get AI to write those for me. Sure, I gotta sift through some garbage, but with CoPilot, Gemini, ChatGPT, I get what I need in seconds. And then, I use AI to make it happen. I haven’t fired anyone, but the way we do things now? It’s faster, it’s better, and it’s ruthless.
Spot on, compadre.
Ahem… The text above was written by AI, converting my story but told to write like a Mexican drug lord..
I was like “this guy is spot on, he gets it, but he certainly comes from a different part of the tech world than I do. Never heard anybody talk like that there.” Well played!
You got it. The game of VC is easy to “game”. The best entrepreneurs have a bias to action, and that action doesn’t need money now. So there’s self-selection, where the entrepreneurs that VCs will want to fund will be the very entrepreneurs that don’t need the money.
"Alright, picture this: I’m sitting there, right, minding me own business, when this geezer strolls up, all swagger, right, talking big about his "genius" idea. He’s got VCs lined up, angels practically flapping their wings to throw money at him – or so he says. He's banging on about how his idea is worth millions, the kind of talk you hear a lot in the City.
So I lean in, all calm like, and I go, **“Yeah, but what have you *sold*?”**. Blank look. So he starts up again, droning on about investors, valuations, and how it's all *about to* take off. All theoretical stuff, no action.
I cut him off, sharper this time, **“Mate, what have you *sold*?”**. You could see it then – the panic, like a rabbit in headlights. Not a sale to his name. All smoke, no fire.
That’s the thing, innit? Doesn’t matter how much air you’re blowing – till you’ve shifted something, till you've actually *sold* something, all the investors in the world don’t mean squat. Selling – that’s where the game really begins.”
Always enjoy your articles, Tomas. And I think this will be the week I upgrade to paid.
This said, I think the evolution of venture capital is more nuanced than you describe. I think smart venture money is going to shift from pure software businesses to complex vertical integrators that need lots of capital to solve problems in industries where incremental solutions are no longer possible.
I wonder what would be the repercussions of a glut of capital from solopreneur wealth creation as well as lower VC investment in software startups.
Presumably an abundance of capital would incentivize risk taking out of necessity such as funding more uncertain or moonshot projects. I can imagine a period of asset inflation if more capital is chasing a limited number of investment opportunities. Or as Andy suggests above, a shift towards more capital incentive vertical integrators if we are indeed in a new TEP area as Packy suggests.
Regarding the potential erosion of public markets (companies staying private longer or indefinitely), I wonder if syndicated marketplaces will form to allow public investment in private companies.
I agree with your thesis about startups leveraging AI requiring less capital and talent. Still, I don't believe most of the bootstrapped or solopreneur startups you mentioned compete with venture-backed startups, at least not today. Most websites and apps you mentioned are creator tools, courses, AI wrappers or toy apps, not B2B saas, social media, platforms, or marketplaces.
In a traditional VC-backed startup, the venture has to have a billion-dollar+ market opportunity to make sense for investors to cover their losses. Most of these solopreneurs are not focused on scaling a massive startup. I would assume that bootstrappers are more interested in a cashflow, lifestyle businesses or being acquired. Most startups in YC are not competing with these businesses, they are focused on bigger markets and disrupting entire industry verticals with innovative products.
If you look at the entire AI space, what is clear to me so far is that AI has benefited LLM incumbents and chip suppliers more than startups, since big tech accounts for the vast majority of AI returns. In the future, the AI Ubers or AirBNBs of the world will emerge to disrupt the incumbent big tech oligopoly. I believe it will be more VC-backed startups that compete and win versus individual or small bootstrapped teams because everyone will have access to AI to leverage their effort but typically the most talented and well-funded teams will be able to scale.
I didn’t say there were only going to be solopreneurs. I said it’s easier to build, which means everything will take less capital. In some cases, the needed capital will be small enough that these startups won’t need capital altogether. That can be solopreneurs or small teams (remember than Instagram sold for $1B and they were 12 employees, and that was was 12 years ago! WhatsApp was tens of billions with just about 50 employees. Imagine what we’ll be able to do now).
The companies you refer to will require less capital and have more competition, two things that are bad for VCs
You write that startups are easy to self-fund so VCs will diminish. But if attention directed from paid content creators is the key differentiator in a startup flooded world, won't you need huge add budgets? So VCs funding will still be important, but pay for advertising cash burn nstead of server and personnel cash burn?
Apart from this, I hope that VCs will invest more in physical technology instead.
