AWS Summit DC 2026 | Logan Sammons, Palantir

Clips
More from AWS Summit DC 2026

Logan Sammons

Strategic Architect

Palantir

play_circle_outline AWS–Palantir deep integration enables day‑one operations, moves AWS DOW provisional authorization to tier‑two provider

play_circle_outline Model‑agnostic AIP enabling selection of different LLMs per use case.

play_circle_outline Avoiding vendor/model lock via architecturally sound, model‑agnostic enterprise foundations.

Info
Transcript

Logan Sammons, Palantir

Logan Sammons

Strategic Architect Palantir

In this interview from AWS Summit Washington, D.C. 2026, Logan Sammons, strategic architect at Palantir Technologies, joins theCUBE + NYSE Wired's Gemma Allen to discuss how Palantir and AWS are helping government agencies move from AI experimentation to mission-ready execution. Sammons explains how Palantir's tier-two service provider authorization pulls directly from AWS's tier-one DOD provisional authorization, giving customers seamless security integration and the ability to operationalize data on day one. He breaks down how AIP, Palantir's model-agnostic... Read more

explore Keep Exploring

What is the relationship between Palantir and Amazon Web Services (AWS), and how is that relationship evolving? add

How does AIP enable organizations to use language models, and why does Palantir emphasize a model-agnostic approach? add

How does your company ensure customers remain flexible in choosing AI models and avoid being locked into a single AI service, subscription, or infrastructure? add

bolt Powered by CUBE AI

Logan Sammons, Palantir

search

Gemma Allen

>> Welcome to The Cube. We are here at AWS Summit in Washington, D.C. 2026. It is a historic week for America and we are about to chat to some of the folks behind AWS, a company making history of its own. We are surrounded by technologists, builders, innovators, all here to answer one simple question. AI is a lot of hype. How do we turn that into mission impact? We have a great lineup today. I'm Gemma Allen and we are going to be bringing you all the conversations to answer exactly that. Joining me now is our first guest, Logan Sammons, senior architect of Palantir. Welcome, Logan.

Logan Sammons

>> Thanks, Gemma. It's great to be here.

Gemma Allen

>> So like I said just there, this is a very unique week in America, right? 250 years.

Logan Sammons

>> Absolutely.

Gemma Allen

>> 250 years of a democratic government. We are thinking a lot about what the future holds right now from the perspective of technology. Maybe start and explain to me the relationship between AWS and Palantir and how that's evolving.

Logan Sammons

>> Yeah, absolutely. When we think about how we partner with cloud service providers like AWS, it's first and foremost to understand that we meet our customers where they're at in their mission. A lot of our customers choose to be on AWS. So it's important for us to meet them there and then to also strengthen our relationship with AWS and others so we understand how we integrate directly with their infrastructure. We're proud to have a relationship with AWS where we pull their tier one DOW provisional authorization into our tier two service provider authorization. There's not many of those and so that means that we sit directly on top of their infrastructure and customers that utilize Palantir, everything from access controls, security controls and authorizations, there's a seamless integration between tier one of AWS's and our tier two, bringing them the opportunity to operationalize data on day one.

Gemma Allen

>> So we want to get into the technology that makes all this possible, right? Palantir has certainly got a very unique technical capability.

Logan Sammons

>> Yes.

Gemma Allen

>> You said a lot of your customers choose AWS. Why do you fundamentally think that is?

Logan Sammons

>> At least within the government space, they were the first mover. So it was the only option for adoption. And as other CSPs or cloud service providers came on board, there's movement there too. But AWS holds core missions that were early adopters of cloud and they continue to exist not only within the U.S. government and federal civilian space but within the defense industrial base as well. And so when we think about their multi-year head start in the cloud business, it's easy to understand why a lot of customers already exist within AWS, missions exist within AWS and their enterprise are powered by some form or a lot of services on the AWS infrastructure.

Gemma Allen

>> So when we think about accreditation and speed and deployment and the race that we're in right now, especially as a lot of public sector organizations are feeling the pressure to keep a pace, it's technological evolution that's scaring even hyperscalers if we're frank.

Logan Sammons

>> Right.

Gemma Allen

>> How do you think about the tech stack for this? Break that down for me from the perspective of Palantir. You guys have some very interesting tech capabilities. We know Foundry. We now have AIP on top of that Ontology. Maybe let's go to AIP first.

Logan Sammons

>> Sure.

Gemma Allen

>> How does that fundamentally change what Foundry was doing for customers 18 months ago?

Logan Sammons

>> It's easy, right? AIP allows us to use agnostic models and LLMs to understand data better, to help our customers understand data better. The key with AIP and all of our Palantir technology is we strive to be model agnostic. And so as we look at what the customer base is doing with AI, many people just think, "Oh, I'll adopt a model. I'll write some prompts and I'll run one or two agents." That doesn't do a lot of change. That's great for experimenting, but where AIP comes into play is giving our customers the flexibility to choose language models that best fits a specific need within an organization and to choose another model that best fits the need for another outcome. And that's the power of AIP in a nutshell.

Gemma Allen

>> And I guess some of the customers you're working with, they don't really have the privilege of risk, right? Risk is not possible inside many of these environments.

Logan Sammons

>> Correct.

Gemma Allen

>> If we're thinking about FedRAMP, IL6, there is no opportunity for any sort of risk there. So it provides a certain level of security, I'm sure. How do you think about it from the perspective of security and of the evolution of that whole space?

