Protein Design Gets an AI Upgrade

NSF awards $32M to fuel AI-driven protein design, strengthening US food, chemical, and materials systems.

21 Aug 2025

The US National Science Foundation has launched a $32mn programme to expand the use of artificial intelligence in protein design, highlighting its growing role in reshaping food and industrial supply chains.

The NSF Use-Inspired Acceleration of Protein Design (USPRD) initiative will fund five teams applying machine learning to areas ranging from sustainable materials to biomass upcycling. The goal is to shorten the process of protein engineering, which traditionally relies on years of laboratory testing. AI models can now predict protein structures within weeks, allowing researchers to move more rapidly from concept to application.

Among the projects, Arzeda is creating enzymes that transform agricultural waste into high-value proteins; Koliber Biosciences is using AI to optimise microbial fermentation; and Novozymes is developing new nutritional compounds such as human milk oligosaccharides. Each approach depends on algorithms capable of exploring vast design spaces beyond human capacity.

Analysts see this as a critical shift in biotechnology. "AI is no longer just a research tool, it is the engine of bio-innovation," said one industry observer, noting that algorithmic design enables a scale and speed unattainable through conventional methods.

Technical challenges remain, including access to advanced computing infrastructure and the task of scaling digital discoveries into affordable commercial products. Startups in particular face barriers in securing the necessary computational resources.

The NSF funding also signals Washington's intention to compete globally. Governments in Europe and Asia have launched similar programmes, seeking to capture strategic advantages in digital biology. By backing AI-led protein design, the US aims to establish leadership in an area that could underpin future food security, chemical production and materials science.

If these efforts succeed, AI-based modelling could become a standard platform technology for the bioeconomy, redefining how proteins are discovered and deployed across multiple industries.