It's true that on one hand there is a feeling of a bubble where it seems that just mentioning AI in your product gets you pre-seed financing. Are most companies seeing real earnings gain when implementing AI today? It reminds me ten years ago, when every company wanted to do Data Science/ML but 90% of corporate projects were failing.
On another hand I totally agree with you that once the first hype passes, it will eventually lead to big changes. It's such a productivity enhancer, and all the solopreneur you mentioned understood this. Having experiencing it myself, the cost of building a SaaS has lowered so much. As a personnal example, before chatGPT it was taking me ages of google/AWS Documentation/stackoverflow to figure how to set up , architecture and build software in the Cloud. Now just talking with chatGPT, and iterating, my learning path suddenly accelerated.
If this trend of solopreneurs taking over materialize, I am curious to see what it means for retail investors. VC will need to shift to more capital intensive sectors but if you are a retail investors investing in S&P 500 ETFs, what happens if all the big tech stocks are beaten by smaller private companies?
I look forward to the premium article of the week!
Those are specifically the thoughts that led me to write this article. The ETF concern is real. The solopreneur path is clear. The value of AI is obvious. Curious what you think of my thoughts from the premium article!
Yes the ETF question is a big question. If a significant portion of the value of large companies is distributed among a large number of smaller companies that are never listed in the S&P 500, the S&P 500 will stop growing at 10% per year and might even experience negative growth. This is quite alarming for someone relying on ETFs to invest in the market. Perhaps this could be offset by ETFs and trackers that, instead of focusing on the top 500 market capitalizations, focus on the top 5000.
Grrr, Tomas, while reading this I've opened half a dozen more tabs/articles to read... not enough hours in the day... wait, never mind, I can get AI to read and summarise them for me... looking forward to the premium article!
The age of small tech businesses is coming but we haven't yet prepared to welcome it, infrastructurally with insufficient data centres and below average internet coverage in most part of the world!
How do you foresee venture capitalists adapting to the rise of solopreneurs and AI-driven startups, where less capital is needed to build and scale businesses? Could this shift lead to a fundamental change in the role of VCs in the tech ecosystem?
Mostly agree. But if things can be done quicker, why wound't a well-funded startup still outcompete individuals if they can do even more with a few people, work around new bottlenecks faster, and use the money on distribution and other things that AI is not as helpful with?
There will be more competitions on simple apps but that's more of a problem whether disintermediation will prevent valuable new companies to be created. But if not, velocity of innovation, talking with customers and iterating, and deploying capital on what matter is still a differentiator, no?
I agree on 20 vs 200. But I’m more thinking on 2 vs 20 (or even better 1 vs 10). A small enough group that is not yet burden by organizational complexity and does not need too much coordination will still benefit by the same proportion from AI. So, 2 people would have a tough time outcompeting 20.
On the other hand, the bar to get seed funding or a Series A would probably be a lot higher: expecting more traction, a better MVP, etc. And given competition from more people building, getting to justify spending large amount of capital would be tough.
But once you are there, deploying more capital and faster ought to still mean a lot.
I think smaller teams means needing less capital. Whether that be counteracted by having more startups getting to good-enough product-market-fit is to be seen.
But there are a lot of 1000+ person companies that now can be disrupted by 20, or 10, or 5 people and VCs would be very happy to provide the capital. So, I’m less worried for VCs not having where to invest but what the value of their current investment would actually end up being.
Yes VCs will demand more MVPs, but MVPs are easier to make. And yes they will want to look for traction but if that happens they might not get to invest anymore (self-funding) so maybe they can’t wait for it.
If 1000+ companies can be disrupted by 20ppl startups, many of these startups will not be funded by VCs because they will be self-funded. Those who do need capital will need less than in the past and will be in a better position to raise it.
You have eaten a bit too much of the hype mushroom Tomas. The generative AI frenzy is already fading. Sure it's useful but it's also incorrect and unreliable. And people are already getting tired of the typically bland and uncreative feel. Receiving an email that was generated feels like you have been cheated. People hate it so much that at some point the market might not be there anymore.
Most developers will also disagree with you on how helpful AI is for coding, for now it's certainly good for junior developers working in their underwear but past that level it's hardly any good. It might change for sure, but we don't know yet. Pieter Levels is part of the 0.1%, this is survivorship bias. All his ideas are built on top of an existing network of thousands of people.
Most of the AI startups are also burning more money than they make buying GPUs and paying electricity bills. At a time where money is so expensive this is not good at all.
