Public transcript

Alex Karp & Jensen Huang on Groundbreaking Palantir NVIDIA AI Partnership - Full Interview CNBC

By: Vampyre Drakul
Length: 12 min
Language: English
Text from: the video's captions
Transcribed: 12 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

CNBC interview with Palantir CEO Alex Karp and NVIDIA CEO Jensen Huang at GTC about their newly announced AI partnership. They explain how Palantir's Ontology platform will integrate NVIDIA's CUDA accelerated computing and AI libraries to automate enterprise decision-making, runnable in any cloud or on-prem. Both executives argue AI strengthens American workers and manufacturing, praise Trump's pro-energy and pro-jobs policies, and note China is currently excluded from NVIDIA's forecast.

Key points

Palantir's Ontology is described as a digital representation of an enterprise, with a security model, that moves decision-making to the front line with or without people.
NVIDIA's CUDA accelerated computing and AI libraries will be integrated into the Ontology platform, runnable in any cloud, on-prem, or in a private cloud.
Karp argues the partnership enables faster, cheaper, more granular decision-making and supports manufacturing in America.
Huang says AI is good for working-class workers, closing the technology divide because everyone can now program a computer via AI.
Both CEOs praise President Trump's pro-energy, pro-manufacturing, pro-jobs policies.
Huang says China is out of NVIDIA's forecast entirely, and any trade deal would be a bonus.
Huang cites David Sacks's framework: aim for American technology to hold ~80% global market share, built on American standards.
Karp says Palantir's primary motivation is customer value creation, not revenue, though the company is doing well on both.

Questions it answers

00:00The partnership

What are Palantir and NVIDIA actually building together?

Palantir's Ontology is a digital representation of an enterprise with a security model that lets large language models drive front-line decisions, with or without humans. NVIDIA's accelerated computing, AI libraries, and open-source models will be integrated into it.

Ontology automates decisions like supply-chain routing for customers such as Lowe's.
The integration makes decisions faster, cheaper, and more granular.

01:52Hardware and deployment

Who buys the hardware and where does it run?

Customers can buy servers from HP or Dell, run in any cloud, on-prem, or in a private cloud — NVIDIA and Palantir both run everywhere. Karp adds that flexibility to use open-weight models and any (or no) hyperscaler is better for America and for workers.

Deployment is cloud-agnostic and on-prem capable.
Open-source model choice is a core selling point.

03:06AI and jobs

Will AI displace workers?

Karp says AI changes jobs but makes workers more valuable and enables US manufacturing; Huang argues AI closes the technology divide since anyone can now program a computer, and NVIDIA itself is hiring more people because of AI productivity.

Humans stay in the loop for defense use cases.
Both frame AI as augmentation that raises worker value and pay.

04:59White House and policy

How do the CEOs view the administration's tech policy?

Both praise Trump's pro-energy, pro-manufacturing, pro-jobs agenda, citing approved power generation and DOE land releases, and framing AI factories as major job creators.

Karp calls it a pro-energy, pro-manufacturing, pro-jobs moment.
Huang says his first meeting with Trump focused on manufacturing in America.

06:48America's AI dominance

Can America stay dominant in AI?

Karp says the US has the chips, software, engineers, culture, and risk appetite to dominate. Huang cites David Sacks's framework of aiming for ~80% American technology market share worldwide, built on American standards.

Both see AI leadership as America's to lose.
Open-source developers in China building on American tech is seen as a win.

09:00China and trade

What about China and trade?

Huang says China is out of NVIDIA's forecast entirely — "at zero" — so any trade deal the president achieves would be a bonus for the country and investors.

Both CEOs express confidence in the president's dealmaking.
The interview ends on the positive note of both companies' share price gains.

Notes

The Partnership

Palantir CEO Alex Karp and NVIDIA CEO Jensen Huang announce a collaboration pairing Palantir's Ontology platform with NVIDIA's accelerated computing.
Ontology is described as a digital representative of an enterprise, including its security model, that takes the abilities of large language models and moves decision-making to the front line — with or without humans in the loop.
Example use case: Lowe's supply chain — automating decisions like getting the cheapest goods to the right place amid disruption.

Deployment Flexibility

NVIDIA's CUDA accelerated computing and AI libraries, plus open-source models the two are working on, will be integrated into Ontology.
Customers can run it in any cloud, on-prem, or in a private cloud; hardware can be bought as servers from HP or Dell.
Karp stresses customers can use open-weight models, any hyperscaler, or no hyperscaler — which he calls better for America, workers, and industry.

Jobs and AI

Karp pushes back on AI displacement fears, arguing AI makes workers more valuable and enables manufacturing jobs in the US.
Huang: AI is the first technology to close the technology divide — anyone can now program a computer; NVIDIA is more productive and hiring more people.
Both praise Trump administration policies on energy, manufacturing, and AI diffusion.

China and Outlook

Huang says China is out of NVIDIA's forecast — "at zero" — and any trade resolution would be a bonus.
He cites David Sacks's framing: aim for American technology to hold ~80% global market share, built on American standards.
Karp calls Palantir the most technically deep enterprise software company and says the collaboration will "supercharge" it.

Text from the video's own captions. Summary and notes written by AI from the transcript, so check anything important against the recording.