Neil Movva

Mentioned in 1 episode

"Cerebras, Groq, and maybe a couple others coming out of stealth have made a very interesting bet on not just building another GPU but actually building a different kind of accelerator that focuses on a different memory hierarchy. They want to maximize the amount of SRAM on the chip and use that as very fast memory for weights and KV cache."

Mentioned in 1 episode

"The new type of data that we cared about a lot was the interaction data from people using ChatGPT and giving feedback signals on what they liked and didn't like."

Mentioned in 1 episode

"All those companies — you could take your pick: Base10, Fireworks, Together — they all focus on low latency inference, and they were pulled in that direction by one very important customer: Cursor."

Mentioned in 1 episode

"It makes perfect sense for OpenAI and Anthropic to do what they do. There's a sensitive topic around distillation which I think is a very core piece of the relationship between closed and open frontier."

Mentioned in 1 episode

"We all make fun of xAI for having some challenges with total flop utilization on its clusters, but the reality for the rest of the world is it's far, far worse."

Mentioned in 1 episode

"Jensen is really good at making his friends billionaires. He's made CoreWeave a many-billion-dollar company. And there's no need for him to destroy that goodwill."

Mentioned in 1 episode

"Deep Seek certainly publishes really interesting work to compress the KV cache further and further. They're making good progress."

Mentioned in 1 episode

"I don't look for CUDA experience at all. That's actually a huge red herring. CUDA as a concept or GPU as a concept have evolved so much in the last 5 years. There's no point asking for 10 years of experience."

Mentioned in 1 episode

"One of our customers, Parallel Web Systems, seeks to build an index over the whole internet and monitor the internet in real time for changes. That is the kind of crazy exabyte-scale task that you need a very different kind of scale of intelligence to achieve."

Mentioned in 1 episode

"The most conservative players in the entire AI stack have got to be the infra players, whether that's data centers or even more conservative is TSMC, the chip infra people."

Mentioned in 1 episode

"It wasn't really something that we could actually bet on until late last year with Opus 4.5. Opus 4.5 was the first agent that was at all suitable for longer horizon tasks."

Mentioned in 1 episode

"Hence why we split the DRAM manufacturing into an entirely different company like Micron, SK Hynix, and Samsung. These are the best companies in the world to do this."

Mentioned in 1 episode

"The best processes that we have in the west, like Intel, not that far behind — at worst maybe 2x worse performance per watt, and the gap is just far smaller than you would make it out to be if you follow the chip board dialogue."

Mentioned in 1 episode

"There's a long tail of other companies that are popping up — net new companies are great — such as Etched, Senova, or dMatrix. All these companies are popping up and I think the main challenge for them is scale."

Mentioned in 1 episode

"Modern deep learning is built on Flash Attention. Modern transformers are built on Flash Attention."

Mentioned in 1 episode

"You can imagine Siri running in the background all the time to understand all the emails you received in a day, all the text messages you receive in a day, and having a much more encyclopedic view of your life and how to be helpful. Right now there are still point solutions and so you end up doing a lot of prompting. Siri is not very proactive."