Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: Are AI Startups a Bubble Poised to Pop?
3 points by razmoket_ 3 months ago | hide | past | favorite | 13 comments
Hey everyone,

I would love to have your opinion about AI startups, specificcaly those riding on the coattails of big-name LLMs like GPT-4, Claude, Llama, Bard .. It's like every other day we see a new startup popping up, aiming to revolutionize everything from legal services, marketing to customer support, chatbots etc. But here's the thing that's been bugging me: how sustainable is this model, really?

When Sam Altman dropped the news about the GPT Store it t felt like a game-changer, and honestly, it kind of spooked me. I can't help but wonder if a bunch of these AI ventures are skating on thin ice. If their whole gig is just tweaking OpenAI models with some fancy prompts, what happens when the big players decide to switch up the game? What value are their bringing ?

I'm super pumped about where AI tech is heading – the possibilities seem endless. But when it comes to these AI startups, I'm not quite on the hype train. It feels like they might be missing a trick or two. Just churning out prompts and feeding data into a pre-built AI – isn't that more of a cool feature than a full-blown business?

It reminds me a lot of what goes down on Amazon. You know, how they jump on the bandwagon of trending products, leaving the smaller sellers in the dust? Kinda feels like the same vibe here.

So, here's what I'm noodling on: Is there a way these AI companies survive long term ? Is this just a big bubble poised to pop ? I'd love to get your take on this!

I don’t have any expertise or experience in the area, but in general I wouldn’t want to underestimate the degree to which it is possible to build a successful product/company around an idea that depends very little on (proprietary) technology. Also, just because something is possible (e.g. ‘creating a GPT’ to do something) doesn’t mean people or organisations are compelled to do it if there’s an app or a service that somehow makes it feel slightly more streamlined and trustworthy. It seems to me that a lot of supposed ‘tech’ companies are in fact marketing successes or general business/logistics successes.

I seem to recall Mark Zuckerberg saying something similar about Facebook early on — that it would quickly be cloned by a behemoth like Microsoft and they’d be eaten. But it never happened.

Many of these companies seem to rely more on the hype around AI than on delivering solid technological innovation or real value to their customers. This trend is reminiscent of what happens in many emerging tech fields: there's a lot of excitement at first, but then reality sets in, and only those with genuinely useful and innovative products or services survive. Just like Mark Zuckerberg's early concerns about Facebook being overtaken, it's not always about the technology itself but how you use it and the unique value you bring to the table. In this light, the AI startup scene will face a shake-up, where only 1/1000 will survive in my opinion.

I agree. Most will quickly be shown to be entirely useless. But there is probably room for more than one company specialising in each AI-powered idea and it isn’t true that some company offering an easy way to create such apps, like OpenAI, will necessarily automatically dominate.

There's a lot of companies that add no real value and will implode, that's for sure.

But at the same time, these new technologies have enormous potential and will bring a shitload of value. Whether it will be tapped on by new startups, OpenAI or big tech — remains to be seen.

There is no such thing as an AI Startup, only a customer startup. To the extent that AI makes a new company better able to serve a customer, they will thrive. If they fail to draw a productive connection between AI and unmet customer needs, they will fail. It doesn't matter if you get funded, this is a universal truth.

VC throwing tones of money because people are making 'AI' startups. That's the point.

Right. I think there's the investor narrative, and then the actual value function offered to customers. Regardless of the former, very few would be an AI startup by the definition of the latter.

> Is this just a big bubble poised to pop ?

What does "popping" mean concretely? That some new companies that aren't profitable... won't last long and won't be profitable, just like it's expected from most startups? Why should anyone who isn't an employee/customer/investor/founder care?

In my view, the bursting of this bubble will mirror the internet bubble of 2000, creating a significant shake-up akin to a seismic event in the realm of tech.

You're grossly overestimating the significance of these businesses - AI startups that integrate with a LLM API are a rounding error in the overall market. They're tiny and there isn't as many of them as you think, nothing like 2000.

Yes, the bubble is poised to pop. The vast majority of AI startups are grasping in the dark for an idea that will capture enough market share that will make them sustainable, and very few will find that. Right now VCs are dumping tons of money into the space to find the things that work, but eventually the vast majority of those bets are going to lose and those startups will fail. The media will go bananas trying to paint this as the end of tech, the end of AI, the end of capitalism, but in reality, it'll just mean a bunch of startup founders and their employees will lose their income stream and move on to other things. The cycle repeats every time a new technology gains traction.

Fully agree but in this case it's a lit bit different. The only advantage you need to have to create an AI startup is too know how to write a prompt. Seems weird ..

everyone and their mother seem to slap the letters AI infront of everything, while i highly doubt it really is.

i feel most of it is a fad, its the same people who flocked first to crypto, then to nft's and now its all AI.

Real products will stay, but most will be gone just as quick as they popped up.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact