Substrate Labs, an AI infrastructure business formed by Stripe, Pulumi, and Robust Intelligence, has launched a $8 million seed investment to provide elegant APIs that allow developers to build more modular AI systems, ushering in a new era of software creation. Lightspeed Venture Partners led the round, which included investment from South Park Commons, Craft Ventures, Guillermo Rauch (Vercel), Immad Akhund (Mercury), Will Gaybrick (Stripe), and others. Led by co-founder and CEO Rob Cheung and co-founder Ben Guo, the firm recently announced the formal launch of its API, which is suited for running ensemble models.
“The AI market will inevitably shift from training models to using inference to put models to work—a massive market that is currently far too inefficient,” said Nnamdi Iregbulem, a partner at Lightspeed Venture Partners. “Rob and Ben are strong engineers with over a decade of experience creating APIs and growing systems for companies such as Stripe, Substack, and Venmo. They’re in a good position to abstract away the complexities of implementing the correct AI model for each job, allowing engineers to maximize development with sophisticated tools previously only available at huge tech companies.”
Today’s AI developers are most successful when they divide a challenge into tasks and install the best model for each one. However, the great majority lack access to internal infrastructure capable of running dozens of AI models in modular pipelines, leaving them with the option of using a single huge AI model for every task or piecing together solutions from many API vendors and GPU service providers. Both approaches produce AI applications that are unreliable, slow, and expensive due to low GPU use.
With Stripe’s developer-first strategy in mind, the Substrate team is creating a powerful AI API that uses curated open-source AI models across multiple modalities such as text, images, audio, and semantic vectors, and optimizes these models to run at scale – with a focus on how quickly one model’s outputs can be used as another’s inputs. Substrate’s foundational infrastructure is tightly integrated with SDKs that make it simple to create “graph relations” between models on the Substrate platform, allowing developers to build complex multi-inference AI workflows in a few lines of code, with fewer data round trips and maximum parallelism. Substack and Maven were among the early customers.
“New AI models offer enormous potential, but the tools developers have access to today are impeding the deployment of intelligence into software systems,” Rob Cheung, co-founder and CEO, said. “We created Substrate to enable developers create AI-integrated systems intuitively, in the same way they create any other program: by connecting discrete semantic tasks to automate work. The tools we’ve developed will allow any organization, not just tech behemoths, to run dozens of branching ML models in a single request.”
“We chose to partner with Substrate because we knew they would enable us to quickly leverage the capabilities of modern ML models without dedicating significant engineering resources to the task,” stated Jairaj Sethi, CTO of Substack. “We’ve integrated Substrate in our internal systems to categorize and recommend content as well as creator-facing tools, and we’re excited to continue exploring new use cases.”
“Over the past year, we’ve explored various tools to power course recommendations on Maven,” stated Maven CTO Shreyans Bhansali. “Substrate has been a breath of fresh air because their product is so clearly focused on providing the simplest possible developer experience.”
Substrate is an API for modular AI. Create fast multi-step AI applications by combining optimized models, colocated file and vector storage, a code interpreter, and logical control flow.
