Amazing write-up, lots of new technology here that I wasn't aware of. I recently thought a lot about Google Gemini, and wrote a little on the subject. Would love your thoughts on it, Vivek and Sabrina!
Currently LLM companies have no switching cost. The switching costs will come from where a company currently stores their data. AWS still has a huge lead over Azure and Google cloud. If Amazon releases an amazing foundational model, it could dominate the enterprise AI space since most companies already have their data there
I see. A 2 trillion parameter will be nothing to sneeze at. I'm super excited to see what specific problems do they try to solve apart from being just another chatbot model. I think given the enterprise focus of a potential Amazon model, low cost per token, action oriented agents, and large context windows similar to Claude 3 and Gemini 1.5 will be table stakes. What gets added on top of these table stakes will be exciting to see.
Thank you for such a useful insight! I am wondering why you say Amazon has no models of their own. They do have Titan and its flavors for text generation. And they are working on another one which is yet to be announced.
Amazing write-up, lots of new technology here that I wasn't aware of. I recently thought a lot about Google Gemini, and wrote a little on the subject. Would love your thoughts on it, Vivek and Sabrina!
https://futuretelescope.substack.com/p/ai-as-a-product
Thanks, Punit! Will take a look at your post this week, appreciate the flag and support!
Great overview of some of the major players. Amazon is the one I’m really interested in watching. Thanks for sharing
Yes, great article. Curious to know - why Amazon specifically? Any innovations they are working on that you're curious about?
Currently LLM companies have no switching cost. The switching costs will come from where a company currently stores their data. AWS still has a huge lead over Azure and Google cloud. If Amazon releases an amazing foundational model, it could dominate the enterprise AI space since most companies already have their data there
I see. A 2 trillion parameter will be nothing to sneeze at. I'm super excited to see what specific problems do they try to solve apart from being just another chatbot model. I think given the enterprise focus of a potential Amazon model, low cost per token, action oriented agents, and large context windows similar to Claude 3 and Gemini 1.5 will be table stakes. What gets added on top of these table stakes will be exciting to see.
Definitely, these models need differentiation. Being slightly better on the benchmark tests aren’t going to shift the playing field significantly
Awesome! BTW saw what you're doing with your newsletter, great stuff! Subscribed right away 🙌
Thank you! I’ll take a look at your stuff as well 🙏🏼🤝
Thank you for such a useful insight! I am wondering why you say Amazon has no models of their own. They do have Titan and its flavors for text generation. And they are working on another one which is yet to be announced.
Antropic Claude is their investment. Claude is a sophisticated general purpose model, purpose-built for Enterprise.
Great post. Anthropic looking more like AWS model to me. Love this space right now.