Amazon Web Services Inc. is launching a technical collaboration with Orbital Materials Inc., a startup that uses artificial intelligence to discover new materials.

The companies detailed the initiative today against the backdrop of the cloud giant’s AWS re:Invent conference. It’s described as a multiyear collaboration. The goal, the companies said, is to develop technologies and materials that will make it possible to operate data centers in a more sustainable manner. 

“Our partnership with AWS will accelerate the deployment of our advanced technologies for data center decarbonization and efficiency,” said Orbital Materials co-founder and Chief Executive Officer Jonathan Godwin.

The company raised a $16 million funding round in February that included the participation of Toyota Ventures. About a month ago, it secured an investment of undisclosed size from Nvidia Corp.’s NVentures. Orbital Materials says that its AI-centric approach to developing new materials makes it possible to carry out research more efficiently than with legacy methods.

It’s pursuing its work using an internally developed AI system called LINUS. According to the company, researchers can enter a set of properties that a new material should possess and the system automatically comes up with potential formulas.

After LINUS proposes a material, Orbital Materials studies its suitability using a second AI system called Orb. The latter software can map out a material’s energy properties, the behavior of its molecules and other features. The company says Orb is more accurate and five times faster than competing AI models.

If a proposed material shows promise, it synthesizes it at its Princeton, New Jersey research facility. The first product it has developed through this process is a carbon removal technology based on a “proprietary active material.” The company says that its researchers have improved the material’s performance by a factor of ten since February.

As part of the collaboration with AWS, Orbital Materials will develop materials and technologies for “data center-integrated carbon removal.” It will also work on products that can help optimize chip cooling and water utilization. Cooling accounts for a significant percentage of data centers’ water and power usage. 

The partnership will see Orbital Materials collaborate with AWS in several areas.

The startup made Orb, the AI model it uses to evaluate new materials, available under an open-source license in September. AWS will make the model available to its customers through AWS Marketplace and Amazon SageMaker Jumpstart. The latter offering provides preconfigured AI software that can be deployed with a few clicks.

AWS sees customers applying Orb lends to materials research projects in several fields. Semiconductors, batteries and electronics are among the products that can be developed with the help of the software.

Orbital Materials, for its part, will develop its future AI models using the Amazon SageMaker HyperPod service. The offering provides preconfigured AI clusters that remove the need for customers to set up machine learning infrastructure manually. Additionally, SageMaker HyperPod automates cluster maintenance tasks such as recovering from hardware failures.

The startup will also explore opportunities to incorporate AWS Trainium into its AI development workflows. Trainium is a series of AWS-developed chips optimized to run AI training workloads. It’s available alongside a suite of development tools, AWS Neuron, that companies can use to deploy their AI models on the cloud giant’s silicon. 

Image: Unsplash