What % of these do you think will be here by 2026?

[-]

omnisvosscio@reddit (OP)

I am sorry, I always forget, https://www.insightpartners.com/ideas/state-of-the-ai-agent-ecosystem-use-cases-and-learnings-for-technology-builders-and-buyers/

[-]

FuzzzyRam@reddit

learnings for technology builders

lol, it's just as dumb as I assumed it would be.

[-]

uwilllovethis@reddit

Depends on when/if the AI bubble bursts. Could be 100%, could be 50%.

[-]

omnisvosscio@reddit (OP)

Maybe, I just feel like AI is way too useful for it to be more then 50%

[-]

GreatBigJerk@reddit

As someone who has been a dev for a long time, it's probably 50%.

Even if it isn't officially that number by 2026, many will be skeleton businesses with a "founder" and sales person. They will try to drum up business by offering things they absolutely can't offer. Then they will slowly disappear as domains and things they registered gradually stop getting renewed.

By 2030, it will be 70-80%. Some of those won't die per se, but will be acquired by a bigger company.

There will also be companies that will shift into an entirely new vertical. An example of this is Slack. The company that originally made it started as a game studio. Now it's just another Salesforce product.

I have a couple boxes of prototype hardware and business cards from dead companies. Some of it was really cool. The tech and VC markets are brutal.

[-]

uwilllovethis@reddit

Only 48% of publicly traded IT companies survived the dot-com bubble collapse. For privately funded startups, I imagine that % is much lower. And you could argue that the internet is as if not more useful than AI.

[-]

omnisvosscio@reddit (OP)

True

[-]

Orolol@reddit

Even if the Bubble isn't bursting and Hype train continue his venture, there's no way that 50% of those survive 1 year, let alone 100%.

[-]

Ok_Phase_8827@reddit

nice

[-]

_pdp_@reddit

Many of these are not even agents - it is just a random graph of companies that do things with AI.

[-]

Mickenfox@reddit

The buzzword treadmill must go on.

[-]

omnisvosscio@reddit (OP)

true

[-]

Mickenfox@reddit

Elefant AI

AI Friends in Minecraft

Billion dollar business I'm telling you.

[-]

ihexx@reddit

ironically the little cockroaches that are just GPT wrappers would probably survive the longest cause yeah they're low margin, but they're also zero cost things that can be run by 1 guy in his basement

[-]

irfantogluk@reddit

Comma AI (OpenPilot)

[-]

mrjackspade@reddit

Its like a vaporware placemat.

[-]

SeymourBits@reddit

Now that is a better thought out product idea than the majority of these... calling Shark Tank!

[-]

nrkishere@reddit

Those who have their own hardware infrastructure will survive. Wrappers like cursor, perplexity are doomed if they can't raise the next round.

[-]

krakoi90@reddit

I thought the same thing about two years ago. I was convinced GitHub Copilot would win the AI coding assistant race before it even started. But then last year I canceled my subscription because, frankly, it seemed to be getting worse while the competition was getting better.

It's a similar situation to the cloud computing boom 10-15 years ago. People argued that Amazon, Google, and Microsoft would crush all other online services because they could deploy and maintain them so much cheaper. That didn't really happen. Instead, some of these services just moved their backends to the cloud platforms of the mentioned companies.

[-]

nrkishere@reddit

You are wrong in your understanding. Things like linode, hetzner, scaleway, ovh existed before aws, long before google cloud and azure. Linode is acquired by akamai, which is itself a big company. And no big company really moved out of aws or azure to hetzner or scaleway. If anything, they became big enough to self host at scale. Things like vercel, netlify are wrappers around aws. People using those services only benefit aws.

Now coming to the copilot thing, I personally find copilots or coding assistant pretty useless. But I don't really see how cursor is any different from vscode w/ gh copilot integrated. Perplexity is a web search engine with AI summary. Every search engines do that now, including the built in search of brave browser. Many of such "AI" startups are not really solving any problem. A small startup can still beat giants if they can provide better product than competitors, have superior UX, better pricing plan etc. One of such example would be deepseek, which is not only equally capable of openAI & claude's model, but also have significantly lower price. But cursor, devin, perplexity don't belong to this list. In a overcrowded market, you need something more than just paid shills doing astroturfing marketing

[-]

krakoi90@reddit

Things like vercel, netlify are wrappers around aws. People using those services only benefit aws.