So yeah, let's step back a bit and chill a little. Let's be more pragmatic and factual.
I hear you and I believe you are wrong. Time will tell!
I believe AI is certainly going to pervade everything we do, we're just starting.
Creative --> Yes because that's the beginning of AI. It's getting so much better so fast. New creators will be unleashed by it, and a golden age of creation will emerge.
Developers --> I linked to sources showing you are wrong. Most developers love it and think it's highly useful.
Yes Pieter is an outlier, but I shared a list of over a hundred other examples.
The example of money-burning startups you use is more for foundational models and the like. I will address these in the premium article this week.
"Step back a bit and chill a little" does not sound like a sentence structured to optimize debate.
Yes, I actually prefer to step back and chill a little when faced with a new hype. This allows me to be a bit more grounded in reality and actually help with the debate. But it's great we don't have the same approach here.
We already saw the damage of this kind of hype with SBF/Crypto/Dotcom Bubble, and we are already seeing the first AI frauds showing their ugly faces: https://www.youtube.com/watch?v=huu_9rAEiQU
Right now, there are already some people preparing NO AI labels: https://notbyai.fyi/
We are 1 year in the hype, people already hate it, and AI is already a difficult product that requires to be hidden, to ensure there is no backlash on the day of release.
Also, you should know, it's not the beginning of AI. LLMs have progressed tremendously and are impressive, but AI has been around for decades. GPTs are 10-15 years old.
Regarding the coding productivity, literally all the sources you have linked are biased: a Github reasearch, a Jetbrains AI research, etc. The article from iLogos Game Studio state: "productivity increase is typically modest, around 5-10%". Instead, check this article that is not biased and was conducted over 1 year: https://newsletter.pragmaticengineer.com/p/ai-tooling-part-2
I personally use AI every day and they are very useful for learning, and searching, but mostly useless for coding. We already know why: lack of context, lack of short term memory, lack of knowledge in new technologies, outdated code and practices. This will hopefully be addressed in a few months/years.
Right now, they are good at doing boilerplate work that a junior developer can do. Your friends spent 2 months building a tool but what if a software engineer could have done that in 2 days, and much better also? Then your friends just wasted their precious time not doing what they are good at, or going out networking and looking for leads and clients. It's certainly easier to code today, but that doesn't mean everybody should do it, or that it's efficient to do it yourself.
Crypto: different animal. Solutions were unclear. Here they’re everywhere. FWIW I wrote 2 articles on crypto and already about 10 on AI.
All new displacement techs get their Luddites. NO AI is just the latest incarnation.
Looked at the pragmatic engineer article. It confirms exactly what I was saying: very valuable, lots of potential, especially for entry-level, less so for ppl doing the same thing at elite level for 30 years
Every single one of the shortcomings of coding you mention are being solved right this time, as you say. You gotta project to that time.
Notice that “knowing the latest” is not a sign of being one of the best engineers.
It does much more than boilerplate. Everything around ramping up, QAing, debugging, documenting… all that is much more efficient than humans already
Tomas, have you read Cory Doctorov’s thoughts on AI? It’s ok to be excited about it, but Eric has a point. There’s always a lull after the first wave until the true value becomes apparent.
AI is basically good for content creation or research, but the real money is in doing things like reading X-rays. While a 5% error rate is acceptable when you’re trying to plan a vacation, a 5% error rate is terrible for a patient and you’d still need a radiologist around to sign off. It’s the same issue with the self driving cars—what’s the point of the system (beyond a safety backstop) if you need somebody monitoring the system, just have him drive!
One other issue for these solopreneurs and VC alike is the increased luck factor. Think of it this way: a car repair guy can write a business plan, do some demographic research and open his shop on a busy intersection in town and have a pretty good shot at success.
VC investing is internet companies have fewer guarantees, but with a good sense of picking smart motivated founders, enough money for marketing and a large portfolio of these investments, luck will be manageable on a macro level.
The problem for your underwear guy is that he makes most of his money from two apps, and even this is survivorship bias as my fellow commenter pointed out, and the income will not last long there are hundreds of photo apps.
He went viral for a couple of months and because the platforms control the app stores, content feed and so on, they can turn off the luck machine for your guy and sudd not promote the competition. Same thing happens with “creators” or “influencers” on YouTube or TikTok. The algorithm decides to promote more bro content and you get this batch of highly rewarded bros making money off of stoicism for five minutes.
This is a terrible time for investments in general. Either luck plays a huge and unpredictable role or monopolies and oligopolies can crush competition, unrank you, sell ads to your direct competitors in search results and so much more. (In Cory Doctorov’s verbiage, the platforms are quick to identify and take the “surplus”.) Forget selling your company to Google anymore. Same goes for selling a new sports drink in the market if you’re not Nestle or PepsiCo by the way.
I wonder if the VC firms have seen the writing on the wall. I hear that they are heavily invested in Crypto at this point, and have turned their lobbing energies on the US Goverment and the presidential candidates to have the fed hold crypto and increase the value of those investments. It’s sad and a little scary when the vibrant economic sectors like Tech/VC seems to get that the fun is over and that they need to solidify their gains by becoming a rent seeker or buying a ton of real estate instead of reinvesting the money back into the ecosystem.
Accuracy: depends on the use case. There are many many many use cases that just need to be better than 90% of people, and thar market is already huge. Driving is an outlier because the error rate must be 0.00000000.1% per mile or something like that. Radiologists are already being routinely beaten, and it will happen faster and better every year.
Agrees on the mechanisms of risk. The result is that many solopreneurs will make a few million and nothing more. You hear plenty of stories of MRRs of $50k that last 1-2 years. Some will be more, other less, but these numbers are substantial.
Also many will be 1-hit wonders, and others will try 50 apps and succeed at 2-3 (like levels)
As a creator, what you say is avoided by pushing the audience to their newsletter or SMS distribution. Hence why you’re here.
Agreed on this being a bad time for investing.
No VCs are not doing that. Crypto is a very tiny piece. Most are all in on AI
The interesting part about accuracy is that it's not an all-or-nothing proposition. The radiologist can check the result of an AI evaluation much faster than they can do the eval themselves.
Ditto for driving. Watching a self-driving car do its thing while you're doing, well, more-or-less whatever, has a lot of appeal. (Unfortunately. I'd rather use a train – except for several annoying reasons, all of them fixable but neglected. why trains end up being the worse choice.) Even if the only effect is that you arrive relaxed instead of dead tired and with a headache.
Luck always was a factor. (30 years ago I founded a company with two friends. It now has 500+ employees. This number could have ended up at 5 just as easily.) It's just more of one in a volatile situation where anybody can make the Next Great App but only some catch the wave.
Interesting take but I fail to see how a batch of solopreneurs using the same AI tools can achieve a sufficient competitive advantage over one another. Sure, you may have a great idea, but if that idea builds upon publicly available AI you're all on a level playing field.
Great post Tomas - I really enjoyed the post. As a senior college student looking forward to breaking to VC or tech investing several years down the road, how can I leverage AI to start developing skills for VC?
Is it possible to use AI for deal sourcing, product understanding, and due dilligence? If yes, what are some freemium platforms/products that I can look at?
Did AI write this? The scary part is the millions who will lose good paying jobs and be forced into shitty ones . And the millions that have no money for all this stuff will continue to serve the new millionaires.
Years ago, me and a couple of buddies, we had this crazy idea. We wanted to start an ISP, a clean one, no ads, just pure internet. But our so-called leader, he got greedy. He wanted to raise five million pounds just to print some damn CDs. Guess what? The whole thing went up in smoke. Not long after, some punks came up with the same idea and called it Facebook.
Then, I put my money in this startup, thought it was gold. They were supposed to sell customers to taxis, easy money, right? But those clowns running the show were all talk, no action. I pulled out before it went south. And guess what? Uber and Grab showed up and made a killing.
Another idea, another failure. The problem with these fools was they thought they needed big investors before making a move. But all the real successes? They started in garages, in bedrooms, with nothing but a dream and grit.
Yesterday, I hit up a SaaS company. Told them straight up, they needed to fix their software. Last paragraph? I threw it in their face, “I know it can be done because I built an app to do it myself last week.” Yeah, I’m technical, but before ChatGPT, I couldn’t code a damn thing.
Now, I use AI for everything. Google Ads? I get AI to write those for me. Sure, I gotta sift through some garbage, but with CoPilot, Gemini, ChatGPT, I get what I need in seconds. And then, I use AI to make it happen. I haven’t fired anyone, but the way we do things now? It’s faster, it’s better, and it’s ruthless.
Spot on, compadre.
Ahem… The text above was written by AI, converting my story but told to write like a Mexican drug lord..
This is hilarious.
I was like “this guy is spot on, he gets it, but he certainly comes from a different part of the tech world than I do. Never heard anybody talk like that there.” Well played!
