Mitesh Agarwal

Mentioned in 1 episode

"Sora first came out — not a lot of people saw it, but I got to see a little bit behind the scenes on the video generation models. The amount of memory — I remember looking at the Google VO model back and it needed four H100s to run like a 10-second clip."

Mentioned in 1 episode

"When we speak with TSMC for fab capacity, they're also wanting to know kind of like can you scale — do you have the balance sheet to do that."

Mentioned in 1 episode

"Lambda builds technology on the cloud and services. I didn't want to leave Lambda — started the company with Stephen there — and Lambda Cloud, but early 2025, late '24, reasoning models had come out."

Mentioned in 1 episode

"I remember looking at the Google Veo model back and it needed four H100s to run like a 10-second clip. Was completely memory bound on bandwidth and capacity."

Mentioned in 1 episode

"People want to use it how they're currently using it and going to continue to use it, which is based on Nvidia's stack but also primarily based on PyTorch."