Lila Sciences has not publicly confirmed a new mega-round, but its verified $550 million funding base still shows how quickly investors are moving toward automated science.
Lila Sciences is being watched because it sits at the point where two expensive markets now overlap: frontier AI and physical laboratory research. That is enough to make venture investors pay attention, even before the company proves that autonomous labs can produce discoveries with commercial value.
The Cambridge, Massachusetts company, spun out of Flagship Pioneering and led by co-founder and Chief Executive Officer Geoffrey von Maltzahn, has built its pitch around what it calls AI Science Factories. These are not ordinary software tools for scientists. They combine AI models, robotics, sensors and experimental feedback loops so the system can form hypotheses, run tests, learn from the results and start again.
That ambition is why the valuation conversation around Lila matters. The latest verified financing shows a company already priced as one of the most closely watched AI-for-science startups. According to Reuters, Lila raised a $115 million Series A extension in October 2025 from investors including Nvidia's venture arm, bringing its total Series A to $350 million, overall capital raised to $550 million and valuation to more than $1.3 billion.
Those figures are still large enough to make the point. Venture capital is no longer treating AI in science as a narrow drug discovery software category. The bigger wager is that automated experimentation can become infrastructure, sitting underneath new medicines, materials, chemicals and industrial processes. If that proves true, the lab itself becomes a platform.
This is the part investors care about most. Many AI companies are fighting over the same raw materials: compute, distribution, talent and access to high-quality training data. Lila is making a different claim. Proprietary experimental data, generated inside closed-loop labs, could become the scarce asset. If a company can run large numbers of physical experiments and feed those outcomes back into its models, it may build a learning advantage that cannot be copied by scraping the internet.
That is also why the company has attracted an unusually broad group of backers. Lila's disclosed investors include Flagship Pioneering, Braidwell, Collective Global, General Catalyst, ARK Venture Fund, March Capital and Nvidia's NVentures. The mix matters because it brings together biotech company creation, growth capital and AI infrastructure. A few years ago, a business combining wet labs, robotics and foundation models might have looked too complex for most venture portfolios. Now it fits the center of the AI investment story.
The hard-science AI market is getting crowded
Lila is not alone. Periodic Labs, founded by former OpenAI and DeepMind researchers, has drawn attention for applying AI and automation to materials science, with Bloomberg previously reporting talks around a roughly $7 billion valuation. Alphabet's Isomorphic Labs is using AI for drug design, while Recursion has spent years building industrialized biological data generation into its platform. Different companies are attacking different parts of the market, but the direction is the same. AI is moving from analysis toward execution.
That shift is important because science gives AI something consumer software often does not: measurable failure. A material works or it does not. A catalyst performs or it does not. A biological system responds or it does not. That makes closed-loop experimentation attractive, because models can be judged by what they cause in the real world, not only by how persuasive their outputs look on a screen.
The risk is just as clear. Autonomous labs are expensive to build and hard to operate. They require elite AI talent, scientific talent, robotics systems, instruments, facilities and quality controls. If the platform does not produce useful discoveries, capital intensity becomes a weakness rather than a moat. Investors can price a story quickly, but customers and partners will still need evidence.
There is also the question of who captures the value. If Lila helps produce a better material, drug candidate or chemical process, commercialization may depend on partners that handle clinical trials, manufacturing, regulation or industrial deployment. That can be powerful if the economics are structured well, but it is not the same as selling high-margin software subscriptions. The business model has to turn faster discovery into durable revenue.
Still, Lila's rise shows where the next phase of AI investing is heading. The first wave was about generating text, images and code. The next wave is trying to connect models to robots, instruments and physical feedback. For Lila and its peers, the test is no longer whether investors believe the story. They already do. The next test is whether automated science can produce repeated, valuable results that make the funding look like patient capital rather than expensive optimism.
Also read: Uber cuts its HR ranks as tech chases leaner operations • Red Hat npm breach shows startup build pipelines need tighter controls • Ideogram 4.0 turns open weights into a startup weapon
