What to Expect at AWS re:Invent 2025 (with Jo Peterson)

It’s time for the biggest cloud and AI event on the planet! AWS re:Invent 2025 is about to kick off.

Leonard Lee of neXt Curve caught up with Jo Peterson, Principal Analyst at ClearTech Research, to share our predictions for the present and future of cloud and AI computing just before we stepped into the analyst pre-brief a day before AWS CEO Matt Garman’s keynote.

Last year, Matt Garman led with cloud and security. Will AWS hold to the foundation of their business in a year of subdued enterprise AI adoption, or will they go all in on AI, AI, AI?

Hit Leonard and Jo up on LinkedIn and take part in their industry and tech insights.

Please subscribe to our podcast which will be featured on the neXt Curve YouTube Channel. Check out the audio version on BuzzSprout or find us on your favorite Podcast platform.

Also, subscribe to the neXt Curve research portal at www.next-curve.com and our Substack (https://substack.com/@nextcurve) for the tech and industry insights that matter.

NOTE: The transcript is AI-generated and will contain errors.

Transcript

Leonard Lee: 0:00

Hey everyone, this is Leonard Lee, executive Analyst at Ncur, and I'm here with

Jo Peterson: 0:04

Joe Peterson, chief Analyst at Clear Tech Research.

Leonard Lee: 0:07

Yes, and we are here. In Las Vegas for AWS reinvent 2025, and it is going to be the year of

Jo Peterson: 0:18

agenda.

Leonard Lee: 0:19

Oh my gosh.

Jo Peterson: 0:21

Just hands.

Leonard Lee: 0:22

Just hands.

Jo Peterson: 0:23

Okay.

Leonard Lee: 0:24

Yeah. So I'm, really excited to be here with you and, we had a wonderful dinner last night. We did. Right? Thank you for recommending the sushi at restaurant with a, really funky chef. We decided Tomago was a chef's choice.

Jo Peterson: 0:39

the things you learn in Vegas?

Leonard Lee: 0:40

Yeah. Well, you know, this year's event is gonna be all about tomago. Right. Really? Yeah. Yeah. Okay. Exactly. So let's unpack agent AI for a moment. Are you expecting that AWS is going to announce or share with the 200 and some odd analysts who are on the globe that are here this year? Which is like, I think a record maybe, right?

Jo Peterson: 1:03

It is a record and., I expect overall that the conference is gonna be quite large. I'd read somewhere that there's gonna be 60,000 visitors per reinvent this year, so crazy. And it's across, what, four or five different hotels?

Leonard Lee: 1:17

That's pretty insane.

Jo Peterson: 1:18

Yeah. Insane. But they have us here at the win, and we're getting a little bit of a taste of our agenda to come. And I think it is, we're gonna talk a lot about agent ai and we're gonna talk about use cases for agent ai and we're gonna talk about. Maybe the idea of super agents And how super agents are going to evolve. and we're starting to see, at least from a consumer side of the equation Companies announce super agents, but more, at least I think with super agent as a personal assistant. So we're seeing them. On the consumer side, again, where consumer, these super agents are gonna learn our preferences and say you and your wife wanna go on a trip to Italy. Well, they're gonna learn based upon your preferences that you, Leonard Lee like to stay in five star hotels.

Leonard Lee: 2:11

Yeah.

Jo Peterson: 2:11

Perhaps. And they're gonna book Gondola rides because they know that you wanna romance your wife while you're in Venice. Maybe. I don't know.

Leonard Lee: 2:20

Okay.

Jo Peterson: 2:21

we're way off track here, so

Leonard Lee: 2:22

that's No, no, no, we're not. Well, I was gonna ask you. what the hell is a super eight? It's,

Jo Peterson: 2:29

it's,

Leonard Lee: 2:29

it's your, so you're answering that? Yeah. No, no. That's wonderful. And so the things that pop into my mind as you went through that, description was privacy and hyper personalization, but personalization that is closed loop because you're actually having the agent act on your behalf, right? Either based on intent or some perception of intent. Or just a prediction. Right?

Jo Peterson: 2:58

But I start to think about all the ways from a security perspective, even the consumer side of the equation. Yeah. Could go haywire, right? Oh, yeah. Because that super agent assumes your identity and it takes on your permissions, it's a proxy, and it takes on your pro permission. So what if it goes rogue back to the jazz hands and says, well, g. I'd like to book, I'd like to approve a$10,000 charge to my American Express to go to a five star hotel in Venice, right? Yeah, yeah. But you personally aren't even aware that it's doing that, but yet you are liable for that expense. Maybe that's just a crazy, notion in my head, but I start to think about permissions. And that acting as a proxy and where are those lines? And we're not hearing about any of that. Maybe it's just too new. But that's what's happening on the consumer side. It's not happening yet on the enterprise side,

