6 ms·
85% of AI Startups Will Be Out of Business in 3 Years, Major Investor Says
- TobiWestside 3y agoSince the general rule of thumb is that only 10% of startups succeed in the long term, this statement is probably true.
- PUSH_AX 3y agoWait, 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.
- bombcar 3y agoI'd say the 5% could be the startups that fail later (after three years). Or like 86% of statistics, they are pulled from the are and only "roughly" correct.
- LatteLazy 3y agoSeems 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... [0]. https://www.bls.gov/bdm/us_age_naics_00_table7.txt https://www.bls.gov/bdm/us_age_naics_00_table7.txt
- ben_w 3y agoIf I had to guess, I would say 85% of all startups don't last 3 years, not just AI and today. Does that sound about right?
- Nux 3y agoYes. https://www.forbes.com/sites/neilpatel/2015/01/16/90-of-startups-will-fail-heres-what-you-need-to-know-about-the-10/ https://www.forbes.com/sites/neilpatel/2015/01/16/90-of-star...
- version_five 3y agoThis 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?
- nubinetwork 3y agoWhat will aicombinator do when ai flops? Go back to crypto?
- Gijs4g 3y agoHow can it flop when there are already millions of people using it every day?
- post-it 3y agoBy making negative money per user, like all flops.
- orbz 3y agoThe secret is having zero users so you make an undefined amount of money per user.
- pid-1 3y agoI don't know anyone who is using a LLM based product besides ChatGPT and Copilot. Moreover, I know two people who are paying for ChatGPT And a handful that are paying for Copilot.
- mathgeek 3y agoThere’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.
- scaredginger 3y agoI mean, duh?
- TrackerFF 3y agoA 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.
- remote_phone 3y agoOnly 85%?
- churchill 3y agoConsidering 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.
- sakex 3y ago0.01% 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.)
- Kye 3y agoBig 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.
- sakex 3y agoFair 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.
- Kye 3y agoGoogle: 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.
- assholeconny 3y ago[dead]
- andrewstuart 3y agoMost 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.
- kneebonian 3y agoYou're forgetting 5. An AI that generates porn.
- jjoonathan 3y agoNo, because that will probably make actual money.
- ourmandave 3y agoIf it's like AI art, it will be the porn of nightmares. AI Art-"Ruining Rule 34 since 2018."
- _gfwu 3y agoAlso "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).
- ryanmercer 3y agoBut, 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." https://www.investopedia.com/financial-edge/1010/top-6-reasons-new-businesses-fail.aspx#:~:text=Data%20from%20the%20BLS%20shows,to%2015%20years%20or%20more 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#:~:text=All%20these%20reasons%20bring%20up,they%20reach%20their%20second%20year https://blog.hubspot.com/the-hustle/how-many-startups-fail#:....
- swyx 3y agoexactly. this investor is unwittingly more bullish on AI than he thinks lol
- Oras 3y agoExactly. 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.
- itsoktocry 3y ago>Purely guessing that startups fail even faster 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.
- anonylizard 3y agoMost 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.
- wwkeyboard 3y agoThat's a pretty good success ratio as far as startups are concerned, right?
- renegade-otter 3y agoConsider the fact that many crypto companies are making a "pivot" to AI. This tells you all you need to know. https://coinmarketcap.com/community/articles/64b07f917043ec2500ea9c5f/ https://coinmarketcap.com/community/articles/64b07f917043ec2... "Hive, a Vancouver-based miner, aims to drive advancements in AI applications. Notably, it plans to support the growing web3 ecosystem." What does that even mean?
- martin_a 3y agoThis resonates well with the fact that all the "Crypto Experts" on LinkedIn have vanished and lots of "AI Experts" have emerged.
- VHRanger 3y agoIt means theyre looking for a16z to fund them
- dooraven 3y agowhy does it matter if they pivot? it's the best interest in their business to pivot
- code_runner 3y agoThe 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.
- dooraven 3y agoI 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.
- Havoc 3y agoClearly. It does seem likely that one of the 15% could be big though
- Hasz 3y agoNo 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.
- miroljub 3y ago15% 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.
- amelius 3y agoIf Nvidia introduces an AI tax, they will. Thanks Apple, for the idea!
- a13o 3y agoWe'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.
- blagie 3y agoMy 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.
- marban 3y ago"dot.com was just sort of like a way for us to interact on the internet and add to our life." Sure; it only killed retail, banks, civil discourse, among a few other minor areas of life.
- brk 3y ago85%, this is one of the more optimistic takes I've seen.
- rco8786 3y agoIs this materially different than the prediction for all startups in 3 years?
- code_runner 3y agoI’ll take the over
- benjaminwootton 3y agoThe 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.
- rewmie 3y agoThis 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?
- JCM9 3y agoWhen 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.
- dangoodmanUT 3y agoWell there are so many so that makes sense, but also I feel like you could say that about all categories...
- karmakaze 3y agoThis stat doesn't say not to invest, but rather to invest in many. Say a dozen so there's 85% chance one will be successful.