Logan Sammons

>> Yeah, that's a great question. If we think about how the government has utilized AI and the progression of AI, it was first a novel concept, like how will we use it in research and development? How do we use it within a sandbox? And they quickly realized that experimentation inside of a sandbox doesn't have great actual tangible outcomes. So as they look to open their understanding of AI and how they manage risk within AI, they bring through using accreditations like FedRAMP, like the DOD IL6 for our impact levels and even the intelligence community using it under ICD 503. And so as those models become accredited, it's easy for us to then pull them in to our products and services and create an opportunity for customers to select which of those accredited models are best suited for their data outcomes.

Gemma Allen

>> So let's talk about the unlock that is Ontology. If you think about it, if you could use the same capability and opportunity that Ontology provides to all of AI, right? That knowledge mapping every single relationship integration.

Logan Sammons

>> Yes.

Gemma Allen

>> I mean, it could unleash so much, right? It could solve for the enterprise conundrum that many find themselves in.

Logan Sammons

>> We definitely think so.

Gemma Allen

>> How do you think about it from the perspective of public sector though? Because we know that oftentimes data's unstructured, knowledge lives in people's heads.

Logan Sammons

>> Yes.

Gemma Allen

>> It hasn't been a space that's been known for agility and even like strong governance, right? We know file shares are decades old for example. How does this actually work in practice?

Logan Sammons

>> Yeah. Being an architect, I always want to go to a whiteboard to explain that, but I think it's much easier if we understand that Ontology sits above language models. They sit above AI services. And so when we think about how it sits across of an enterprise, regardless of agentic AI, generative AI, or a baseline of AI models, it helps the enterprise understand and map their data and enterprise structure and architecture into a machine language that all of them understand. And that's the power of Ontology. We don't strive for our customers to focus on subscription or token usage and it's quite the opposite. How do we best utilize specific language models, specific data models to drive to specific outcomes? When we think about how it completely integrates from AIP, Mission Manager, Ontology and Foundry, what we focus on is our customer's outcome for success. It's not just good enough to write language models to integrate agentic AI for it to produce more data that a human has to then synthesize and drive to a decision. How do we think about it when we deeply integrate it into an enterprise and it's pulling forward those decision making processes, those behaviors that help make decision makers more informed. And I think that's really the goal that we focus on. We're outcome driven rather than just being engineering driven.

Gemma Allen

>> And when we think about the role of data and how that's evolved and some of the historical challenges in unlocking data, sometimes that relates to vendor lock, right? Vendor lock made systems quite tight, quite constrained, you're locked in, they're sticky.

Logan Sammons

>> Yes.

Gemma Allen

>> This new world of AI, we can't really figure out yet how sticky any of these LLM preferences are. It's kind of 2BD, right?

Logan Sammons

>> Yeah.

Gemma Allen

>> From the perspective of Palantir and what you're developing and building, how do you think about that unlock? How do you think about setting these projects up for success?

Logan Sammons

>> Yeah. And I think you phrased it right, Gemma. We don't know what the future of AI looks and we don't know how really sticky it could be for customers that span financial services to healthcare to governments. The best approach that we continue to drive towards is being model agentic, right? We want agnostic approaches to where our customers have the flexibility to use the models that they need for the outcomes that are desired. More importantly, we focus with our products and services to build a foundation of an enterprise that is architecturally sound to where it doesn't matter what model that you use. It doesn't matter what subscription you have and it doesn't matter what token services that you really want. As long as we have a strong foundation of an enterprise architecture that supports what the business is driving towards, then we have the opportunity through AIP and Ontology to then select models to have our customers say, "This works best for us for project B, another one works best for project C." And that's the flexibility that is really, I think, misunderstood by a lot of people with Palantir. They think, "Oh, we're locked in to Palantir. We're locked into an AI service. We're locked into an infrastructure provider." We help our customers drive to an architecture that is above all of that, that is sound for their enterprise and their business for their outcomes.

Gemma Allen

>> So let's talk about the priorities. It is a very interesting time in the world of tech and the world of government. A lot is happening very fast. We saw today that Mythos, that pullback is now back for certain customers. We know there's a lot of threats on the horizon. How do you guys think about the next 12 months? Because I think it's hard to think beyond that and what are the priorities from the perspective of your relationship with AWS?

Logan Sammons

>> Yeah. Specifically with AWS, we're highly focused on continuing to strengthen the relationship of understanding what our customer's needs are over the next horizon. Engineering opportunities and platforms and services that are going to meet those that are seamless to our customer, that requires a trusted relationship, specifically with AWS to where we understand and we're communicating as two companies working together to drive towards a community of outcomes. And so when we think about capacity planning, a new thing in government, a lot of talk is around multi-region. What does multi-region really mean? How do we get there? What are the cost considerations? What's parity of services? And so it's great to have a partnership and relationship with AWS that are consistently looping us in to understanding what is happening on the ground, what is needed for the next future, what are some things that we can do better and what are things that we can more importantly, especially within the government space, pull forward and get pre-authorized, pre-accredited, have audits and security reviews and make sure that they're transparent, not only with us as we fold in, but they're transparent with their customers? Because again, our goal is not to have something that's desired and ready to go and then spend the next 12 or 18 months seeking a government authorization. It's really important for companies in B2B relationships like we have with AWS to strive towards a goal of operational use day one.

Gemma Allen

>> Wow. But some of those things you listed off there are problems that felt like they could take decades to solve two years ago, right? So it's certainly a fascinating time. Logan Sammons, thank you so much for joining us on TheCUBE.

Logan Sammons

>> Thanks, Gemma. I appreciate it.

Gemma Allen

>> I'm Gemma Allen here with TheCUBE at AWS Summit in D.C. We have a phenomenal lineup coming up today. We're talking sovereign AI, cybersecurity, the world of technologists in public sector. Don't go anywhere.