When it comes to building cutting-edge technology companies, Erik Bernhardsson's journey with Modal exemplifies the mix of passion, technical ingenuity, and adaptability required to succeed. From tackling infrastructure challenges in the data and AI space to navigating the nuanced dynamics of scaling a business, Modal’s story provides a masterclass in modern tech entrepreneurship.
“I start out by saying [Modal is] infrastructure for data, AI, and machine learning teams. We run a lot of infrastructure in the cloud—GPUs, CPUs—and make it easy to use through a Python SDK. Customers like Suno, which does AI-generated music, use Modal at pretty large scales.”
“I quit my job during the pandemic… I started thinking a lot about the data, AI, and machine learning space. It’s always been frustrating to work with from an infrastructure point of view. I thought I’d just build a better version of Luigi… but then realized, to get workflow scheduling right, you have to nail code execution first.”
“This started as a personal project during the pandemic. My kids were running around screaming in the background, and I was writing code 12 hours a day. I loved geeking out—it was a return to IC work after managing large teams.”
“I partnered with Akshat, who I didn’t know well initially, but we hit it off. He moved to New York, and we started hacking together. It’s not advice I’d give to everyone, but it worked for us.”
“We invested heavily in infrastructure—our own container runtime, file system, etc. A year and a half in, we had zero revenue and zero customers. Then Gen.AI exploded, and stable diffusion became a killer app. People started coming to us because we had GPUs and a great platform for scalable compute.”
“I think it’s good to listen to customers, but in the first year, we didn’t. We had a strong point of view about the infrastructure the world lacked, shaped by my decade of experience.”
“Scaling meant solving unique challenges like managing GPU capacity and optimizing for real-time AI applications. We aggregated GPU capacity across regions and vendors to avoid being the ‘WeWork of GPUs.’”
“Multi-tenancy and resource pooling have been key for us. Hundreds or thousands of de-correlated users smooth out workloads, making capacity planning easier.”
“We focus on fast feedback loops, minimal configuration (no YAML), composable SDK design, and great documentation. To me, everything should be programmable.”
“We built for machine learning engineers who love Python and want to write code, not for low-code solutions or prompts. That focus has shaped our approach to developer experience.”
“Good people attract more good people. Early hires included exceptional engineers from competitive programming backgrounds. We also leveraged my network from Spotify and other places.”
“Modal operated for years without traditional product or go-to-market teams. Engineers designed products for engineers, but as we matured, we started layering on sales and marketing.”
“2025 goals include 3-5x revenue growth, expanding into enterprise segments, and building for new use cases like distributed training. We aim to materialize the vision of an end-to-end platform for machine learning development.”
.svg)