Google Unveils AI System That Can Predict Protein Structures in Real Time
DeepMind's new AI processes uncharted protein families with 98% accuracy at 40 times the speed of its predecessor, offering hope for faster drug discovery.
Image: GlobalBeat / 2026
Google DeepMind on Wednesday announced a next-generation artificial intelligence system capable of predicting complex protein structures in real time, a development scientists say could accelerate drug discovery by years if not decades.
The system, which builds on the company’s groundbreaking AlphaFold model, can now process previously uncharted protein families with 98 percent accuracy at speeds 40 times faster than its predecessor. Researchers published the findings in the journal Nature.
“We are entering a new era of biological computing,” said the lead researcher at DeepMind’s structural biology division. “What once took a laboratory months can now be modeled in minutes.”
Pharmaceutical companies have already expressed strong interest in licensing the technology. Industry observers estimate that faster protein mapping could reduce the cost of bringing a new drug to market by up to 30 percent.
Critics, however, caution that computational predictions still require laboratory validation before clinical application. Regulatory agencies in the US and EU said they are monitoring the development closely.