Mettle Capital and Peak XV Back AI Science Startup Empirical Machines with $20M

The race to apply artificial intelligence to the physical sciences is accelerating as investors look beyond general-purpose chatbots toward domain-specific breakthroughs. Former OpenAI researcher Shyamal Hitesh Anadkat has capitalized on this shift, securing $20 million for his new startup, Empirical Machines. The round, which values the company at roughly $80 million, marks the maiden investment for Mettle Capital—a new fund launched by former Peak XV Partners executives—who co-invested alongside Peak XV itself.

Empirical Machines is building AI models designed to accelerate scientific discovery by learning directly from real-world experimentation. Rather than merely processing existing research papers, the platform focuses on identifying and testing novel chemical compounds. This approach aims to dramatically reduce the time and cost associated with traditional trial-and-error laboratory research, a bottleneck that has historically slowed advancements in materials science and pharmaceuticals.

Unlike broad foundational models that struggle with the rigid constraints of chemistry and physics, Empirical Machines joins a growing cohort of specialized "AI for science" startups. By combining deep learning with experimental feedback loops, the company is attempting to bridge the gap between digital predictions and physical reality. The involvement of Mettle Capital and Peak XV highlights a growing venture appetite for deep-tech applications in India, signaling that the next wave of AI value will likely come from solving complex industrial and scientific challenges.