But, won't most anything startups be out of business in 3 years?
Data from the BLS shows that:
"approximately 20% of new businesses fail during the first two years of being open, 45% during the first five years, and 65% during the first 10 years. Only 25% of new businesses make it to 15 years or more."
It also depends on how one defines "fail." A majority of startup I know* are "living dead." This means:
- There are 1-20 employees
- Big-O, investors have seen no returns
- Big-O, they're cash-flow neutral and growth/decline neutral
A VC defines that as a "fail." Founders are often very happy running a small business in their domain of passion. It's a lot more fun than a big business.
* Yes, that's a very strong sample bias. I don't mean to imply a statistical sample, and explaining the types of startups I typically interact with would be an off-topic essay.
Do they? Most businesses in the real world have to generate cash or they fold. Venture capital can keep small startups going far beyond their useful life.
But how many startups receive any appreciable amount of VC money? How many are operating on the savings of the founder(s) and/or money from family and friends rounds? How many founders are going into personal debt to fund their startup?
And how many of them that do raise funds go crazy on a hiring spree once they get any remotely reasonable funding, while not having a functional product or a remotely sustainable user base?
According to Crunchbase only 1 in 3 startups even make it to Series A (between 2011 and 2018, the % was fairly consistent each year)
Looking at "funding data from around 15,600 U.S.-based technology companies founded between 2003 and 2013" TechCrunch comes to the conclusion that only about 40% that close a Pre-Series A round make it to a Series A.
Most businesses can get loans, including unviable ones, by either
1. Burning existing equity: Owner uses own house as security to get loan, common story
2. Political pressure to get loans: This is how zombie companies are born, and they are very common outside of anglo saxon countries.
I would not say startups are subject to more or less business discipline compared to normal companies of the same caliber. Your average tech startup founder has a lot more resources and credentials to burn in emergencies compared to an immigrant starting up a restaurant, so it has to be a like for like comparison.
There's an important difference here in that AI company is basically a new market. If 50% of Mexican restaurants or low-cost database companies that start this year go out of business by 2026 it doesn't matter much because there's a pre-existing mass of them. If 85% of AI startups go out of business by 2026 that's basically 85% of ALL AI companies.
If, last year, we had seen a headline that "85% of all Blockchain Startups Will Be Out of Business in 3 Years" I don't think anyone would be shocked at that prediction.
A startup is a specific category within the broader framework of "new businesses". Based on my knowledge, that about 90% of startups do not survive beyond their initial 5 years. If a startup cannot find repeatable and scalable business model in 5 years then it should close: it is just waste of time.
Exactly.
The article seems to focus on AI startups making their own algorithms and platforms (OpenAI, Meta, Google) but doesn't mention startups who are using these platforms for niche use cases. This is similar to saying a hosting company will fail because AWS, Azure and GCP will always have more resources.
This. It would be great to have an AI product to catch this reasoning-baits.
Offtopic: I understand this is not a fallacy but how is it call in English when someone applies a logic that is truth for the universe but only to a subset? It is not a tautology.
No, it is not cherry picking because with cherry picking the implication is false. It is like because A is true and A is included in B, B is true. In this case it is B is true, A is included in B so A is also true but you only name A instead of the bigger set B.
Most AI startups, whether they know it or not are following a methodology that I call:
Artificial Intelligence Machine Learning Entrepreneurial Startup System (A.I.M.L.E.S.S).
In this methodology, all startups create the same basic four products:
- 1: an AI logo generator
- 2: an AI avatar/profile generator
- 3: an AI pretty picture maker
- 4: an AI thing that lets you be a lazy writer
The first to market makes $100,000 in a month then joins the rest in making nothing, whilst loudly complaining that everyone copied their idea.
Then, just like Web 1.0, the founders throw up their hands and go bust, declaring that “there’s no way to make money from AI”, and shut down their companies, only to be startled when five years later someone comes up with the AI equivalent of Facebook and everyone kicks themselves for not seeing the opportunity earlier.
Also "AI" working on "chatbots". I have a friend who works as a recruiter at a place that helps companies sell their trash products via whatsapp (what a disgusting invasion of capitalist greed into formerly personal communication spaces) that pivoted to doing it via LLMs. It's run by two grifters who couldn't tell the difference between a LLM and "ChatGPT", it's all the same for them. yet the roles they are hiring for are called "AI engineer" and such. When it was founded, the company was a clothing brand, before pivoting to ecommerce. Can't wait for that useless piece of shit "startup" to go out of business (after my friend finds a better job).
