Will any company form a defensible moat around LLM-based AI before 2025?
29
172
590
2025
18%
chance

This market resolves YES, if in my opinion, any company has formed a defensible moat around an LLM-based AI, and it is not yet 2025.

What does a defensible moat look like? There's no single perfectly objective test, but it generally looks like this:

  • The company is making a lot of money

  • They are far ahead of their closest competition both in terms of market share and/or in terms of profits (market share is a really strong indicator tho)

  • Other companies have a hard time catching up and putting a dent in their market share, even if their technology is "just as good" or "almost as good"

  • Alternatively, if nobody else's product/service is anywhere close to as good, that's another good sign of a defensible moat

  • There's a strong network effect or other source of natural or artificial lock-in that makes it hard for users to leave the ecosystem

  • The business press talks about them having a defensible moat and even identifies what it is

If a legal or natural monopoly is clearly established, that would also probably do the trick.

Get Ṁ1,000 play money
Sort by:

if i successfully dig a ditch around openAI’s corporate headquarters and fill it with water by 12/31/24 will this resolve YES

What if open LLM's render centralized LLM's less effective so moat-building becomes easier?

@PatrickDelaney Explain please

@LarsDoucet Certainly. So if you check out this: https://github.com/imaurer/awesome-decentralized-llm you can see that there is a huge list of decentralized LLM's. I have some markets which are tracking various aspects of AI capabilities which I list here: https://www.patdel.com/ai-ml-capabilities-markets/ some of which are LLM-dependant, all of which require the usage of an independent inference to be uploaded for judging. Essentially, benchmarking is being done to compare LLM's to centralized services such as GPT-4 to compare various aspects of their performance. If we make the assumption that decentralized LLM's are reasonably close in performance to centralized LLM's, it's reasonable to believe that many other apps will be developed around those LLM's from many developers, not just OpenAI, making OpenAI's moat movement mostly moot. Well, not necessarily mostly moot but I couldn't resist the alliteration.

@PatrickDelaney So where is the easier moat development in this scenario?

@ian Google's already got a moat, in the form of all of its users and products whereas OpenAI has ChatGPT. If open source LLMs develop sufficiently that renders ChstGPT useless, handing the throne and further moatbuilding back to Google and hence why OpenAI is 10% owned by Microsoft, it's a hedge.