Public transcript

Palantir CEO Alex Karp: "It's Like No Glove, No Love. No Ontology, No Love."

By: Amit Kukreja
Length: 16 min
Language: English
Text from: the video's captions
Transcribed: 13 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

Palantir CEO Alex Karp sits down with Amit Kukreja to discuss why open-source models matter to Palantir, arguing that the value of AI is created at the application and ontology layer rather than in the frontier models themselves. Karp explains that enterprises are frustrated paying high prices for tokens while their proprietary alpha migrates to frontier labs, and that open source plus ontology provides safety, fine-tuning, and sovereignty. He defends Palantir's pro-American stance, criticizes Europe's protectionist approach to tech, and says Palantir is at the beginning of a massive opportunity with demand beyond what it can supply.

Key points

Karp says enterprises are paying large sums for tokens that are hard to value, while the alpha of their business migrates to the frontier labs they pay.
He argues models are excruciatingly effective when used with an application layer, and that almost all the value is created in the ontology with those models.
Open source plus ontology allows fine-tuning on proprietary data and insights without going through a frontier lab, protecting enterprise alpha.
Karp says the phrase 'no ontology, no love' has gone viral in corporate America, comparing it to 'no glove, no love'.
He questions outsourcing the moral veracity of the Department of War to frontier lab people who have never met anyone in uniform.
Karp calls Europe a template for how not to do tech policy, saying closing borders to US tech screws European companies.
Palantir has a strong preference for American models, though some clients use models from all over the world, including China.
Karp says Palantir's biggest problem right now is scaling to meet demand it cannot fulfill in both US government and US commercial.
He states Palantir is at the very beginning of a revolution and believes it will play a disproportionate role in American industry accelerating forward.

Questions it answers

00:00The open-source revolution

Why did Karp's recent CNBC appearance change how the Valley thinks about AI models?

Karp says the appearance, building on work with Nvidia, exploded into a revolution because enterprises realized they were paying absurd sums for tokens with unclear value while their alpha migrated to frontier labs. Technical participants in the market know Palantir gets it right, and now they are copying its approach.

Enterprises are paying a lot for tokens whose value is hard to see.
The alpha of a business is migrating to the frontier labs being paid.
Palantir's approach is now being copied by others.

03:00Ontology and safety

What does Karp mean by 'no ontology, no love'?

He says the phrase has gone viral in corporate America. Ontology serves a safety function, protecting enterprises from models that may leave unwanted products behind, and it also helps businesses expand their tribal knowledge and alpha while protecting it.

Ontology provides a safety layer when deploying models.
The safety function matters more as models come from all over the world.
Businesses need to expand their alpha while protecting it.

06:12Open source and fine-tuning

Why is open source plus ontology important for orchestration?

Open source allows enterprises to fine-tune models without dealing with a frontier lab, exploiting the alpha deep in their own proprietary data and insights. Karp gives the example of building a cancer drug on proprietary data and insights into how the human body works.

Fine-tuning is very hard to do without going through a frontier lab.
Open source lets enterprises build on their own proprietary alpha.
Valuable enterprises have specialized ways of doing things.

07:46Sovereignty and the Department of War

Why does Karp object to outsourcing military AI to frontier labs?

He asks in what world the Department of War, which he calls the most credible institution in America, outsources its moral veracity to people far away who have never met a man or woman in uniform. He says Palantir has always been on the side of protecting individual rights, data, and sovereignty.

Palantir has always protected individual rights and data sovereignty.
Karp questions frontier labs' fitness for military decisions.
Sovereign AI is already working in the US government context.

09:43Customer response and scaling

How have customers responded to Palantir's open-source sovereign narrative?

Karp calls the response colossal, saying it is what customers want and need, and that there is a valve release of anger at the options previously provided. Palantir's biggest problem is now scaling to meet demand it cannot fulfill in US government and commercial.

Customers are livid at the options previously provided.
Palantir is rebuilding infrastructure and fine-tuning capability.
Demand exceeds what Palantir can supply.

11:06China and being pro-American

Is Palantir endorsing China by using open-source models?

Karp says Palantir is very pro-American and partnered with Nvidia, with a strong preference for American models, though some clients use models from all over the world including China. He argues Europe's approach of closing borders to US tech is the template for how not to do it, producing weak companies.

Palantir has a strong preference for American models.
Europe's protectionism screws European tech companies.
All models should be scrutinized with better oversight.

14:00Palantir's future

What does it mean to deliver value as an enterprise software company in the age of AI?

Karp says Palantir is at the very beginning of a massive revolution and believes it will be a trillion-dollar company. Orchestrating compute, ontology, product, and open models is more art than science, and demand for what Palantir does is beyond what it can supply.

Karp believes Palantir will be a trillion-dollar company.
Orchestration of all these elements is more art than science.
Demand for Palantir's work exceeds its supply.

Notes

Context and the open-source shift

Karp credits Amit Kukreja with helping people understand Palantir since its DPO, saying millions participated in the public markets because of his missives.
He describes a recent CNBC appearance as sparking a revolution in how the Valley thinks about frontier versus open-source models, building on work with Nvidia.

Why enterprises are turning away from frontier labs

Enterprises are paying large sums for tokens whose value is hard to see, while the alpha of their business migrates to the frontier labs they pay.
Frontier companies are also behaving in the public sphere in ways that blow up American politics, Karp argues.
Models are excruciatingly effective when used with an application layer, and almost all the value is created in the ontology with those models.

Ontology: safety and alpha

The phrase "no ontology, no love" (like "no glove, no love") has gone viral in corporate America.
Ontology serves a safety function, protecting against models that may leave unwanted products behind, which matters even more as models come from all over the world.
Businesses need to expand their tribal knowledge and alpha while protecting it, which requires compute, ontology, and owning the weights.

Open source and orchestration

Open source allows fine-tuning without dealing with a frontier lab, exploiting the alpha deep in your business on proprietary data and insights.
Karp cites the US government context, asking in what world the Department of War outsources its moral veracity to people far away who have never met anyone in uniform.
Palantir has always been on the side of protecting individual rights, data, and sovereignty, from PG and Foundry to Ontology, Maven, and sovereign AI.

China, Europe, and being pro-American

Karp says Palantir is very pro-American and partnered with Nvidia, but some clients use models from all over the world, including China.
He calls Europe the template for how not to do this: closing borders to US tech produces weak companies nobody believes can compete.
All models should be scrutinized with better oversight, but overt protectionism screws the people you are trying to help.

Outlook

Palantir's biggest problem is scaling to meet demand it cannot fulfill in US government and commercial.
Karp says Palantir is at the very beginning of a massive revolution, repeating his belief it will be a trillion-dollar company, and that orchestrating compute, ontology, and product is more art than science.

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