Yes, but those companies and projects aren't part of Amazon. They use AWS, so yes, they benefit AWS, too. But they are legitimate companies that haven't been killed off by a competing service from Amazon, Google, or similar, even though they are "just wrappers of AWS" according to your logic.

In the same way, Cursor could become a legitimate and profitable company using Anthropic models in the background, while still providing a better user experience than competitors like Copilot. They aren't necessarily doomed to fail within a few years.

but I don't really see how cursor is any different from vscode w/ gh copilot integrated.

That's because you haven't used it yet, particularly not the composer part. I'm not a Cursor shill, by the way. We could also talk about Windsurf, a similar product that I also think is better than GitHub Copilot.

[-]

nrkishere@reddit

You forgot the "can't raise next round" part I mentioned earlier. Vercel and the likes are heavily funded, still operating at loss with barely making 25 mil in revenue. There were many of such aws wrappers, like gatsby cloud which failed once vc money dried out, but nothing value was lost from aws' side.

Even during funding summer, VC investments start to consolidate after some point. This is why uber and lyft are the only major player in the segment, airbnb have no legit competitor etc. One or two of the AI wrappers will get big, while the rest will either shut down or be acquired by bigger players. This is how things have worked in past, the trend will follow even if there's no bubble. If there's a bubble, then most of these will be wiped out overnight like in .com bubble.

[-]

Ok-Cucumber-7217@reddit

In 10 years less 10%, but in 2026: 200% There are no signs of slowing down, VC is still pouring into these AI agentic whatever

[-]

Expensive-Apricot-25@reddit

next to none. "agents" are trivial to make unless your designing your own model from scratch

[-]

Any-Demand-2928@reddit

My view is that most of the guys who got there early (2022 - 2023) and raised hundreds of millions of $ will die off. Many new players coming on the scene and the big guys don't really hold any advantage, infact in some cases their advantages eroding with new LLM developments. Consolidation will happen a few years out but by that time new startups will have caught up and only then will it begin.

I'm a believer in AI and LLMs to bring a lot of value but there's no doubt about it there's been a hype the past 2 years and companies who should not have been raising a lot of money did. So many overvalued startups who aren't going to be able to perform as well as they hoped and won't be able to raise their next Series C or Series D. What happens then? They'll just die off.

Most of these startups have been raising money by distorting the truth at best or just out right lying at worst. Most companies that have implemented AI in some shape or form has been trialing a bunch of different ones from different startups and the amount that have long term commitments is quite low. You can start to see how these startups could've made big promises.

[-]

-Olorin@reddit

It’s so strange to live in a time where even the prospect of total automation can’t unseat the giants. All R&D is either done within or consolidated by just a few large companies. It seems it was Joseph Schumpeter, not Marx, who best predicted the current form of capitalism.

[-]

ErikBjare@reddit

I at least know that my own gptme will still be here :)

[-]

Only-Letterhead-3411@reddit

Most people give the classic "the bubble will burst and only the biggest a few remain" answer but we have to keep in mind that in other fields of technology, there's millions of companies we never ever heard about and they do fine. World is bigger than you think. It shouldn't surprise you that there's so many companies to list.

Technology field is extremely competitive and fast moving. So the companies that can adapt and use their cards fine will do well, the ones that don't will disappear.

[-]

NickNau@reddit

I like "SuperAGI" branding. Should be better than regular shitty AGI we all are forced to struggle with daily.. . . . right?

[-]

djstraylight@reddit

This image is a little old. Under Search, it still shows Bard. Already, some of these are gone.

[-]

throwAway9a8b7c111@reddit

Just skimming through ~70%

First, its a bad list generally when it puts something like Meta in multiple categories which have nothing to do with their actual revenue drivers.

Secondly, We haven't quite hit the M&A phase in the AI Market yet so I doubt we have a ton of consolidation in 2025.

Third, most of these companies have runways/funding that will extend beyond 12months.

[-]