You got it. The game of VC is easy to “game”. The best entrepreneurs have a bias to action, and that action doesn’t need money now. So there’s self-selection, where the entrepreneurs that VCs will want to fund will be the very entrepreneurs that don’t need the money.
"Alright, picture this: I’m sitting there, right, minding me own business, when this geezer strolls up, all swagger, right, talking big about his "genius" idea. He’s got VCs lined up, angels practically flapping their wings to throw money at him – or so he says. He's banging on about how his idea is worth millions, the kind of talk you hear a lot in the City.
So I lean in, all calm like, and I go, **“Yeah, but what have you *sold*?”**. Blank look. So he starts up again, droning on about investors, valuations, and how it's all *about to* take off. All theoretical stuff, no action.
I cut him off, sharper this time, **“Mate, what have you *sold*?”**. You could see it then – the panic, like a rabbit in headlights. Not a sale to his name. All smoke, no fire.
That’s the thing, innit? Doesn’t matter how much air you’re blowing – till you’ve shifted something, till you've actually *sold* something, all the investors in the world don’t mean squat. Selling – that’s where the game really begins.”
Ahem… said it before, but different. LOL.
Now I’m in a Guy Ritchie movie
Always enjoy your articles, Tomas. And I think this will be the week I upgrade to paid.
This said, I think the evolution of venture capital is more nuanced than you describe. I think smart venture money is going to shift from pure software businesses to complex vertical integrators that need lots of capital to solve problems in industries where incremental solutions are no longer possible.
See this week's article from Packy: https://www.notboring.co/p/vertical-integrators-part-ii
Thanks for sharing! I noticed the article but didn’t read it. I’ll make sure to read it and incorporate the insights into my article if I missed them.
And welcome!!
I wonder what would be the repercussions of a glut of capital from solopreneur wealth creation as well as lower VC investment in software startups.
Presumably an abundance of capital would incentivize risk taking out of necessity such as funding more uncertain or moonshot projects. I can imagine a period of asset inflation if more capital is chasing a limited number of investment opportunities. Or as Andy suggests above, a shift towards more capital incentive vertical integrators if we are indeed in a new TEP area as Packy suggests.
Regarding the potential erosion of public markets (companies staying private longer or indefinitely), I wonder if syndicated marketplaces will form to allow public investment in private companies.
Looking forward to the next article!
Interesting
The syndicates would be illegal in the U.S. as of now. Gotta change the law
FYI Tomas small typo: OpenAI is worth around $100B not $100M
I rewrote part of the article without giving time to my editors. This is the poor result. I don't learn.
Not sure I would've spotted that one anyway! $100B? Unbelievable!
I agree with your thesis about startups leveraging AI requiring less capital and talent. Still, I don't believe most of the bootstrapped or solopreneur startups you mentioned compete with venture-backed startups, at least not today. Most websites and apps you mentioned are creator tools, courses, AI wrappers or toy apps, not B2B saas, social media, platforms, or marketplaces.
In a traditional VC-backed startup, the venture has to have a billion-dollar+ market opportunity to make sense for investors to cover their losses. Most of these solopreneurs are not focused on scaling a massive startup. I would assume that bootstrappers are more interested in a cashflow, lifestyle businesses or being acquired. Most startups in YC are not competing with these businesses, they are focused on bigger markets and disrupting entire industry verticals with innovative products.
If you look at the entire AI space, what is clear to me so far is that AI has benefited LLM incumbents and chip suppliers more than startups, since big tech accounts for the vast majority of AI returns. In the future, the AI Ubers or AirBNBs of the world will emerge to disrupt the incumbent big tech oligopoly. I believe it will be more VC-backed startups that compete and win versus individual or small bootstrapped teams because everyone will have access to AI to leverage their effort but typically the most talented and well-funded teams will be able to scale.
I think we agree.
I didn’t say there were only going to be solopreneurs. I said it’s easier to build, which means everything will take less capital. In some cases, the needed capital will be small enough that these startups won’t need capital altogether. That can be solopreneurs or small teams (remember than Instagram sold for $1B and they were 12 employees, and that was was 12 years ago! WhatsApp was tens of billions with just about 50 employees. Imagine what we’ll be able to do now).
The companies you refer to will require less capital and have more competition, two things that are bad for VCs
You write that startups are easy to self-fund so VCs will diminish. But if attention directed from paid content creators is the key differentiator in a startup flooded world, won't you need huge add budgets? So VCs funding will still be important, but pay for advertising cash burn nstead of server and personnel cash burn?
