VAST Data

“I can forget that it even exists.” That is how Nic L. at Texas Advanced Computing Center (TACC) describes the shift after moving critical HPC and AI workloads to VAST. For a supercomputing team, the file system should not dictate when users run jobs, when admins perform upgrades or how mixed workloads get isolated. It should absorb hardware failures, handle unpredictable I/O, rebalance as systems expand and keep researchers moving. At TACC, VAST supports the mixed reality of modern research computing: simulation, AI training, preprocessing pipelines, repeated model and environment loading, and high-concurrency access patterns running together on the same platform. Nicole Hemsoth Prickett shares how TACC stopped treating I/O instability as inevitable: https://lnkd.in/eUXwHkHu

Scott Howard explains why the VAST architecture is more than just an eye-catching diagram. It’s a sophisticated, shared-everything design that can handle 11 terabytes per second and makes multi-protocol access seem effortless. From indestructible objects to snapshots with zero performance impact, this session covers the tools that actually make an admin’s life easier: https://lnkd.in/gqkr22_G Read more here: https://lnkd.in/e7bvwdr8

Video has become one of the most important data sources in public safety. The challenge is turning that footage into real-time understanding. At MTX - Milipol TechX in Singapore, VAST joined NCS to highlight how the VAST AI OS can help transform video from passive footage into a reliable input for real-time understanding – helping teams see what is happening, reason over context and act faster. And across the broader APJ region, partners like Firmus Technologies are advancing another critical piece of the story: sovereign AI cloud infrastructure with the performance, security and control required for national-scale AI strategy. Hear from VAST APJ leader Sunil Chavan and NCS Group CEO Sam Liew.

Listen to insights from Leidos’ Josh Salmanson and Rob Linger, and VAST Data’s Jeremiah Hinrichs, as they examine the evolving Federal IT landscape. Their discussion focuses on the transition toward massively parallel GPU environments, which are currently enhancing cyber defense and government mission operations. Learn how the partnership between Leidos and VAST Federal is leveraging unified data architectures to accelerate cyber threat responses: https://lnkd.in/g-kQ3_dJ

The model is the easy part now. The hard part is everything around it. Keeping checkpoints from blocking training. Holding a vector index consistent at a trillion rows. Isolating tenants on shared GPU infrastructure without noisy neighbors. Answering who is trashing my system right now in seconds, not hours. Closing the loop from a robot at the edge back to the AI factory by morning. AI used to be a training job. It is now a continuous system. Data preparation, streaming pipelines, inference, KV cache, search, observability, agents, all running at once on the same infrastructure. That is what separates a demo from a system. And it is where most "AI infrastructure" stories quietly fall apart. VAST CTO Alon Horev and Field CTO Andy Pernsteiner on what it actually takes: https://lnkd.in/ecWFncv9

Building a modern AI cloud often feels like trying to assemble furniture without the manual. Most providers are stuck managing 13 different copies of the same data while their expensive GPUs sit idle. The industry is officially moving past the simple training era and heading straight into a world of agent swarms and massive inference demands. Jason Vallery breaks down the eight essential stages of the AI pipeline. You will discover how to scale capacity without being held back by throughput and why a global data namespace changes everything. Stop settling for many different storage systems when you only need one unified platform. Watch the full session and see how the top GPU clouds stay dominant: https://lnkd.in/gb6UGY_7

A video archive can hold millions of hours of footage and still be nearly impossible to understand. Not because the data is missing. Because the meaning is buried across motion, speech, sound, scene changes, timing and sequence – and most systems were never designed to preserve those relationships at scale. TwelveLabs solves the model problem with video-native AI that can search and reason across multimodal signals. VAST solves the infrastructure problem around it: keeping massive video libraries, metadata, embeddings and derived intelligence accessible, governed and continuously available in customer-managed environments. TwelveLabs’ Maninder Saini explains why next-generation video intelligence needs a new architecture for scale, continuity and control: https://lnkd.in/eEECY2CN

Stop managing silos. Start managing outcomes. The VAST DataBase bridges the gap between the efficiency of a warehouse and the scale of a lake. From native vector search to integrated event streaming, the VAST AI OS provides the foundation for the thinking machine. Listen to Fouad Teban talk about High-Performance Analytics with VAST DataBase: https://lnkd.in/gWF5nkxu

NFS isn't the problem. Your architecture is. VAST Data is breaking the laws of physics that traditionally hold back GPU clusters. No re-architecting applications. No manual tuning. Just high-performance I/O that stays smooth even when patterns get messy. Watch the lab results and see how VAST keeps the lights on at exascale: https://lnkd.in/gacz4d42

Eight sessions. One complete picture. Most data teams operate across multiple systems: warehouses, lakes, streaming layers, vector databases, and ML infrastructure. Each introduces its own pipelines, latency, and operational overhead. Data is constantly moved, transformed, and duplicated just to keep systems in sync. By the time data is ready, it is already stale. This webinar series examines what changes when analytics and AI are built on a unified architecture instead of stitched-together systems: https://lnkd.in/gXy_st6p