The changing construct of MOAT in AI-powered businesses
A leaked memo from Google ignited many discussions on the existence of a true moat for AI-powered businesses. It said:
βWe do not have a moat. Neither does OpenAIβ!
This is completely a contradiction to the popular belief that LLMs are built with unbeatable technology and dataset powers and it would be impossible to match the powers of the big ones.
Hereβs what I think.
In the age of AI-powered products, true βtechnology MOATβ will be a myth. 3 reasons.
1. The LLMs have democratized building mind-blowing applications 100x easier than before. Anything you build, 100 others are building the same thing, and probably out there already.
2. Open-source movements are already proving to be much more powerful than the walled gardens, making it tough for even the current LLMs to keep their moat. (No, I donβt believe that 100s of millions of dollars would be needed to build an LLM and I donβt believe that it is going to be one or two winners ruling the market. Still in doubt? Checkout Vicuna from Huggingface)
What can be the real moat?
- Deep specialization β Build for a space that you understand, go deep!
2. Personalization β This is something that is hard to come by. But can definitely bring about a difference.
3. GTM β The smarter your distribution is, the better. Includes your community
4. The UX β How seamless your UX is with the journey of the user? AI calls for blending in, not changing behavior.