Apart from this, I hope that VCs will invest more in physical technology instead.
Clearly the article is not clear enough. Sorry for that.
I did say that startups will still need capital for distribution, to access attention (the scarce resource).
What I also added is that creators will be the recipient of that money more and more, in a parallel to solopreneurs.
Great article!
It's true that on one hand there is a feeling of a bubble where it seems that just mentioning AI in your product gets you pre-seed financing. Are most companies seeing real earnings gain when implementing AI today? It reminds me ten years ago, when every company wanted to do Data Science/ML but 90% of corporate projects were failing.
On another hand I totally agree with you that once the first hype passes, it will eventually lead to big changes. It's such a productivity enhancer, and all the solopreneur you mentioned understood this. Having experiencing it myself, the cost of building a SaaS has lowered so much. As a personnal example, before chatGPT it was taking me ages of google/AWS Documentation/stackoverflow to figure how to set up , architecture and build software in the Cloud. Now just talking with chatGPT, and iterating, my learning path suddenly accelerated.
If this trend of solopreneurs taking over materialize, I am curious to see what it means for retail investors. VC will need to shift to more capital intensive sectors but if you are a retail investors investing in S&P 500 ETFs, what happens if all the big tech stocks are beaten by smaller private companies?
I look forward to the premium article of the week!
Those are specifically the thoughts that led me to write this article. The ETF concern is real. The solopreneur path is clear. The value of AI is obvious. Curious what you think of my thoughts from the premium article!
Yes the ETF question is a big question. If a significant portion of the value of large companies is distributed among a large number of smaller companies that are never listed in the S&P 500, the S&P 500 will stop growing at 10% per year and might even experience negative growth. This is quite alarming for someone relying on ETFs to invest in the market. Perhaps this could be offset by ETFs and trackers that, instead of focusing on the top 500 market capitalizations, focus on the top 5000.
Open question to me:
How do VCs re-focus their efforts on the businesses that do need capital more readily?
As mentioned, there is a shrinking demand for external venture capital.
I think the trend right now is to chase AI wrapper startups and move away from capital-intensive deep tech businesses like robotics.
This seems to be at odds with logic until we remember interest rates are rising and VC purse strings are much tighter.
So, self-preservation is winning over chasing opportunity?
(Could be off but these are first thoughts)
Addressing in the premium article this week!
Grrr, Tomas, while reading this I've opened half a dozen more tabs/articles to read... not enough hours in the day... wait, never mind, I can get AI to read and summarise them for me... looking forward to the premium article!
Hahaha I write them so you don’t need to check all the sources, but yes, AI is welcome for this type of thing!
The age of small tech businesses is coming but we haven't yet prepared to welcome it, infrastructurally with insufficient data centres and below average internet coverage in most part of the world!
How do you foresee venture capitalists adapting to the rise of solopreneurs and AI-driven startups, where less capital is needed to build and scale businesses? Could this shift lead to a fundamental change in the role of VCs in the tech ecosystem?
Premium article tomorrow! The short is they can.
Mostly agree. But if things can be done quicker, why wound't a well-funded startup still outcompete individuals if they can do even more with a few people, work around new bottlenecks faster, and use the money on distribution and other things that AI is not as helpful with?
There will be more competitions on simple apps but that's more of a problem whether disintermediation will prevent valuable new companies to be created. But if not, velocity of innovation, talking with customers and iterating, and deploying capital on what matter is still a differentiator, no?
Because 20 outstanding individuals can now out-execute 200 average ones.
And 20 outstanding individuals are much easier to self-fund than 200.
Put in another way: capital still matters, but it matters *less*, and that’s dangerous for VCs
Wdyt?
I agree on 20 vs 200. But I’m more thinking on 2 vs 20 (or even better 1 vs 10). A small enough group that is not yet burden by organizational complexity and does not need too much coordination will still benefit by the same proportion from AI. So, 2 people would have a tough time outcompeting 20.
On the other hand, the bar to get seed funding or a Series A would probably be a lot higher: expecting more traction, a better MVP, etc. And given competition from more people building, getting to justify spending large amount of capital would be tough.
But once you are there, deploying more capital and faster ought to still mean a lot.
I think smaller teams means needing less capital. Whether that be counteracted by having more startups getting to good-enough product-market-fit is to be seen.
