MegaRobo raises nearly $200 million to plug AI agents into real labs, making experiments callable by software
Synopsis
MegaRobo announced a new round of nearly $200 million, with the Oman Investment Authority participating, existing investors Asia Investment Capital, Granite Asia and Sinovation Ventures continuing their support, and CMB International joining. The funding will go to core technology, global expansion of its AI for Science business, and infrastructure that lets AI operate in increasingly complex physical environments. Its own One Intelligence Stack combines models and agents, data, workflows, software, scientific instruments, robotics, automation and real-world feedback into one reusable technology base, with science as the first proving ground.
Interpretation
MegaRobo has raised nearly $200 million in new funding to build infrastructure that lets AI act in the physical world, with science as its first application. The company's funding previously came from multiple rounds totaling about $600 million, including a Series C of nearly $300 million co-led by Goldman Sachs Alternatives, Asia Investment Capital and Granite Asia, with Pavilion Capital participating. The company says it has supplied products and technology to more than 500 customers, including global pharmaceutical companies, drug-discovery organizations, biotech firms and research institutions, covering stages such as early research, target and mechanism identification, drug screening and candidate optimization.
What supports this shift is the One Intelligence Stack, which connects intelligence with execution rather than being a set of isolated AI applications or automation products. The first generation of AI learned mainly from digital information such as text, images, video and code; the next generation will increasingly learn by interacting with the physical world. The architecture is organized around five steps, think, design, execute, observe and learn, and brings models and agents, data, workflows, software, scientific instruments, robotics, automation and real-world feedback into one reusable base.
Through Megalaxy Lab, real-world experimentation is becoming programmable: researchers and AI agents can configure experiments, send them to automated laboratory systems, track execution and retrieve structured, machine-readable results. Those results can feed directly into the next cycle of scientific reasoning, shifting experimental capacity from lab work organized around individual projects toward capacity that can be standardized, automated and requested on demand. Megalaxy Lab has served more than 200 customers; in Poreva Open, a global AI protein-design challenge with partners including Tencent Cloud and Twist Bioscience, Megalaxy Lab will carry out laboratory validation of more than 10,000 designed protein sequences.
Perspective
The demonstrated scope is scientific discovery and pharmaceutical R&D: more than 500 customers across early research, target and mechanism identification, drug screening and candidate optimization, more than 200 Megalaxy Lab customers, and a Poreva Open plan to validate more than 10,000 designed sequences in the laboratory. The intended readers are researchers and engineers following AI and automated experimental infrastructure.
The company's judgment that the next generation of AI will learn through physical-world interaction is forward-looking; Poreva Open's laboratory validation is described as forthcoming and no results are reported; customer counts and supported stages are company statements.