This sounds very optimistic and probably reflects the VC view more than reality. If 15% of current AI startups are succeeding 3 years from now (granted not the same as "in business", but meaning roughly they survived to another funding round) investors will have done very well. I'd invest in a portfolio I thought had that kind of upside. Personally I think it will be way lower.
How many web3 startups will still exist next year?
There'll be something else along. Perhaps metaverses will come back into fashion, or, er, data lakes or something? At any given time there is usually one thing that VCs are weirdly excited about; occasionally it pans out. Much as about 5-10% of VC-backed companies are big successes, 5-10% of VC-backed fads turn into something real.
There’s a real potential market in AI driven support bots, but time will tell if they catch on or are just fads. “Things people have to do but don’t want to do” is an area to keep an eye on.
stackai, botpress, elevenlabs, midjourney, the title didn't say llm it said AI. This would also include vector db companies and ancillary services that only exist to add value to llm companies, that would probably fail or be way lower valued if there weren't an llm use case.
The word business is doing a lot in this sentence. Crypto finally has been exposed for what it was all along and now the next hot thing is AI.
AI is real though… but everyone who thinks forking over massive amounts of money to OAI is good business is going to be sorely disappointed when prices increase and/or apis go away all together.
You need to be building something of your own based on the tech. The hallmark of crypto bro turned ai influencer is doing none of the work themselves… just like they let everyone else mine the crypto and then basically just stole it.
I think it'll depend on company / product, there is awful crypto influencers and token shills that will jump on to the next hype train but I doubt you'd call Ethereum as a useless product.
There will be tangible real use cases for Gen AI as there are tangible use cases for blockchains and crypto and there will also be fakes and peddlers who want to sell you bs.
I think the meaning behind OPs words is that you find some type of startup who tend to pivot & circle around "hype tech", rarely managing to produce any product or service of value before pivoting to next hype bubble.
When the peak of the hype cycle starts to fall (and there are signs of that now) it’s going to be a very hard road for the vast majority of these startups. 85% is probably optimistic.
Few have done anything that has defensible IP, there are no moats in this business (if someone builds a better model you just move to that), most startups have no viable business model given the bad unit economics of running generative AI training and inference, the regulatory environment is also ramping up and its looking ugly. Once one looks past the initial wow factor there’s just not much there there from a business standpoint. All that combined with the fact that many VCs are just starting to lick their wounds from some terrible investment decisions of the 2019-22 era and there won’t be much of a net to catch those that stumble.
I find the tech really interesting, but this is all looking quite terrible from a business sense.
Considering that 90% of startups fail, I'd say those figures are even too optimistic. It's more like:
95% will fail,
4% will become stable projects that throw out enough cash to support their developer (plus a team of 2 - 4),
0.99% will become venture-scale companies, and
0.01% will grow to Google/Microsoft/Apple-scale.
> 0.99% will become venture-scale companies, and 0.01% will grow to Google/Microsoft/Apple-scale
The only relevant question for VCs is whether or not that top 1% has already launched or not. If they have they're too late. If they haven't then continued investing in AI makes sense from their perspective. Given the feeding frenzy it looks as though the general consensus is that that 1% hasn't launched yet (nor an appreciable fraction of it).
In other words, VC analysts are expecting one or more key breakthroughs are still in the pipeline, and given how broad the investments are they haven't a clue as to where those breakthroughs will happen. I'm privy to some of these conversations (but in a European context) and even though I think that in principle they are right I'm not necessarily bullish on the idea that the larger part of that 1% is still to come. What GPT-5 will do when it is released will determine to what degree free money will still be thrown at AI start-ups, if it underwhelms I expect this source of funding to be reduced fairly quickly. It's the .com boom all over again: everybody wants to get on the train but nobody has any idea where the train is going to.
But in the very long term (beyond the horizon defined by the typical VC cycle) the impact will be massive.
0.01% is very optimistic too I think. It could very well be 0%. The reason Google /Microsoft/Apple got to that scale is because incumbent corporations like IBM let them do so, totally missing the bandwagon on the Internet, personal computing and mobile. I seriously doubt Google/Microsoft/Apple will let small startups get to their scale without a fight (or aggressive buyouts.)
Big corporations don't miss disruptive trends. They crush any internal efforts to capture them out of fear of compromising the main profit lines and end up missing out. IBM didn't miss the bandwagon on mobile, Internet, or personal computing.
