I think there are serious risks for LLM development (i.e. a better DeepSeek could be released at any points), but also some serious opportunities.
1. The game is still early. It’s hard to say what moats might exist 5 years from now. This is a chaotic field. 2. ChatGPT/Claude spend a lot of attention on their frontends, the API support, documentation, monitoring, moderation, lots of surrounding tooling. It’s a ton of work to make a high production-grade service, besides just having one narrow good LLM. 3. There’s always the chance of something like a Decisive Strategic Advantage later.
Personally, if I were an investor, both would seem promising to me. Both are very risky—high chances of total failure, depending on how things play out. But that’s common for startups. I’d bet that there’s a good chance that moats will emerge later.
I think there are serious risks for LLM development (i.e. a better DeepSeek could be released at any points), but also some serious opportunities.
1. The game is still early. It’s hard to say what moats might exist 5 years from now. This is a chaotic field.
2. ChatGPT/Claude spend a lot of attention on their frontends, the API support, documentation, monitoring, moderation, lots of surrounding tooling. It’s a ton of work to make a high production-grade service, besides just having one narrow good LLM.
3. There’s always the chance of something like a Decisive Strategic Advantage later.
Personally, if I were an investor, both would seem promising to me. Both are very risky—high chances of total failure, depending on how things play out. But that’s common for startups. I’d bet that there’s a good chance that moats will emerge later.