Research

AI engineering at the quantum scale

A fully solvated COVID spike protein with half a million atoms simulated at quantum accuracy on a single GPU.

This scale is only possible with Orbital's research.

Research

We live in a material world. The chips powering the computer you're reading this on are advanced materials called semiconductors, crafted from the silicon that gave Silicon Valley – the epicenter of technology – its name. The battery in your electric car, the composites in spaceships, and the LEDs in your screen are also advanced materials. New advanced materials were the cornerstone of 20th-century prosperity – and could power the 21st century.

But discovering and scaling up new materials is hard. Not only do their remarkable properties derive from complex quantum interactions that are too difficult for traditional, bottom-up computer simulations to solve accurately, but translating a promising material from the lab to industrial-scale production presents significant hurdles. Designing efficient hardware, optimizing process parameters, and ensuring consistent quality at scale are all complex challenges. This means that materials science has not benefited from the exponential improvements that computers have brought to other fields. A materials lab today would feel largely unchanged from one 30 years ago.

Read More
Theme 1

Pretraining & generative models for advanced materials

This research program aims to unlock the extraordinary emergent properties of large materials datasets, mirroring the transformative impact of scale seen in natural language processing, to revolutionize materials discovery. By developing pretraining strategies that harness the complexity of vast chemical and physical data, we seek to create models capable of predicting and designing novel materials with unprecedented accuracy and efficiency, fundamentally changing how we approach materials innovation.
Theme 2

Large-scale simulation

This research program tackles the critical challenge of simulating metallic systems and quantum properties at scales between 10,000 and 100,000 atoms—regimes where traditional quantum methods fail to scale. Many key experimental phenomena, from phase transformations to defect dynamics, emerge at these large scales, yet remain out of reach for conventional quantum simulations. By using accelerated AI simulations, these critical phenomena - often the things you care most about - become within reach.

Contact us

By submitting this form, you agree to our Privacy Policy
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.