The IBM PC was what everyone outside a few niche companies copied. IBM Simon beat the iPhone by over a decade and probably could have been significant with a few iterations. Almost every modem used to connect to the early commercial internet slotted into an IBM clone. They built the wagon, but every time they had to make a choice on how to proceed, they chose poorly.
Fair point. But don't you think that companies today have learned from the mistakes of the past? Also they have way more cash for aggressively buying out disruptors.
Google: Invented the transformer, but didn't do much with it. Google Assistant probably uses this tech and has been around, but it was ChatGPT that took the world by storm. Bard was useless at launch, and doesn't seem much better today. Google's recent history is littered with heralded launches and unceremonious closures shortly after. Time will tell if they can turn it around.
Microsoft spent a long time resting on its laurels before starting to innovate again, but then they fell back into their old ways. Every smart move is hampered by the continuing need to subsume innovation to serve a desktop OS monopoly with declining relevance. LLMs are headed away from the silos that birthed the revolution toward on-board purpose-built LLMs that are entirely within the user's control, so Copilot's days are numbered. Even integrated GPUs will have enough power and memory to run the better LLMs in 5-10 years.
Apple: back to selling overpriced PCs. Bringing SOCs in from mobile to the desktop is an interesting move, but Microsoft and Windows PCs are likely to beat them on price and features once that world catches up. They're in for a period of mistaking profit growth for innovation the way all big companies do. The pivot to services will be fraught, hindered by corporate inertia, and possibly kill the company. They've had on-board ML cores in all their mobile devices for years, and the best they could do was Siri.
My prediction is that this will be like the .com boom. 99% of AI startups will fail. A handful will succeed.
Google Series A was $25M at a $75M valuation (source: random web site. Doesn't need to be correct for the point). 1/3 of Google is worth $500B today. Add in dilution, and it big-O perhaps a 1000x ROI.
For every Google or Amazon, there's probably nearly a hundred pets.com.
I suspect AI might be similar. It's hard to predict which is going to be which. A portfolio doesn't seem crazy to me.
The space just seems too bubbly to even try anything entrepreneurial yet. Slapping a UI around someone else’s API feels like snake oil or at best it would have no competitive moat. And how would you ever get through the noise right now?
AI is interesting but my approach is to look properly at commercial opportunities and real applications when the hype has died down and when the tech settles.
Seems pretty standard, many businesses don't make it 3 years. The figures for ALL businesses [0] are not too drastic but that includes simple, low risk, cash positive ones. Start ups are worse of course...
Wait, this means AI startups actually will fare better than other startups then? Seems like the inverse of what the title is trying to convey which is pretty funny if true.
No way it's that low. <10% of startups succeed in a nurturing environment with cheap money. The current env is decidedly average, not cheap. I would suspect <5% will survive, and 0.1-0.5% will give a "good" return if they sell or go public.
This is a very poorly written article with a clickbait title.
The word "fail" appears only once in the whole text, which is in the title description itself.
This is what the article actually states:
> Smythe expects, however, 85% of AI startups to be out of business in three years, either because they were swallowed up by big companies or simply because they ran out of cash.
Running out of cash can be classified as failure, but being "swallowed up by big companies" is often the end game for startups. I would not describe a buyout as a failure. Did Mojang failed as a company when Microsoft handed over 2.5 billion dollars for it?
We've reached the part of the hype cycle where VCs realize LLMs are yet more aimless technology with very narrow product potential; so they start tricking an even less informed investor class into fund matching.
A very fundamental question in the AI/ML business is how we are going to avoid monopolies, as so much of the capital (and thus computing power) gets centralized to a few companies. The fact that so much of this work is built on open research makes it even more questionable.
You can only get that far with theoretical research - when it comes down to it, money / funding is what actually realizes the product.
15% survival rate for any startup category is pretty good. Why would one expect that specifically AI startups would do better than the rest? Personally, I would even expect lower rates, since now people see AI as a gold rush.
Data from the BLS shows that:
"approximately 20% of new businesses fail during the first two years of being open, 45% during the first five years, and 65% during the first 10 years. Only 25% of new businesses make it to 15 years or more."
https://www.investopedia.com/financial-edge/1010/top-6-reaso....
Purely guessing that startups fail even faster. This Hubspot article states:
"All these reasons bring up one question: How many startups fail? The reality is that 90% of startups fail."
https://blog.hubspot.com/the-hustle/how-many-startups-fail#:....