Leonard Lee: 3:55

but I think the ambition is there. And quite frankly, I'm not quite sure they've cracked a code on that for the consumer either. yes, you have personalization engines that are driving, the delivery let's say. Ads is probably the most common thing. but I think you're using much more deterministic, agents, RPA ish kind of stuff, for instance on Apple you have, shortcuts, things that. are more like triggers that you're attaching to specific events. this is old school stuff, But it's a consumer that can now do that within the context of their, device in this case. Which might be an iPhone or an Android phone of whatever variety, on the enterprise we're looking at, something much more complex, I think. Yep. In terms of an agent and not. That is a proxy of you, but they're more of like, like little worker beads, right. That you're creating to, take on tasks that might be part of a workflow or a process or even a function. What are the possibilities there in terms of what you're, you articulated earlier about things going haywire. Yeah. Happening and so the important thing here is that we're seeing a lot of vendors, start to put together all these frameworks, the ENT frameworks to try to act, actually take away agencies from these agents.

Jo Peterson: 5:26

You know, my advice to business units spinning this up outside the purview of their security teams is stop get them involved. Identity is gonna become the new black in Agentic ai. Right. and what I mean by that is maybe your company hasn't installed microsegmentation yet.

Leonard Lee: 5:49

Yeah.

Jo Peterson: 5:50

If they haven't, that's a problem because now there's these agents running around being able to go and do things. In systems that maybe the security guys don't know are there? Yeah. Maybe if you're a great big company with a lot of smart engineers and a lot of money, well you've got those things in place. Yeah. But maybe you're a mid-market client. Yeah. And you've got a director of it that is wearing the security guy's hat because you don't have a security guy or gal. and your marketing team has gone out and installed some sort of agentic. Or your HR team has gone out and Whichever team, it doesn't matter which team. The point is, be a good citizen of your company and get your IT folks involved in what it is you're installing or putting in. So that's one of the things that you can do tangibly as a worker and a good citizen of a company, don't create shadow ai.

Leonard Lee: 6:50

Everyone's still on a learning curve, even the vendors. Sure. And this might be one of the issues and, taking whatever kind of lessons learned that they do have that can be productized. And I know that a lot of'em are doing it and trying to do it, but also recognizing fundamentally what the limitations of these technologies are. Right. I mean, there's a lot of shoehorning going on. they're trying to. Fit AI into, let's say, a solution where AI might not even be part of the solution. You don't need it. But there's this notion, and we hear it everywhere, that AI is going to take over everything, but that may not actually be the case. Maybe the more sensible thing is looking at. You know what Amazon is, kind of famous for is working your way backward from the problem and figuring out what are the technologies that really make sense. To solve the problem rather than having this fixation on AI that everyone seems to have, which is overriding almost every single agenda out there. And you know, no one's talking about cloud anymore.

Jo Peterson: 8:00

Well think, which is really weird I think. I think what's happening, and you bring up a really good point. And I think. That's perfect. We're here to be wowed. We're here to hear about what they're thinking about. We're here to think about innovation, and they want us to be thinking about innovation. Right. And, that's all great. But underneath, as we pull back the layers

Leonard Lee: 8:22

Yeah.

Jo Peterson: 8:23

What you said is really important. What I see is a starting a re-imagining of our cloud footprints. And with ai. And how does that new design look? And is it in fact more of a container, less environment? I think when we start to build out these new solutions or these new footprints, they're gonna look very different both from a network perspective, A server perspective, the things that we think about and have thought about for the past 15 or 18 years in cloud design are being shifted.

Leonard Lee: 9:01

Yeah.

Jo Peterson: 9:02

Right.

Leonard Lee: 9:02

And you're making a great point because what you're probably looking at is a need for both container based architectures and serverless. Yeah. and we are seeing that. But the tricky thing is, how the AI technology at the model level continues to shift and how it plays out. On top of hardware. Yeah. What it implies in terms of virtualization and what you virtualize, how you virtualize it, whether or not it's container based or microservices based, or serverless. These are all things that haven't been quite worked out yet. And then there's this whole topic of economics, well, how much does it actually cost to run all this stuff? And is it economical? Right? Yeah. In aggregate. And so there's still a lot of open questions. So it'll be interesting. I'm really looking forward to what, AWS has to say. And I'm sure it's going to, have reverberations everywhere. So. And we're

Jo Peterson: 10:02

just at day one, so catch back up with me on day four. Yeah. Maybe I'll have a different spin on things. Yeah,

Leonard Lee: 10:06

same here. Same here. All right. Well it's great chatting with you. Thank you. And we're gonna have fun. We're gonna have some fun. So anyways, thanks for tuning in and we will see you later. Um, probably three days, right? Bye all. Yeah. Bye.