What challenges did Schrödinger face before AI implementation?

Drug discovery simulations are compute-intensive and bursty, and legacy on-premises clusters could take up to three weeks for a single calculation. Schrödinger needed a virtual-lab platform to speed iteration and expand datasets for research programs such as lymphoma research.

What was the impact of the AI solution?

Computations were reduced from weeks to hours. The platform enabled a 4,000x larger dataset for lymphoma research. The approach reduced physically built molecules in the lab by 97%. In one program, the field was narrowed from billions of molecules to 78 candidate molecules. The company said the platform can help move programs from years to around 10 months for human readiness.

Source

GCP

Schrödinger: Physics-based molecular simulation platform on Compute Engine, Cloud GPUs, BigQuery

Use case typeDrug discoveryUpdated Jun 13, 2026

Schrödinger uses Google Cloud as the foundation for its physics-based molecular simulation platform to accelerate drug discovery and reduce the cost and time of lab work. The company runs compute-intensive simulation workloads that previously took up to three weeks on on-premises clusters; with Google Cloud, it can scale horizontally across more GPUs and CPUs to finish computations within hours. Schrödinger added BigQuery as a data hub for hundreds of billions of molecules, reagents, and iterations, and uses machine learning to narrow simulation results toward the most promising candidate molecules.

Organization: Schrödinger
Industry: Pharma
Location: United States
Published: May 2026

Reported outcomes

quantified impactOther quantified impact

−97%quantified impact

Strategic outcomes

Speed & agilityAccelerated simulation turnaroundScale & capacityExpanded research dataset scaleNew product / capabilityBuilt a virtual-lab discovery platformBetter decisions & insightNarrowed candidates to promising molecules

Primary read

Use case focus

Showing 3 of 4

1Drug Discovery
2Scientific Computing
3Simulation Platform

Drug discovery simulations are compute-intensive and bursty, and legacy on-premises clusters could take up to three weeks for a single calculation.
Schrödinger needed a virtual-lab platform to speed iteration and expand datasets for research programs such as lymphoma research.

Schrödinger built a platform on Google Cloud using Compute Engine, Cloud GPUs, and BigQuery.
The platform scales simulation workloads horizontally to use more GPUs and CPUs.
BigQuery serves as a central data hub for hundreds of billions of molecules, reagents, and iterations.
Machine learning is used to narrow the field of candidate molecules from very large datasets to the most promising options.

Technologies

Compute Engine Cloud GPUs BigQuery

Computations were reduced from weeks to hours.
The platform enabled a 4,000x larger dataset for lymphoma research.
The approach reduced physically built molecules in the lab by 97%.
In one program, the field was narrowed from billions of molecules to 78 candidate molecules.
The company said the platform can help move programs from years to around 10 months for human readiness.

Architecture

A virtual-lab molecular simulation platform on Google Cloud that uses Compute Engine and Cloud GPUs for bursty scientific workloads and BigQuery as a large-scale data hub for molecules, reagents, and simulation iterations.

Sources & evidence1

Groundedness: 4/5Type: Customer StoryPublished: May 26, 2026Publisher: Google CloudEvidence: PrimaryConfidence: High

Primary source

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Customer evidence