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

Ask HN: What is a 2024-2030 moat for AI

news.ycombinator.com

Post reply on HN
Print view View original

1–10 of 25 posts

Ask HN: What is a 2024-2030 moat for AI

#1

Post by jonas_kgomo »

What are the key moats and competitive advantage in the AI industry(specifically if one is working in the Generative AI space) for the next 6 years? Some simple examples, you are building on powerful technologies: AI forecasting based on graphcast, or an assistant based on GPT or computer vision based on Roboflow and Scale API. Products like Perplexity is built on GPT and others and Rabbit is built on Perplexity.

Looking ahead to the years 2024 to 2030, it becomes increasingly crucial to identify the moats that can help AI companies maintain a sustainable competitive advantage. Some say this is users, data or compute. But what makes sense for 1-3 person team.

Re: Ask HN: What is a 2024-2030 moat for AI

#2

Post by nextos »

It's hard to say. Good moats, at least mid-term, are rarely algorithms. It's usually data and infrastructure.

However, for some applications I am interested in, I think that robust representation learning solutions could give a significant edge.

But that is mostly an open problem in high-dimensional spaces.

Re: Ask HN: What is a 2024-2030 moat for AI

#5

Post by stuartaxelowen »

nextos wrote: post #2

It's hard to say. Good moats, at least mid-term, are rarely algorithms. It's usually data and infrastructure. However, for some applications I am interested in, I think that robust representation learning solutions could give a significant edge. But that is mostly an open problem in high-dimensional spaces.

And not just data, but hard to match data generators. I’m skeptical of the defensibility of any given licensed dataset.

Re: Ask HN: What is a 2024-2030 moat for AI

#7

Post by uf00lme »

Has a 1-3 team moat ever really existed with new technologies? My best guess would be to somehow get your tech embedded into the defence industry.

My goto moat test: is there an obvious ‘death by Amazon’ route? Amazon can strike fear into entire industries with scale and capital. Very few tech companies could withstand a full frontal assault by Amazon. I can’t see how this trend won’t get worse if AGI occurs during our lifetime.

The only reason there is not more AI or robots doing the majority of the work available to humans, is either it’s cheaper to hire humans or companies just haven’t gotten around to it yet.

Re: Ask HN: What is a 2024-2030 moat for AI

#8

Post by tqi »

I think successful AI companies in 2030 will get there through a combination being above some minimum threshold of technical competence and execution and luck (right place, right time, right professional network). I think the former is probably lower than we'd expect, while the latter is much more important than we expect. I also think that in 2023, we will tell ourselves that those companies were inevitable because XYZ obvious reasons created a moat.

Re: Ask HN: What is a 2024-2030 moat for AI

#9

Post by bhag2066 »

The moats are 1) network effects, 2) switching costs, 3) economics of scale, 4) low cost producer, and 5) brand. So an AI technology where each new user makes it more valuable for the existing users, there are data lock-ins that prevent customer switching, training costs are spread across more users, there is access to lower cost chips/storage/compute and a brand that attracts more users.

OpenAI already appears to hit a number of these categories.

Re: Ask HN: What is a 2024-2030 moat for AI

#10

Post by throwbadubadu »

Love to look stupid.. as foreign speaker and after reading all comments and also all translations in the dictionary, I still have only a vague meaning of "moat" in this context. Could someone define precisely please?

Things like "network effects" vs "1-3 team moat" vs "human written text" now totally confused me.. what will protect it? What will be the most precious? Who succeeds?

Post reply on HN