Stephen Balaban

Because we’re a technical team, we get to build for ourselves and we’re able to spot trends before they happen. One thing that really sets Lambda apart is how differently we’re approaching the business compared to many of our competitors. A lot of them have built their AI offerings around bare metal infrastructure and long-term contracts – three- or five-year deals.

Lambda, on the other hand, is moving toward a usage model based on dollars per million tokens instead of dollars per GPU hour. We’re also launching products like Lambda Chat, which puts us more directly in the space of companies like Perplexity and Anthropic. That’s a very different direction from most of the Neo cloud providers.

I think that shift is only possible because we’re a deeply technical team. If we wanted to, we could pivot into being a full foundation model lab ourselves. But my personal view is that it’s better to leverage open source. You avoid having to amortize R&D and training costs over every API call, and the gap between the best proprietary and open source models is shrinking fast. These models are becoming commoditized.

Our partnership with NVIDIA has been amazing. We’ve worked with them since joining their partner program in 2017, so it’s been over half a decade of close collaboration. Everyone there is incredibly helpful, all the way to the top. That culture starts with Jensen [Huang, NVIDIA’s CEO], who genuinely cares. He’s helped us personally on multiple occasions. As a fellow entrepreneur, I think he enjoys seeing founder-led companies working hard, delighting customers, and deploying NVIDIA’s tech in the world.

We’ve seen similar value from other strategic partners like Supermicro, Pegatron, Wiwynn, and Mitac – these are major OEMs who build servers using NVIDIA chips. We look at where our cash flows are going and talk to those vendors: “You’re excited about AI. How can we work more closely together?” I think it’s a really important mix, and a lot of people probably don’t give strategic investors as much attention as they should.

Lambda is a capital-intensive business, and going public could significantly reduce our weighted average cost of capital by opening access to a much larger pool of capital. Public markets represent around $103 trillion compared to $13 trillion in private equity. That increased demand for our shares would lower both the nominal and economic costs of capital, speeding up fundraising and improving flexibility. Somebody once advised me, “if you can continue to successfully raise in the private markets, then you should stay private as long as you can”. It’s really just that Lambda is such a capital intensive business and growing so fast that it just makes a lot of sense to tap public markets.

Regarding milestones, there’s almost nothing that would stop us from getting to a $1B revenue run rate. The question in my mind is “how do we get to $5-10B”. That requires a laser focus on core execution – providing that direct-line compute. But the point of the business is not revenue – it’s market capitalization. The hardware business with a 1-2x revenue multiple would never achieve the share price we have, because the cloud business trades at a higher multiple. So you will start seeing Lambda go into things like Lambda Chat to bring more of a pure-play software multiple mix into our EV. Can you imagine what this business could look like at $1.B in cloud revenue, a couple hundred million dollars of annual subscription revenue for something like a ChatGPT competitor, based on open source software that’s all running on our own infrastructure? With strong vertical integration and technical sophistication, Lambda is building something fundamentally different from anything else in the space.