AWS GAI Update

By Patrick Moorhead - October 3, 2023

The Six Five Team discusses AWS GAI Update.

If you are interested in watching the full episode you can check it out here.

Disclaimer: The Six Five Webcast is for information and entertainment purposes only. Over the course of this webcast, we may talk about companies that are publicly traded and we may even reference that fact and their equity share price, but please do not take anything that we say as a recommendation about what you should do with your investment dollars. We are not investment advisors and we ask that you do not treat us as such.


Patrick Moorhead: Hey, and because we can’t get enough generative AI here, Amazon AWS brought a multitude of generative AI updates. Dan, what’s going on here?

Daniel Newman: Well, for more information on that, please follow Patrick Moorhead on, because he probably put out one of the better tweets. It got exactly two likes.

Patrick Moorhead: I think got 40. What are you talking about?

Daniel Newman: No, no. I clicked on the one that you shared. It shows two. I don’t know. Am I wrong? Anyway.

Patrick Moorhead: Well, Dan, it’s not all about the likes.

Daniel Newman: I don’t know. My children told me it is. So sometimes when I post stuff, they’re like, “You’re verified, but only five people like your tweets.” It’s not very nice. My kids are kind of mean. But I can say that here because I know they’ll never listen to my podcast.

Patrick Moorhead: Yeah.

Daniel Newman: By the way. They won’t read my books either. You think the one person you could count on to read your books would be… All right.

Patrick Moorhead: Disappointing.

Daniel Newman: So in the gen AI world, we are going to talk quite a bit about GA this week. And another bedrock, which is the managed service for the various foundational and large language models offered by AWS. And Pat, you know what I said in my tweet? Amazon is further democratizing the ability for enterprises and users to be able to implement and scale their artificial intelligence and generative AI ambitions through making and democratizing these foundational models and large language models. So if you’re already running your data and running your compute in AWS, it’s a shortcut to being able to simplify and implement your Gen AI ambitions. So you saw that, they have their Titan, which is a bunch of search and personalization engines that are going to be made GA as well. They’ve now decided to quickly adapt Llama 2, and then they’ve got a few other things that they came out with.

Probably the most important thing, Pat, that I noticed in this week’s announcement is they’re starting to want to communicate that hey, we have big companies, big customers that are committed to and leveraging AWS and Bedrock to take their gen AI strategies to market. So they put this in here and Pat, that’s why I said I loved your tweet because it was a nice consolidation of these thoughts for me. But big companies like BMW, big companies like LexisNexis, Rocket Mortgage, PGA Tour and many others are basically saying they’ve already committed to building generative AI on top of AWS Bedrock. Amazon Bedrock. Is it AWS Bedrock or Amazon Bedrock?

Patrick Moorhead: I don’t know. I always get confused there. I think Amazon confuses folks with that.

Daniel Newman: Okay well, the fact is, whichever one it is, this is being built on top of Bedrock. And I think what the company is doing right now is really twofold. One, they are really leaning into we are the open… It’s interesting, Pat, because AWS for a long time being the no hybrid, no multi-company has kind of 180 this whole Gen AI thing and said, “Look, we’re not going to try to build it in terms of ground up doing all the big large language model and then centralize it. We’re going to use all Anthropic and Cohere and we’re going to use Llama and we’ll… Whatever it is, you can use it.” And they’ve created that very open approach, which I think in Gen AI is good, as I do believe some of these bigger language models that were initially moats aren’t really going to be moats long-term. The foundational models, the smaller models, the industry specific models, that’s where it’s going to be at. We’re seeing it somewhat here. We’ve heard it from SAP, we’re hearing it from Oracle, we’re seeing it from other companies. Good progress from Amazon, AWS in terms of gen AI offerings.

Patrick Moorhead: All right. I’m glad you took all my content off my notes.

Daniel Newman: Your team was so good. I mean, look, as long as I gave you a pat on the back, I get to steal all your ideas.

Patrick Moorhead: You do. And sometimes I’ll steal your ideas and not give you any credit, so I appreciate you giving me the credit there.

Daniel Newman: I know, dude, I know.

Patrick Moorhead: So listen, I’m going to hit this maybe from a tangential angle, which was the meme out there was that AWS was way behind in generative AI. And that was issued by people that I respect out there. And so I really put my nose to the grindstone to say, what does late mean? They were the last one to announce, they were the last big one to go GA. But then again, they also have the largest enterprise AI estate out there of anybody. So I’ve also said, and we said this even right after the first Microsoft launch, that this is a marathon and not a sprint. If you’re not too far behind to go GA and getting customers on this, but AWS came out with all guns ablaring and they listed out 21 customers. And these are not just web folks who do nothing but share content, right? You have financial folks out there, legal folks out there, right? You’ve got Intuit, you have LexisNexis, NatWest, you have pharmaceutical companies like Merck, Rocket Mortgage.

So these are not just a bunch of folks who are going to jump off and take these massive risks without having payback. So I think that counts for something. One thing that I didn’t hear is Titan Embeddings is something that there was no description on what are the data sources? And not only do I not know the data sources, Amazon is not at least publicly indemnifying people if there’s something in there that have a copyright on it or not be owned by Amazon. So to be interesting as we jump into the IBM one and you cover that, I think it’ll talk about more of the why. But folks, AWS is very much going to be an IaaS and a PaaS play with generative AI. Don’t convince yourself that they’re not. They’re the largest purveyor of ML services for IaaS.

I’ve never measured that on PaaS, but it’s just a fact. And while generative AI technologically is very different from machine learning and deep learning and analytics, the type of things you’re trying to get out of it are very similar. Generative AI just does it better and more accurately. One final editorial here, I wish they didn’t have separate names for generative AI, Bedrock versus SageMaker. I know that technologically, SageMaker is a machine learning basically a complete IDE end-to-end process flow, but I think AWS will confuse people by having those things as separate. Out of the other side of my mouth, I’ll say that Bedrock is an AI service, not an ML service as defined. So yeah, maybe it deserved a different name. I just think it’s confusing.

Patrick Moorhead
+ posts

Patrick founded the firm based on his real-world world technology experiences with the understanding of what he wasn’t getting from analysts and consultants. Ten years later, Patrick is ranked #1 among technology industry analysts in terms of “power” (ARInsights)  in “press citations” (Apollo Research). Moorhead is a contributor at Forbes and frequently appears on CNBC. He is a broad-based analyst covering a wide variety of topics including the cloud, enterprise SaaS, collaboration, client computing, and semiconductors. He has 30 years of experience including 15 years of executive experience at high tech companies (NCR, AT&T, Compaq, now HP, and AMD) leading strategy, product management, product marketing, and corporate marketing, including three industry board appointments.