Menu
Sign In Search Podcasts Charts People & Topics Add Podcast API Blog Pricing

Nathan Lambert

๐Ÿ‘ค Speaker
1665 total appearances

Appearances Over Time

Podcast Appearances

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

Unless they're so good at models, which they are.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

Unless they're so good at models, which they are.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

It depends on where you think AI capabilities are going.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

It depends on where you think AI capabilities are going.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

It depends on where you think AI capabilities are going.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

The whole idea is human data is kind of tapped out. We don't care. We all care about self-play, verifiable data.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

The whole idea is human data is kind of tapped out. We don't care. We all care about self-play, verifiable data.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

The whole idea is human data is kind of tapped out. We don't care. We all care about self-play, verifiable data.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

You have to believe that, and you know, there's a lot of discussions that tokens and tokenomics and LLM APIs are the next compute layer, or the next paradigm for the economy, kind of like energy and oil was. But there's also like, you have to sort of believe that APIs and chat are not where AI is stuck, right? It is actually just tasks and agents and robotics and computer use.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

You have to believe that, and you know, there's a lot of discussions that tokens and tokenomics and LLM APIs are the next compute layer, or the next paradigm for the economy, kind of like energy and oil was. But there's also like, you have to sort of believe that APIs and chat are not where AI is stuck, right? It is actually just tasks and agents and robotics and computer use.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

You have to believe that, and you know, there's a lot of discussions that tokens and tokenomics and LLM APIs are the next compute layer, or the next paradigm for the economy, kind of like energy and oil was. But there's also like, you have to sort of believe that APIs and chat are not where AI is stuck, right? It is actually just tasks and agents and robotics and computer use.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

And those are the areas where all the value will be delivered, not API, not chat application.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

And those are the areas where all the value will be delivered, not API, not chat application.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

And those are the areas where all the value will be delivered, not API, not chat application.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

If model progress is not rapid, yeah, it's becoming a commodity, right? DeepSeek v3 shows this, but also the GPT-3 chart earlier, Kurt chart showed this, right? Lama 3B is 1200x cheaper than GPT-3. Any GPT-3, like anyone whose business model was GPT-3 level capabilities is dead. Anyone whose business model is GPT-4 level capabilities is dead.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

If model progress is not rapid, yeah, it's becoming a commodity, right? DeepSeek v3 shows this, but also the GPT-3 chart earlier, Kurt chart showed this, right? Lama 3B is 1200x cheaper than GPT-3. Any GPT-3, like anyone whose business model was GPT-3 level capabilities is dead. Anyone whose business model is GPT-4 level capabilities is dead.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

If model progress is not rapid, yeah, it's becoming a commodity, right? DeepSeek v3 shows this, but also the GPT-3 chart earlier, Kurt chart showed this, right? Lama 3B is 1200x cheaper than GPT-3. Any GPT-3, like anyone whose business model was GPT-3 level capabilities is dead. Anyone whose business model is GPT-4 level capabilities is dead.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

I don't think they care about that at all.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

I don't think they care about that at all.

Lex Fridman Podcast
#459 โ€“ DeepSeek, China, OpenAI, NVIDIA, xAI, TSMC, Stargate, and AI Megaclusters

I don't think they care about that at all.