But there are a lot of 1000+ person companies that now can be disrupted by 20, or 10, or 5 people and VCs would be very happy to provide the capital. So, I’m less worried for VCs not having where to invest but what the value of their current investment would actually end up being.
I agree with a lot of what you say.
Yes VCs will demand more MVPs, but MVPs are easier to make. And yes they will want to look for traction but if that happens they might not get to invest anymore (self-funding) so maybe they can’t wait for it.
If 1000+ companies can be disrupted by 20ppl startups, many of these startups will not be funded by VCs because they will be self-funded. Those who do need capital will need less than in the past and will be in a better position to raise it.
You have eaten a bit too much of the hype mushroom Tomas. The generative AI frenzy is already fading. Sure it's useful but it's also incorrect and unreliable. And people are already getting tired of the typically bland and uncreative feel. Receiving an email that was generated feels like you have been cheated. People hate it so much that at some point the market might not be there anymore.
Most developers will also disagree with you on how helpful AI is for coding, for now it's certainly good for junior developers working in their underwear but past that level it's hardly any good. It might change for sure, but we don't know yet. Pieter Levels is part of the 0.1%, this is survivorship bias. All his ideas are built on top of an existing network of thousands of people.
Most of the AI startups are also burning more money than they make buying GPUs and paying electricity bills. At a time where money is so expensive this is not good at all.
So yeah, let's step back a bit and chill a little. Let's be more pragmatic and factual.
I hear you and I believe you are wrong. Time will tell!
I believe AI is certainly going to pervade everything we do, we're just starting.
Creative --> Yes because that's the beginning of AI. It's getting so much better so fast. New creators will be unleashed by it, and a golden age of creation will emerge.
Developers --> I linked to sources showing you are wrong. Most developers love it and think it's highly useful.
Yes Pieter is an outlier, but I shared a list of over a hundred other examples.
The example of money-burning startups you use is more for foundational models and the like. I will address these in the premium article this week.
"Step back a bit and chill a little" does not sound like a sentence structured to optimize debate.
Yes, I actually prefer to step back and chill a little when faced with a new hype. This allows me to be a bit more grounded in reality and actually help with the debate. But it's great we don't have the same approach here.
We already saw the damage of this kind of hype with SBF/Crypto/Dotcom Bubble, and we are already seeing the first AI frauds showing their ugly faces: https://www.youtube.com/watch?v=huu_9rAEiQU
Right now, there are already some people preparing NO AI labels: https://notbyai.fyi/
We are 1 year in the hype, people already hate it, and AI is already a difficult product that requires to be hidden, to ensure there is no backlash on the day of release.
Also, you should know, it's not the beginning of AI. LLMs have progressed tremendously and are impressive, but AI has been around for decades. GPTs are 10-15 years old.
Regarding the coding productivity, literally all the sources you have linked are biased: a Github reasearch, a Jetbrains AI research, etc. The article from iLogos Game Studio state: "productivity increase is typically modest, around 5-10%". Instead, check this article that is not biased and was conducted over 1 year: https://newsletter.pragmaticengineer.com/p/ai-tooling-part-2
I personally use AI every day and they are very useful for learning, and searching, but mostly useless for coding. We already know why: lack of context, lack of short term memory, lack of knowledge in new technologies, outdated code and practices. This will hopefully be addressed in a few months/years.
Right now, they are good at doing boilerplate work that a junior developer can do. Your friends spent 2 months building a tool but what if a software engineer could have done that in 2 days, and much better also? Then your friends just wasted their precious time not doing what they are good at, or going out networking and looking for leads and clients. It's certainly easier to code today, but that doesn't mean everybody should do it, or that it's efficient to do it yourself.
I thought I had responded!
Crypto: different animal. Solutions were unclear. Here they’re everywhere. FWIW I wrote 2 articles on crypto and already about 10 on AI.
All new displacement techs get their Luddites. NO AI is just the latest incarnation.
Looked at the pragmatic engineer article. It confirms exactly what I was saying: very valuable, lots of potential, especially for entry-level, less so for ppl doing the same thing at elite level for 30 years
Every single one of the shortcomings of coding you mention are being solved right this time, as you say. You gotta project to that time.
Notice that “knowing the latest” is not a sign of being one of the best engineers.
It does much more than boilerplate. Everything around ramping up, QAing, debugging, documenting… all that is much more efficient than humans already
Tomas, have you read Cory Doctorov’s thoughts on AI? It’s ok to be excited about it, but Eric has a point. There’s always a lull after the first wave until the true value becomes apparent.
