Published: Apr 23, 2025
Accelerating Drug Discovery with AI: How 310 AI is Building the Future of Therapeutics

Can AI design life-saving therapeutics? At Beyond CUDA Summit 2025, Kooshiar Azimian, Founder & CEO of 310 AI, delivered one of the most compelling talks of the event—showing how generative AI, supercharged by AMD GPUs, is now tackling biology at scale.
🧠 Protein Design Is the New Frontier
Forget text and image generation—310 AI is pushing the boundaries of generative biology.
- Prompt in, protein out: Their proprietary model turns natural language biological descriptions into complex 3D protein structures.
- Why it matters: Proteins underpin every drug we use today. The therapeutic protein market is worth $300B+, with massive implications for diseases like cancer, diabetes, and more.
Azimian compared today’s stage of protein generation to the GAN days of 2014—early, imperfect, but promising.
🧪 Meet MP4: Molecule Programming, 4th Gen
310 AI’s MP4 model is a generative evolution of AlphaFold. Rather than predicting structure from sequence, it designs proteins end-to-end.
🔢 Training Details:
- 300M+ sequences, 5B+ data points
- Trained on AMD MI250 and MI300X GPUs using ROCm
- Peaked at 200 sequences/second throughput
- Completed over 72K GPU hours in training
⚙️ Engineering wins:
- Switched from Nvidia A100 → H100 → MI250 → MI300X
- ROCm migration took one engineer, one day
- Tools like Cursor enabled fast porting
🔬 Real-World Results: From Code to Clinic
MP4 isn’t just theoretical—it’s producing lab-validated drugs.
In the past 90 days alone, 310 AI generated 96 GLP-1 analogs, a class of promising diabetes and weight loss therapeutics.
Expressed with 84% success in Swiss labs—significantly outperforming historical benchmarks.
Clinical testing begins next month.
🛠️ Why ROCm Matters for AI Biology
Transitioning from Nvidia to AMD is no longer a blocker.
310 AI’s experience shows that modern AI dev tools + ROCm = smooth scaling.
💡 Takeaways:
- Training performance matched or exceeded Nvidia counterparts
- Minimal code refactoring required
- AMD MI300X enables faster iteration at lower cost
🔁 Final Thoughts: AI Isn’t Just Generating Text Anymore—It’s Generating Medicine
Azimian’s closing words hit home:
“If you can train a vision model, you can train a biology model.”
310 AI’s MP4 is proof that the future of drug discovery lies at the intersection of AI, high-performance compute, and domain expertise.
📺 Watch the full session: Accelerating Drug Discovery with Kooshiar Azimian | Beyond CUDA Summit 2025
🚀 Run Efficient Models on AMD GPUs
Deploy your optimized models on TensorWave’s AMD-powered AI cloud—built for training, inference, and experimentation at scale on MI300X and MI325X GPUs.
About TensorWave
TensorWave is the AI and HPC cloud purpose-built for performance. Powered exclusively by AMD 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.