# Palantir and Nvidia CEOs discuss their latest partnership

*CNBC Television* · https://www.youtube.com/watch?v=cQV12w2z7jQ

## Summary

CNBC talks with Palantir CEO Alex Karp and Nvidia CEO Jensen Huang about their new partnership, which integrates Nvidia's accelerated computing and AI libraries into Palantir's Ontology platform. They explain how it automates enterprise decision-making, runs in any cloud or on-prem, and can be used with open-weight models or any hyperscaler. Karp argues it benefits American manufacturing and workers, and both push back on fears that AI simply takes jobs.

## Key points

- Palantir and Nvidia are partnering to combine Palantir's Ontology platform with Nvidia's accelerated computing and AI libraries.
- Ontology is described as a digital representation of an enterprise, including its security model, enabling decision-making to move to the frontline with or without people.
- Jensen Huang says the stack can run in any cloud, on-prem, or in a private cloud, on servers from vendors like HP or Dell.
- Alex Karp says the partnership enables faster, cheaper, more granular decisions and supports American manufacturing and workers.
- Karp argues AI will change jobs but make companies more productive, leading them to hire more people.
- Neither executive commits to near-term revenue accretion from the partnership.

## Notes

### Partnership Overview
- CNBC interviews Palantir CEO Alex Karp and Nvidia CEO Jensen Huang about their collaboration.
- The partnership combines Palantir's **Ontology** platform with Nvidia's **CUDA accelerated computing** and AI libraries, plus an open-source model they are working on together.

### What Ontology Does
- Karp describes Ontology as a digital representation of an enterprise, including its security model.
- It takes the capabilities of large language models and moves decision-making to the frontline, with or without people depending on the use case.
- Using Lowe's as an example: automating decisions like how the supply chain works, how it would work if it changed, and getting the cheapest goods to the right place amid disruption.

### Deployment Flexibility
- Huang says the stack runs everywhere: in any cloud, on-prem, or in a private cloud.
- Customers can buy servers from HP or Dell, or set up private clouds in other countries.
- Karp says customers can use open-weight models, open source models, any hyperscaler, or no hyperscaler at all.

### Speed and National Impact
- Karp expects the partnership to enable "2x" speed, and even faster in the future, at lower cost.
- He argues it is better for America: enabling domestic manufacturing, strengthening workers, and increasing competition.

### Jobs and Humans in the Loop
- Karp clarifies that "humans not in the loop" doesn't mean humans are never needed; in defense contexts humans must be in the loop.
- He suggests you can create more manufacturing jobs and make humans more valuable.
- Both argue AI will change every job but make companies more productive, which leads to hiring more people.

### Revenue Question
- Asked whether the deal is accretive to revenue near-term, Karp deflects, saying value creation for customers comes first and Palantir is already doing well on revenue and profit.

## Transcript

[00:00] And I'm just going to piggyback off of that conversation, Brian. We're talking about AI and an important collaboration between Palunteer and Nvidia, the fact that they're joining forces not to just analyze data, but actually to uh uh provide business decisions. And I I just Alex, I'm actually going to start with you uh the CEO of Palunteer just to explain does that mean no human intervention? Can you just explain how this would work and use Lowe's as an example? Well, first of all, I'm delighted to be on the stage with uh what many people I believe to be one of the great entrepreneurs of of my lifetime and a patriot. Um and it means a lot to me to be here. Um uh so we we have a product called Ontology. You

[00:41] could think of it as uh a essentially a digital representative of your enterprise including security model which essentially takes the disabilities that have large language models and allows you to move decision-m to the front line with people without people depending on how you're doing it. um in the context of loaves and and other enterprises allows you to essentially automate decision- making like how would your supply chain work how would it work if it changed uh how do you get the cheapest good to the right place given that there may be disruption but in general what's very very special I believe about our our partnership is we're going to be able to move 2x and I think in the future even faster so you

[01:16] get a much much faster and actually because you can control it in a more granular way much faster much cheaper and then the macro thing that I'm actually most excited about which for our country is because of that you can do manufacturing in America. You can make the workers stronger and better and you have more competition because you can use openw weight models, open source models, you can use you can use any hyperscaler. You can decide not to use a hyperscaler. This is much much better for America, much better for workers, much better for industry. It's cheaper, it's faster and obviously they have one of the world's top tech teams and we work very well with them. So Jensen just in in regards to that collaboration then

[01:53] who would actually be procuring the hardware? Is this like a company like Lowe's? Do they have to put the upfront cost uh you know with their existing GPUs and then working with Palunteer? Is that how it works or >> first of all it's amazing to think that the Palunteer ontology platform has achieved all of this and they're the fastest enterprise growing platform in the world. Um uh in a lot of ways it gives enterprises what we all want which is intelligence and and the ability to make good decisions fast and and now we're going to supercharge it with Nvidia's accelerated computing for the first time. And so Nvidia's CUDA accelerated computing is going to get integrated. We're going to be able to

[02:32] put uh Nvidia's AI libraries on top of that and open source model that that we're working on together in into the ontology platform and you'll run it in the cloud any cloud or you run it on prem or even a private cloud and so as you know Nvidia runs everywhere and so you could buy a server from HP or Dell uh you could be in another country and you want to set up your own private cloud we got no trouble with that and and so it doesn't matter we run everywhere and Palunteer runs everywhere >> but you're talking about the future then so this isn't something that would be accretive to revenues just in the near term. You're still working it out. And like I know Lowe's is an example, but are there other customers?

[03:06] >> Well, our challenge here, we just do what's best for our customers. We work downstream for value creation. So, we believe we know when it's faster, better, more accurate, and more flexible and uh creates a better future for them because they can choose how they want to run their applications, we'll do very well. And at Palunteer, we we're doing very well on revenue and profit. And we're, you know, that's not actually our primary motivation now. And the reason we've done so well on profit and on growth is because you just have to create value. This creates more value for the end user be >> and but you just said before too that the human may not even be involved with some of these decisions and I guess that

[03:40] is a major concern in this day and age that uh AI efficiencies could take away from or displace the job force. So couldn't something like this collaboration do that? >> Uh you know it's interesting when you say humans don't have to be in the loop. That doesn't inherently mean there aren't cases where humans need to be in the loop. For example, 40 years ago, we had 40x that we had 40 we 40 with 40x more manufacturing jobs than we have now. So it's like you can make more manufacturing jobs which are humans workingass humans. You can build things you couldn't build in the past. And those humans become more and more valuable than they were in the past. In the defense context, you have situations where humans have to be in the loop.

[04:19] Offense situations where humans shouldn't be in the loop like defense. Our collaboration makes all those use cases better for workers, for manufacturing, for defense, for offense, and in general because it allows people to use any kind of model, any kind of compute structure, it makes America stronger and better. Whenever a business is more productive, whenever they can move faster, their business grows, they hire more people. And so, so I think this this idea that that AI uh is going to take jobs is going to change everyone's job. That's for sure. But I think it's going to make companies more productive. We're more productive today because of AI.