– VantAI will leverage its proprietary large scale structural data and NVIDIA H100 GPUs on Google Cloud to enable a step-change in accuracy for all-atom protein interaction prediction –

– VantAI reports a 3x speedup in training times by switching from NVIDIA A100 to H100 GPUs, leveraging Google Cloud's A3 virtual machines for large-scale distributed training –

NEW YORK: #AI--VantAI, a pioneer in generative AI-enabled drug discovery, is working with Google Cloud to power large scale distributed training and further improve internal operations. This collaboration is set to provide VantAI with NVIDIA H100 GPUs, a move that VantAI believes will help accelerate the training of geometric foundation models for making protein-protein interactions programmable.

VantAI's mission to exponentially expand the space of druggable targets by making protein interactions programmable takes a strong leap forward with this collaboration. Through its proprietary structural proteomics platform, VantAI is on the path to generating structural data for the protein fabric making up entire cells and tissues - a shift in scale that will help unlock significantly larger datasets than previously accessible. By combining this large proprietary dataset with the computational power of NVIDIA H100 GPUs, VantAI will continue to build industry leading all-atom protein interaction foundation models, key to its mission to unlock a new chapter in medicine driven by Proximity Modulators (ProMods), innovative molecules designed to modulate protein interactions. VantAI accomplished a threefold increase in training speeds by transitioning from NVIDIA A100 to H100 GPUs for the same number of nodes already and expects to orchestrate the computational resources efficiently as VantAI uses Google Kubernetes Engine (GKE), leveraging Google Cloud’s A3 virtual machines for the large scale distributed training, as a part of the collaboration with Google Cloud.

"The ability to rationally design molecules that modulate protein interactions has been held back by the lack of data needed to train foundation-scale models,” said Luca Naef, CTO and co-founder of VantAI. “Now that we’ve finally removed this bottleneck through our proprietary data platform, the obvious next step was to apply scale. We are thrilled to work together with Google Cloud in this incredible problem space and believe our collaboration will lay the basis for significant scientific breakthroughs in the near future."

“Google Cloud is committed to providing the most open ecosystem and helping customers accelerate their digital transformations,” said Shweta Maniar, global director of Life Science Strategy & Solutions, Google Cloud. “Through this new collaboration, VantAI will use Google Cloud’s purpose-built infrastructure to power large scale distributed training that can improve operations and create real-world value for businesses.”

About VantAI

VantAI is at the forefront of applying artificial intelligence technologies to drug discovery. With a cross-disciplinary team of world-class scientists and engineers, VantAI is uniquely suited to power a best-in-class solution for proximity modulation, where the nuance of protein-protein interactions, such as those in protein degradation, has outsized influence on drug success. VantAI has collaborated with and achieved milestones for numerous leading biopharma partners on difficult to drug targets leading to the launch of new development programs with exciting properties. For more information, please visit www.vant.ai.

Fonte: Business Wire