Innophore needed to predict protein-ligand interactions and off-target effects much earlier in drug discovery, but the scale of biomolecular computation made traditional CPU workflows too slow and too expensive for practical screening.
Reported impact
Innophore cut protein-protein binding affinity predictions from 2,000 CPU hours to 15 minutes at about 5% of the original cost, ran millions of molecular computations per day, evaluated more than 15 million SARS-CoV-2 variants, and reached 5 million off-target predictions per second.
Innophore built its Cavitomix platform on NVIDIA GPUs, NVIDIA RAPIDS, NVIDIA BioNeMo, and AWS to encode proteins and nucleic acids into digital catalophores that can be searched and compared at scale. The company trained convolutional neural networks on thousands of protein-binding-site catalophores using NVIDIA T4 and V100 GPUs, while cloud elasticity on AWS let it burst through 270,000 GPU hours of work instead of maintaining the hardware itself. The platform also generated atomistic structures for more than 42,000 human proteins through AlphaFold, OpenFold, and ESMFold so Innophore could assemble a proteome-wide structural library for virtual off-target screening. With that stack, Cavitomix can run millions of molecular computations in seconds, compare binding sites across biomolecules, and feed drug-safety and efficacy predictions back to researchers before they commit to slower wet-lab work.
Protein and nucleic-acid structures are generated and encoded into catalophores, processed on AWS GPU infrastructure, and then compared across a large structural library so Innophore can predict off-target effects and binding affinity before wet-lab validation.
Innophore has spent 270,000 NVIDIA GPU hours on AWS.
Innophore's AI algorithms reduced computation time considerably, from 2,000 CPU hours down to 15 minutes per protein-protein binding affinity prediction, and just 5% of the original cost.
NVIDIA boosted our performance so that we can run five million off-target predictions per second.
This enabled Innophore to evaluate more than 15 million new variants of SARS-CoV-2.
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
The Organization JSON-LD lists addressLocality 'Graz', and addressCountry 'AT', and numberOfEmployees value 18.
Related
NVIDIA Sample Sep 22, 2026 DELFI Diagnostics DELFI Diagnostics needed to speed genomic analysis for cancer screening and therapeutic monitoring, but CPU-bound alignment and somatic variant-calling steps created turnaround-time and cost bottlenecks that limited how… 10–20 business days DeepSomatic · MutectCaller · NVIDIA Parabricks NVIDIA Sample Jul 10, 2026 Astellas Pharma Inc. Astellas needed to shorten the slow handoff between experimental scientists and computational chemists in early drug discovery so more researchers could use structure prediction and affinity screening without waiting on… More than 50 regular users and over 2,000 total users within four months NVIDIA BioNeMo · NVIDIA Boltz-2 NIM · NVIDIA DGX H100/H200 NVIDIA Sample Apr 23, 2026 Biofy Hospitals needed pathogen identification and antibiotic-resistance analysis much faster than five-day culture and antibiogram workflows, but CPU-bound genomic processing and limited AI expertise made it hard to turn… 100,000+ synthetic DNA sequences generated by the biology model Abby · H100 Tensor Core GPUs · NVIDIA CUDA-accelerated libraries 