AI could slash drug discovery times from months to days

Drug companies spend years and billions of dollars bringing new medicines to market. Most of that time is wasted on experiments that fail. Now, Persistent Systems says it can compress months of early-stage research into days using AI that simulates molecules before they ever enter a real lab.

The company has built a system called GenMolVS that runs on NVIDIA's AI platform. It creates virtual versions of molecules and tests how they might work as drugs. Only the most promising candidates move to expensive physical experiments.

How does it work?

The GenMolVS system uses large AI models trained specifically on biological and chemical data. These models can predict how molecules will behave in the human body without actually creating them in a lab.

The system works in several steps:

• AI generates new molecular structures that could work as drugs
• Virtual screening tests millions of these molecules against disease targets
• The system ranks candidates based on their likelihood of success
• Researchers get a prioritized list for physical testing

The AI agents can make decisions in real-time, continuously updating their recommendations as new data comes in. This creates what Persistent calls "simulation-led intelligence" that guides researchers toward the most promising experiments.

Why does it matter?

Traditional drug discovery is brutally inefficient. Companies test thousands of compounds that never make it to human trials. Each failed experiment costs time and money that could be spent on better candidates.

Virtual screening changes this equation. Instead of synthesizing molecules to test them, researchers can simulate their properties first. This approach offers several benefits:

• Faster identification of promising drug candidates
• Lower costs in early research phases
• Better success rates in later clinical trials
• More resources available for promising compounds

The technology could be especially valuable for rare diseases, where traditional research economics don't work. If AI can identify good candidates faster and cheaper, companies might pursue treatments they would otherwise ignore.

The context

This partnership reflects a broader shift in pharmaceutical research. Major drug companies are investing heavily in AI-powered discovery tools. The pressure to find new treatments faster has never been higher, especially after the pandemic showed how quickly the industry can move when necessary.

NVIDIA has been positioning itself as the infrastructure provider for this transformation. Its AI chips already power much of the machine learning used in drug discovery. The company's BioNeMo platform specifically targets life sciences applications.

Persistent Systems brings domain expertise in healthcare and life sciences to the partnership. The company has over 26,500 employees across 18 countries and has made AI-driven solutions a core part of its business strategy.

The collaboration also includes plans to use NVIDIA's Nemotron models and NIM microservices to deploy these tools at scale. This suggests the companies are planning for production deployments, not just research experiments.