Lila Sciences: наука как данные для ИИ — автономная лаборатория за $550 млн

Startup Lila Sciences revealed on July 16, 2026, in the Latent Space podcast how its autonomous laboratory works: AI controls robots and equipment, independently designs and runs experiments, and has already generated more than 10 trillion experimentally verified tokens for training scientific models.

Why science, not the internet?

Lila Sciences is betting that the scientific method is the last untapped internet-scale dataset. The company's logic is simple: text from the web for training large models is nearly exhausted, while verified experiments are a practically endless source of new data.

Training here is built as reinforcement learning, where nature itself acts as the verifier: the model proposes a hypothesis, a robot runs the experiment, and the result from the physical world confirms or refutes the prediction. Through this, Lila has accumulated more than 10 trillion tokens of scientific reasoning grounded in real measurements rather than internet text.

How the robot lab works

Lila's laboratory is AI-controlled robotics layered on top of ordinary scientific equipment, not an assembly line built for speed. The company emphasizes that it optimizes not for throughput but for flexibility and generalizability: a single setup must be able to run the widest range of different experiments.

• Samples ride on pucks that float on magnetically levitating tracks
• Vision-language models operate legacy instruments, "looking" at them the way a human would
• Experiment orchestration works like a SLURM queue — modeled on a computing cluster
• Gas sorption measurement was reengineered to run roughly 2500 times faster
• The model proposed platinum-group-metal-free electrocatalysts that outperformed previous results
• Within six months, the team reached in vivo data on CAR-T therapy in primates

"The lab of the future should feel like a data center," is how

Andy Beam, CTO, and Rafa Gómez-Bombarelli, Chief Physical Sciences Officer at Lila Sciences, put it in the Latent Space episode.

How much has the company raised?

Lila Sciences has raised a total of about $550 million and reached a valuation of roughly $1.3 billion. The company grew out of the Flagship Pioneering venture incubator — the same one behind Moderna.

The first round, worth $200 million, closed in March 2025, led by Flagship Pioneering. In October 2025, as Reuters reported, Lila raised another $115 million in an extension of its Series A round with participation from Nvidia's venture arm — bringing Series A to $350 million and the valuation to ~$1.3 billion. The company is led by CEO Geoffrey von Maltzahn, with geneticist George Church as chief scientific officer.

What this means

If Lila's bet pays off, the AI race will shift from parsing the internet to producing proprietary scientific data: the winner will be whoever has not more servers full of text, but more robots running real experiments. For now, this is backed by individual results — faster measurements and new catalysts — but not yet a universal "scientific superintelligence."

Frequently asked questions

What is Lila Sciences?

Lila Sciences is a startup out of the Flagship Pioneering incubator building autonomous AI laboratories for "scientific superintelligence": AI designs and runs experiments through robotic hands. The company was founded by Geoffrey von Maltzahn together with a team that includes Andy Beam and Rafa Gómez-Bombarelli.

How much has Lila Sciences raised?

A total of about $550 million: a $200 million seed round (March 2025) and a $115 million Series A extension with participation from Nvidia (October 2025). The company's valuation is roughly $1.3 billion.

Why does Lila need robots and a lab?

Robots are needed to generate verified experimental data for training models: the company considers experiments, not internet text, to be the primary inexhaustible source of data for AI. This has already produced more than 10 trillion verified scientific tokens.