Rodrigo Liang

"at 10 kilowatts in SN40 and 19 kilowatts in SN50, they're so low in power that we can use air- cooled. And the beauty of air- cooled is 85 % of data centers that exist in the world today are air- cooled, and so that's what already exists. You're not forcing the buyers to come in and retrofit data centers every time they want to buy somewhat of a rack"

"SambaNova, deploying a rack at 10 kilowatts compared to 140 kilowatts of NVIDIA, 10 kilowatts goes into most of your existing data centers."

"a lot of our customers are cloud service providers. They're trying to build token factories. They're trying to provide AI services. They're trying to offer inference services."

"the next three to five years, I think you're going to see inference become 90 plus percent of everything we do, right? Training is, in the world of search, training is creating a new search algorithm. Inferencing is using the search, right? And so all of us will be doing inferencing and so that's 95%."

"as agents are coming in, everybody wants their own customized versions, so it's really hard to share. Sharing models on infrastructure becomes nearly impossible, because you want an agent, I want an agent. Everybody uses a slightly different agent. And so now you have this problem where, in order to serve this, the economics of the cloud providers is really poor"

"Once you get to reasoning, it's not 10x more. It's 100x more. People are generating 20, 30 page documents. And what's scarier is agentic systems, another 10 to 100x more per prompt"