“We’re at the beginning of a completely different era of computing.” – Rodrigo Liang

Duration: 2:37 573 transcript words EN summary

Video summary

SambaNova is positioning itself at the forefront of a new era in computing by focusing specifically on inference infrastructure, which aims to drastically reduce costs and democratize access for everyone. The company recently achieved a significant milestone with a billion-dollar funding round, bringing its total valuation to 11 billion dollars. This impressive financial standing is driven by investor recognition of the critical need for additional chips and data center racks that operate at much lower power consumption and cost structures. By addressing these fundamental issues, SambaNova enables organizations to deploy AI solutions more efficiently than ever before. The core argument presented is that inference has become the central pillar of the current AI economy, shifting the focus from model training to the practical application of those models in real-world scenarios. As the industry moves into this inference phase, long-standing challenges regarding energy consumption, operational costs, and data center space have moved to the forefront. SambaNova's technology is designed to solve these specific pain points, allowing businesses to run hardware that is best suited for their specific class of models. This approach not only optimizes performance but also directly tackles the sustainability and economic constraints that are currently limiting widespread AI adoption. The conversation also touches on the broader implications for the industry, including the recent calls from major tech leaders like Musk, Dario Amodei, and Sam Altman to slow down AI development for safety reasons. Rodrigo Liang agrees with this sentiment, emphasizing that safety is crucial not only for frontier models but equally important for the underlying infrastructure. He notes that the challenges faced in building safe and efficient hardware mirror those of model development, suggesting that paying attention to both sides of the AI world is essential for responsible progress. This balanced perspective highlights how infrastructure innovations can support safer and more sustainable AI ecosystems. Looking ahead, the team identifies a major opportunity in the next six months centered on heterogeneous infrastructure, which allows for mixing and matching different types of hardware to optimize performance and efficiency. The ability to select the best hardware for specific model classes represents a significant leap forward in solving the cost and power issues that many companies are currently facing. This flexibility is expected to redefine how data centers operate, enabling them to handle increasing workloads without proportionally increasing energy usage or expenses. Ultimately, SambaNova's journey reflects a commitment to continuous innovation, ensuring that their technology evolves rapidly enough to meet the dynamic needs of the market while keeping safety and efficiency at the heart of their mission.

Read the full video transcript

So, what do we got? Just off CNBC, what exactly does SambaNova do? >> Well, we work for on inference infrastructure, you know, driving the cost [music] structure down so that you can democratize inferencing for everybody to take to make use. >> [music] >> How much money have you guys raised? What's the valuation? >> So, we just closed a billion-dollar round in total, two and a half billion in the company. [music] The latest valuation is 11 billion. >> What are the numbers behind 11 billion-dollar valuation? Give it to me in 10 seconds. >> This shows [music] us an incredible valuing by the investors of infrastructure, the need [music] the additional chips, additional racks that go into data centers at a much, much lower cost, much, much [music] lower power. >> What was the key inflection point? >> Inference is at the center of the AI economy now. And so, as the world [music] is moving into inference, all the issues we've been talking about for years around cost and power and data center space is becoming front and center. >> [music] >> And so, now people are using SambaNova technologies to actually help them solve those problems. >> Three big names, Musk, Dario, Altman said, "We need to slow things down." What are your thoughts? >> I think it's great that the leadership start to think about safety and the importance of making sure that the technologies that we're building on [music] the frontier model side are in fact safe for everybody to use. Same on the infrastructure side, we have same challenges that [music] and that's where SambaNova plays around power and data center and cost. And so, those two sides of the AI world in fact have challenges and I think it's great that we're actually paying attention to it. >> Hardest moment building SambaNova so far? >> It's an incredibly challenging knowledge to continue to invent and innovate. And you look at all the different players out there. And so, you know, we SambaNova we're hitting this inflection point because, you know, our power and cost structure really hitting kind of the pain points that customers have today. >> What's one thing that the best founders do differently? >> You know, to be able to look at the market needs and continue to respond and react and making sure that your technology solve the problem that's needed today. And so, that ability to actually keep looking and getting the feedback from the market and making adjustments incredibly valuable. >> One thing you wish you knew 25 years ago. >> 25 years ago I wish we could have seen kind of where where AI was going. AI is a technology that's been around for 30 some years, you know, and we haven't kind of only now come back to it. I think if we'd have been investing the [music] whole time I think we'd be in a different world. >> What's the biggest opportunity ahead in the next 6 months for you and the team? >> Look, I think you know, the the world's now looking at heterogeneous infrastructure and so the ability to actually run hardware that's best for that class of models and so I think we're super excited about the opportunity to mix and match and being able to use our technology [music] to solve these cost and power issues that people are seeing.