Can $1 billion turn startup Scale AI into an AI data powerhouse?

Scale and its investors are confident the company can grow beyond a tool to help customers prepare data for AI into a software platform that plays a deeper role in helping customers build their own custom AI.

Scale’s latest round was led by venture firm Accel and includes additional commitments from investors including Y Combinator, Founders Fund and Tiger Global Management.

New strategic investors included Cisco Investments, Amazon.com Inc. and the venture arms of semiconductor companies Qualcomm Inc., Intel Corp. and Advanced Micro Devices Inc. Existing investor Nvidia Corp. also participated in the San Francisco-based startup’s investment round.

Field Chief Technology Officer Vijay Karunamurthy spoke to The Wall Street Journal last week at the Collision conference in Toronto about Scale’s ambitions and how the company plans to deploy new capital in the AI ​​arms race. The interview has been edited for length and clarity.

The Wall Street Journal: You’ve spoken about how this will be the platform to achieve “artificial general intelligence,” or AGI, where machines can learn and think like humans. What does this mean for Scale AI and how are you working with AI labs like OpenAI and Anthropic?

Karunamurthy: Our role has changed to become a “data foundry for AGI.” This is a journey that will continue for the next few years. [AI] At some level, research institutions are starting to get interested in AI. Not just in specific functions, but a lot of interest in how do we get models that can reason generally about the world and answer questions reliably at the same level that humans can trust? This has huge implications for society. We’re trying to stay on target just as research institutions are trying to stay on target.

The Wall Street Journal: Genesis Why is data so important in advancing AI, AGI, and other future technologies?

Karnamurthy: As these models get more powerful and are able to store more knowledge and inferences, the amount of data needed to saturate them grows exponentially. That data needs to be diverse. We are now often asked to power sophisticated datasets about how the world around us works. How do you answer questions about this table? If I tilt this cup at a certain angle, will the coffee stay in the cup? This is crucial to enable robots to interact with the world around them. But the data is still scarce. Even if we believe that AGI is still far away, in the next few years we will see advances in embodied agents that can use robotic arms and hands in physical environments.

The Wall Street Journal: What are the characteristics of your recent $1 billion fundraise?

Karnamurthy: We wanted to add more strategic investors, as their clients are all asking about the data side of the equation. [venture investments that companies are making] We’re focused on how we can improve our enterprise products. I’ve met with some of the more strategic investors to see how we can work with them. When you talk to Intel and Cisco, there’s a very strong focus on security and trust that their private knowledge of tweaking those models doesn’t leave the boundaries of the enterprise. They want to be able to audit how those models are accessed and have strict access controls.

The Wall Street JournalScale has new capital. How will it be allocated across the company?

Karnamurthy: We are hiring across the board and we are growing. [Scale’s full-time employees] That’s up more than 20% year-over-year. We’re also growing internationally. We announced that London will be our first official international office. We know there’s a lot of talent in London and a wide range of AI talent across Europe, so some of the funding will go towards expanding that as well. A lot of our funding is helping build out the human side of the equation and the model side of the equation. Both humans and technology play hybrid roles.