Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why AI Infrastructure Startups Are Insanely Hard to Build

nextword.substack.com

Post reply on HN
Print view View original

171–180 of 190 posts

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#171

Post by eropple »

devoutsalsa wrote: post #163
eropple wrote: post #154

Earlier quoted context omitted.

That's getting into perfectly-spherical-cow territory, though. Investors aren't logical and neither are founders, and there's a real chance that an investor-fair deal isn't going to get any bites. The BATNA for most founders who don't secure funding is "go get a job that pays a lot of money"; while most VC investment is lopsided in the investor's favor through other means (equity preferences etc.), not landing deals…

One could argue that illogical investing is called speculation or gambling.

One indeed could. Excuse me while I channel Kermit the Frog and take a long sip of my tea.

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#172

Post by nextworddev »

diwank wrote: post #147

While I agree that AI infra startups are hard to build, I strongly disagree with the idea that they are harder than foundational or application layer startups. I think it boils down to what you know and what resources you can muster . For instance, foundational AI startups are also ridiculously hard to build. You need an insane amount of funding, spend it pretraining models to stay competitive only to find that gains…

Author here. Yes, I explicitly called out the danger of thinking application layer startups are easier, because it totally depends on the founding teams' backgrounds and interests.

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#173

Post by jononor »

swalsh wrote: post #46
jononor wrote: post #26

Earlier quoted context omitted.

Anything SaaS that solves a painpoints for established industries. Those that have billions of turnaround for decades already, are not good at building tech themselves, and buy solutions/services to run their business. Bonus for low barriers to entry. Agriculture, logistics, real estate, energy, etc.

I have a theory that the days of established businesses that don't know tech is dwindling. A lot of companies which has adopted tech has started building a small foundation of talent internally. I think you're seeing this trend accelerate with the large tech companies laying people off. I have heard about top grade data science talent landing at some small sized health plan. My companies fastest growing competitor is…

Yes computer savvyness is on the rise, and have been for decades, and this will continue. But there are many levels: Ability to be a competent user, and competent buyer, ability to build it themselves. Then there are big difference in buy vs build culture. And preferences for type of solutions that are default buy vs default build. And finally, smaller and medium sized organizations are less likely to have internal teams. All this should be analyzed for the specific market, product and customer segment one targets.

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#174

Post by surgical_fire »

rapsey wrote: post #152
ToucanLoucan wrote: post #146

Earlier quoted context omitted.

> AI is a long term trend, next is more products on top of AI. Just like the internet. This is legitimately just the same damn hype train the tech sector is constantly attempting to create. Now AI is the next internet. Before that it was the metaverse. Before that it was NFTs. Before that it was cryptocurrency. Before that it was quantum. Before that it was VR. Before that it was AR. None of those were the revolution…

> This is legitimately just the same damn hype train the tech sector is constantly attempting to create. Now AI is the next internet. Before that it was the metaverse. Before that it was NFTs. Before that it was cryptocurrency. Before that it was quantum. Before that it was VR. Before that it was AR. Neither of those things you listed were mainstream trends. If you can not distinguish between fads and major trends th…

> Neither of those things you listed were mainstream trends

They certainly were sold as trends about to go mainstream.

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#175

Post by pas »

rapsey wrote: post #79
pas wrote: post #68

Earlier quoted context omitted.

if they are not profitable with these prices ... what the fuck they are doing!? do they just have company coke-athons all day every day?

It is a sales and engineering heavy business. Very difficult to generate large returns.

They are bundling open source stuff (Spark and Delta and so on), sure, they have their fancy IDE and whatnot on top, but they can let the community maintain things, scale back R&D, focus on things that matter to existing clients.

They have 1.6B revenue, a 50% YoY growth. And still not profitable. Hm, okay, recent acquisition on ML stuff, and of course probably burning hundreds of millions on cloud-GPU-AI shit.

Well, I guess as long as they have so big growth it makes sense to invest and raise ... and yeah that probably completely obscures the actual profitability of their core business. (Not to mention that they are probably spending all that money to try to expand their core business. To upgrade their value prop from cloud version of less-dumb-data-pipelines to 1-800-data-4-AI.)

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#176

Post by dvt »

marcosdumay wrote: post #161
dvt wrote: post #33

Earlier quoted context omitted.

In my humble opinion, a chat interface (API or not) does not a product make. Not to mention that Llama is free and competitive with both (same with Mistral, heck the 7B model works great on my RTX 3080). If you started a company that blew up because you made a badass product (and let's say you used ChatGPT under the hood), you would just eventually train and deploy your own model because an LLM is not a product .

> an LLM is not a product I imagine you meant to say that LLMs are comoditized. Getting the correct words here is important, as you can see by all the people disagreeing on the literal interpretation of your post. And yeah. LLMs have got the fastest transition from highly innovative singular product to plain commodity I've ever seen or read about. BSD licensed software libraries do not move that quickly. They were mo…

> I imagine you meant to say that LLMs are comoditized.

Yep, much better way of putting it!

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#177

Post by borroka »

threeseed wrote: post #7

Way too many founders don't understand the impact of competing with cloud vendors. Almost all enterprises have pre-committed budgets for cloud which means unless your product is FOSS it's going to be hard to convince someone to bet their business on it. Especially given that in this fundraising environment there is a 95% chance they won't be around in a year or two anyway. It's going to be a brutal few years especial…

That's the observation. I work for a huge company, both in terms of number of employees and geographical reach. I/we receive emails, offers, and declarations of interest each month/year from dozens of start-ups selling some AI infra system, labeling service, GenAI whatever.

We are not buying anything and we won't. We already have contracts in place with cloud vendors--the usual suspects: Google, Microsoft, Amazon--and the rest of the infra is developed in-house. Why should we buy subscription-based and high-maintenance products from an a16z-funded AI infrastructure company that might shut down operations in two years and say "bye-bye, it's been great"?

The former Uber/Michelangelo team that built Tecton, Netflix's Metaflow that later became Outerbounds, I have no idea who they are selling their AI infra products to.

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#179

Post by sirspacey »

ToucanLoucan wrote: post #146
rapsey wrote: post #71

Earlier quoted context omitted.

AI is a long term trend, next is more products on top of AI. Just like the internet. Biotech and space will both be trends but slower and less bubbly because the cost to play is high, though the returns are possibly huge in both.

> AI is a long term trend, next is more products on top of AI. Just like the internet. This is legitimately just the same damn hype train the tech sector is constantly attempting to create. Now AI is the next internet. Before that it was the metaverse. Before that it was NFTs. Before that it was cryptocurrency. Before that it was quantum. Before that it was VR. Before that it was AR. None of those were the revolution…

You could lump all of those trends together and they aren’t 1/10th of the adoption of chatGPT alone.

Re: Why AI Infrastructure Startups Are Insanely Hard to Build

#180

Post by rapsey »

surgical_fire wrote: post #174
rapsey wrote: post #152

Earlier quoted context omitted.

> This is legitimately just the same damn hype train the tech sector is constantly attempting to create. Now AI is the next internet. Before that it was the metaverse. Before that it was NFTs. Before that it was cryptocurrency. Before that it was quantum. Before that it was VR. Before that it was AR. Neither of those things you listed were mainstream trends. If you can not distinguish between fads and major trends th…

> Neither of those things you listed were mainstream trends They certainly were sold as trends about to go mainstream.

To suckers.
Post reply on HN