AI is basically good for content creation or research, but the real money is in doing things like reading X-rays. While a 5% error rate is acceptable when you’re trying to plan a vacation, a 5% error rate is terrible for a patient and you’d still need a radiologist around to sign off. It’s the same issue with the self driving cars—what’s the point of the system (beyond a safety backstop) if you need somebody monitoring the system, just have him drive!
One other issue for these solopreneurs and VC alike is the increased luck factor. Think of it this way: a car repair guy can write a business plan, do some demographic research and open his shop on a busy intersection in town and have a pretty good shot at success.
VC investing is internet companies have fewer guarantees, but with a good sense of picking smart motivated founders, enough money for marketing and a large portfolio of these investments, luck will be manageable on a macro level.
The problem for your underwear guy is that he makes most of his money from two apps, and even this is survivorship bias as my fellow commenter pointed out, and the income will not last long there are hundreds of photo apps.
He went viral for a couple of months and because the platforms control the app stores, content feed and so on, they can turn off the luck machine for your guy and sudd not promote the competition. Same thing happens with “creators” or “influencers” on YouTube or TikTok. The algorithm decides to promote more bro content and you get this batch of highly rewarded bros making money off of stoicism for five minutes.
This is a terrible time for investments in general. Either luck plays a huge and unpredictable role or monopolies and oligopolies can crush competition, unrank you, sell ads to your direct competitors in search results and so much more. (In Cory Doctorov’s verbiage, the platforms are quick to identify and take the “surplus”.) Forget selling your company to Google anymore. Same goes for selling a new sports drink in the market if you’re not Nestle or PepsiCo by the way.
I wonder if the VC firms have seen the writing on the wall. I hear that they are heavily invested in Crypto at this point, and have turned their lobbing energies on the US Goverment and the presidential candidates to have the fed hold crypto and increase the value of those investments. It’s sad and a little scary when the vibrant economic sectors like Tech/VC seems to get that the fun is over and that they need to solidify their gains by becoming a rent seeker or buying a ton of real estate instead of reinvesting the money back into the ecosystem.
Thanks! Reacting pt by pt
Accuracy: depends on the use case. There are many many many use cases that just need to be better than 90% of people, and thar market is already huge. Driving is an outlier because the error rate must be 0.00000000.1% per mile or something like that. Radiologists are already being routinely beaten, and it will happen faster and better every year.
Agrees on the mechanisms of risk. The result is that many solopreneurs will make a few million and nothing more. You hear plenty of stories of MRRs of $50k that last 1-2 years. Some will be more, other less, but these numbers are substantial.
Also many will be 1-hit wonders, and others will try 50 apps and succeed at 2-3 (like levels)
As a creator, what you say is avoided by pushing the audience to their newsletter or SMS distribution. Hence why you’re here.
Agreed on this being a bad time for investing.
No VCs are not doing that. Crypto is a very tiny piece. Most are all in on AI
The interesting part about accuracy is that it's not an all-or-nothing proposition. The radiologist can check the result of an AI evaluation much faster than they can do the eval themselves.
Ditto for driving. Watching a self-driving car do its thing while you're doing, well, more-or-less whatever, has a lot of appeal. (Unfortunately. I'd rather use a train – except for several annoying reasons, all of them fixable but neglected. why trains end up being the worse choice.) Even if the only effect is that you arrive relaxed instead of dead tired and with a headache.
Luck always was a factor. (30 years ago I founded a company with two friends. It now has 500+ employees. This number could have ended up at 5 just as easily.) It's just more of one in a volatile situation where anybody can make the Next Great App but only some catch the wave.
Interesting take but I fail to see how a batch of solopreneurs using the same AI tools can achieve a sufficient competitive advantage over one another. Sure, you may have a great idea, but if that idea builds upon publicly available AI you're all on a level playing field.
Exactly. Plenty of competition! Good for customers, bad for builders that aren’t nimble.
Great post Tomas - I really enjoyed the post. As a senior college student looking forward to breaking to VC or tech investing several years down the road, how can I leverage AI to start developing skills for VC?
Is it possible to use AI for deal sourcing, product understanding, and due dilligence? If yes, what are some freemium platforms/products that I can look at?
Thanks a lot Tomas!
Did AI write this? The scary part is the millions who will lose good paying jobs and be forced into shitty ones . And the millions that have no money for all this stuff will continue to serve the new millionaires.
Or they can get on it, learn AI, and produce instead of crying
That’s a very simplistic and ignorant statement.