Lila Sciences raised $200 million in a seed round that closed in March 2025. The round was led by Flagship Pioneering, with participation from Abu Dhabi Investment Authority subsidiary, General Catalyst, March Capital, ARK Venture Fund, Altitude Life Science Ventures, Blue Horizon Advisors, State of Michigan Retirement System, and Modi Ventures.

This is among the largest seed rounds in the AI-driven scientific discovery sector, driven by investor interest in the company's vertically integrated model that combines large language models with autonomous laboratory infrastructure.

The company originated from Flagship Pioneering's incubation process, which follows the venture studio's framework of developing companies around foundational scientific concepts.

Lila Sciences develops AI Science Factories, fully automated laboratory facilities that integrate large foundation-scale scientific language models with general-purpose lab robots, sensors, and data infrastructure. Each facility functions as a closed-loop system in which AI agents generate hypotheses, design experimental protocols, operate laboratory equipment, capture multimodal data streams, and update the models with results in real time.

The process begins when a human scientist or partner company uploads a research objective, such as identifying an antibody that binds to a specific target or discovering a catalyst for hydrogen electrolysis. Lila's domain-specific language model analyzes proprietary and public datasets, including research papers, patents, and molecular databases, to produce ranked candidate solutions.

An orchestrator agent translates each candidate into executable laboratory instructions, detailing parameters such as volumes, temperatures, and analytical endpoints. Robotic units, including liquid handlers, automated bioreactors, high-throughput reactors, and analytical instruments, autonomously execute these protocols continuously with minimal human oversight.

Results are parsed and reintegrated into the system, enabling the model to learn error margins and physical constraints while fine-tuning itself on the new experimental data. The updated model then generates the next set of experiments, establishing a self-improving cycle capable of conducting hundreds of experimental iterations per week.

Lila Sciences operates a vertically integrated model that combines proprietary AI software with physical laboratory infrastructure. Instead of licensing software or selling data, the company owns and manages the entire stack, from AI models to robotic equipment, offering a distinct solution in the scientific discovery market.

The business model focuses on providing accelerated R&D services to pharmaceutical companies, materials science firms, and other research-intensive organizations. Customers use the platform to address specific discovery challenges, such as drug development or catalyst design, without investing in their own AI or automation capabilities.

Revenue generation follows a project-based model in which customers pay for individual discovery programs. The company also plans to introduce lab-as-a-service offerings, enabling organizations to purchase access to automated experimental capacity on a subscription or usage basis.

Vertical integration results in higher capital requirements but also the potential for higher margins compared to software-only businesses. By controlling the entire experimental process, Lila captures value from both AI-driven insights and the physical execution of experiments, while building proprietary datasets that enhance model performance over time.

The company's approach contrasts with traditional contract research organizations by leveraging automation to prioritize speed and scale, potentially completing discovery programs in weeks rather than months or years.

Recursion Pharma operates BioHive-2, which features GPU compute power and robotized cell-imaging pipelines, with a focus on small-molecule drugs and phenotypic screening. The company has developed multimodal datasets and maintains partnerships with NVIDIA. Its expertise is more concentrated in computational biology than in autonomous laboratory operations.

Insilico Medicine has raised funding to expand automated laboratory capabilities and advance drug candidates into clinical trials. The company competes in therapeutics discovery, spanning target identification, generative chemistry, and robotics integration.

Ginkgo Bioworks employs a foundry model for cell engineering and has acquired AI capabilities through companies such as Reverie Labs. Its focus on synthetic biology and bio-foundry throughput overlaps with Lila's biological applications, though the two employ different technological approaches.

Emerald Cloud Lab and Strateos provide remote access to automated laboratory equipment, competing in the infrastructure segment of scientific automation. These platforms enable researchers to use high-end instruments but do not incorporate the AI-driven hypothesis generation and experimental design that Lila emphasizes.

LabGenius integrates machine learning with robotic protein engineering through its EVA platform, concentrating on antibody and nanobody optimization. While its closed-loop methodology is similar to Lila's, LabGenius operates within a narrower domain.

Isomorphic Labs utilizes Alphabet's computational resources and AlphaFold technology for structure-based drug discovery, securing partnerships with pharmaceutical companies valued in the billions. However, the company depends on external partners for wet lab validation rather than running autonomous facilities.

DeepMind's AI capabilities in protein folding and biological systems present potential competitive pressure, though its focus remains on foundational research rather than commercial discovery services.

Lila operates across biological therapeutics, genetic medicine, materials science, and chemical catalysis using a unified platform. Structuring these capabilities into distinct commercial offerings could increase addressable markets in biopharma, clean energy, and advanced materials, industries that collectively allocate hundreds of billions of dollars annually to R&D.

The platform's cross-domain functionality enables engagement with industrial customers outside traditional pharmaceutical sectors, including chemical manufacturers, energy companies, and materials producers seeking innovative compounds and processes.

The company plans to use recent funding to establish additional commercial AI Science Factories, enabling the sale of automated experimental capacity through subscription or usage-based models. This approach would diversify the business model, moving beyond project-based discovery services to include ongoing laboratory infrastructure offerings.

Expanding laboratory facilities near customer clusters in major research hubs could address data sovereignty requirements and reduce sample logistics costs. This strategy could unlock markets in Europe and Asia, where local presence is often preferred or mandated.

Lila's ability to generate differentiated assets, such as novel antibodies and advanced materials, creates potential for capturing downstream value through licensing or spin-out arrangements. By retaining ownership stakes in promising compounds or materials, the company could extend its revenue streams beyond discovery services.

Collaborations with academic institutions, including initiatives like the AI residency program, could provide access to advanced research challenges while fostering relationships with emerging industry talent. These partnerships may expand both the customer base and the talent pool.

Capital intensity: The business model requires substantial upfront investment in AI infrastructure and physical laboratory equipment, resulting in high fixed costs and extended payback periods. Scaling operations across multiple facilities necessitates considerable capital deployment before revenue generation, exposing the company to risks associated with funding market conditions.

Regulatory validation: Autonomous laboratory systems used in pharmaceutical and materials discovery must comply with complex regulatory requirements, which could delay adoption or incur significant compliance costs. Adjustments to regulatory frameworks governing AI-driven research may limit the company's ability to access certain markets or necessitate costly operational changes.

Technical execution: Integrating large language models with autonomous laboratory equipment involves considerable technical complexity. Failures in AI reasoning or physical automation could undermine experimental validity. The closed-loop learning system relies on accurate data capture and processing across diverse scientific domains, introducing multiple potential failure points.