Lila's Closed-Loop Science Factories

The core economic bet is that owning the lab lets Lila sell outcomes, capacity, and data instead of just software seats. A software only AI company can charge for predictions, but Lila can also charge for running the experiment itself, which pushes more revenue through each customer program. That same setup also creates proprietary results data from every robotic run, which can make the models better and increase margins as factory utilization rises.

• Lila is building closed loop AI Science Factories where models design experiments, robots run them, sensors capture the results, and the system learns from the output. That requires real lab buildout and equipment spend up front, which explains why the company launched with $200 million in seed funding to build labs and infrastructure, not just hire software engineers.
  1 sacra 2 lila 3 acs
• This is different from a typical software layer in drug discovery and different from a traditional CRO. A CRO usually bills people and lab time to execute customer designed work. Lila aims to compress that workflow into an automated system that can run many more experiments faster, so it can capture both the intelligence margin and the execution margin in one product.
  2 lila
• The planned lab as a service model matters because it turns a lumpy project business into something closer to infrastructure revenue. Instead of paying only for one discovery program, customers could buy recurring access to automated experimental capacity, which lifts utilization of expensive lab assets and can make gross margins look more like a scaled platform than a services shop.
  2 lila

If this model works, AI for science will shift from selling copilots for researchers to selling autonomous experimental throughput. The winners will be the companies that keep their labs busy, compound proprietary data fastest, and turn fixed lab costs into a high volume discovery engine across biotech, chemistry, and materials science.