PostHeaderSection

Symbol

Global NavBar

PostContentSection

Global Footer

Blog Detail Page

Staff Writer

AstraZeneca accelerated drug discovery and medical imaging by running advanced life sciences models on TensorWave’s AMD Instinct cloud, cutting training times up to 49%.

Accelerating Life Sciences with TensorWave: How AstraZeneca Improved Model Training on AMD Instinct MI300X GPUs

AI is reshaping drug discovery, medical imaging, and life sciences research at an unprecedented pace. But breakthroughs don’t just come from better algorithms; they come from the compute platforms that make next-generation models possible.

That’s why we’re proud to share that AstraZeneca’s recent study on advanced life sciences models was powered by TensorWave, an AI cloud built exclusively on AMD Instinct™ GPUs.

Read the original article from AMD Silo AI here


From NVIDIA to AMD Instinct on TensorWave

In collaboration with AstraZeneca, the AMD Silo AI team successfully ported three life sciences models - REINVENT4, SemlaFlow, and SwinUNETR - to run on AMD Instinct™ MI300X GPUs.

Previously, these workloads ran in AstraZeneca’s environment based on NVIDIA GPUs. By moving to TensorWave’s AMD-powered infrastructure, the team tapped into higher memory bandwidth, seamless ROCm™ integration, and a modern GPU architecture designed for the most demanding AI workloads.

The result? Faster training times, smoother workflows, and effortless scaling for R&D efforts in drug discovery and medical imaging.


Key Model Improvements

 * SemlaFlow (3D molecular structure generation): Optimizations reduced training time by 49%, accelerating the path to new drug candidates.
 * REINVENT4 (de novo molecular design): Speed improved by an average of 41% across four configurations, making molecule generation more efficient.
 * SwinUNETR (3D medical image segmentation): Training time reduced by up to 1.8x, enabling faster tumor detection in CT and MRI scans.

These improvements were achieved with no code changes required. The ROCm open software stack acted as a drop-in replacement for CUDA, allowing AstraZeneca’s PyTorch-based workflows to run out-of-the-box on TensorWave’s AMD Instinct GPU cloud.


Why TensorWave Matters

TensorWave is the only AI cloud built exclusively on AMD Instinct GPUs. That means:

 * Performance without compromise: High-bandwidth memory and next-gen GPU architecture purpose-built for large AI models.
 * Open ecosystem: ROCm integration ensures developers avoid lock-in and maintain flexibility across frameworks.
 * Enterprise scale: From healthcare to finance to autonomous systems, TensorWave enables R&D teams to go from prototype to production at speed.

As Dr. Hannes Loeffler, Machine Learning Researcher at AstraZeneca, put it:

“We were able to achieve these results without any significant changes to our software. Our results showed substantial improvements in the speed of training.”


Looking Ahead

Building on this success, AstraZeneca, AMD, and TensorWave are exploring deeper opportunities to co-create the next generation of AI-driven drug development workflows.

But the impact goes far beyond life sciences. The same tools and infrastructure powering molecular design and medical imaging can accelerate breakthroughs in automotive, robotics, and climate modeling.

At TensorWave, our mission is simple: power the next wave of AI innovation. AstraZeneca’s results are proof that with the right infrastructure, science moves faster.

👉 Ready to build your next breakthrough on AMD Instinct GPUs? Get started with TensorWave.


About TensorWave

TensorWave is the AI cloud purpose-built for performance. Powered exclusively by Instinct™ Series GPUs, we deliver high-bandwidth, memory-optimized infrastructure that scales with your most demanding models—training or inference.

Ready to get started? Connect with a Sales Engineer.

Accelerating Life Sciences with TensorWave: How AstraZeneca Improved Model Training on AMD Instinct MI300X GPUs

Product

Solutions

Resources

Company

Product

Resources

Solutions

Company