Company Overview - Nscale is a vertically integrated AI stack provider, offering end-to-end solutions from infrastructure to cloud [1] - The company customizes data centers for customers, optimizing for specific workloads, similar to a hyperscaler approach for private clouds [2][3] - Nscale is building the largest supercomputer cluster with Microsoft in Europe, comprising approximately 23,000 nodes [4] Technology and Services - Nscale supports diverse AI workloads including model training, fine-tuning, and inference, accommodating various parameters [5] - The company embraces Kubernetes and SLURM for orchestration, providing managed services and bare metal as a service [9][10] - Nscale offers an open AI API compatible interface, enabling scaling and deployment of open source or proprietary models, along with fine-tuning services [12] - The platform supports both Nvidia and AMD GPUs, catering to different customer requirements [13] Future Directions - Nscale aims to provide a global fleet management solution, integrating on-premise and public/private cloud solutions for a consistent customer experience [14] - The company plans to further diversify its AI services, focusing on open source systems and enterprise features like fine-grained access controls [15] - Nscale supports the open-source community through Hugging Face, acting as an inference provider [16]

the founding uh quote from Danny Thomas, if you're familiar with Danny Thomas in the 60s was no child should die in the dawn of life. That I think is a mission we can all get behind. It's a it's a statement we can all agree with.And so it's a really amazing organization and as a children's research hospital, we don't really do product endorsements, but we do have a very close collaboration with DDN. They are they are provider of a lot of our equipment. We work with them closely to make sure we're getting th ...

uh but there's a uh you know massive amount of uh proof of concept massive amount of gathering of use cases right uh I was talking to a fortune 500 and they've gathered literally over a thousand use cases but really the challenge is to will those through these thousand use cases extract which ones are the most important and implement them right so uh so the the we have to make it simpler for enterprise to adopt AI. ...

Welcome. Today I'll be showing how the Exosscaler data intelligence platform supports multi-tenency. Exoscaler has always had strong capabilities with regard to the networking it supports and exoscaler multi-tenency takes advantage of this by using VLANs along with X's ability to divide its data space up into secure partitions.We now have the ability to create a secure multi-tenant environment. For additional security, we've also implemented client access controls to prevent unauthorized access to a tenants ...

Product Overview - Google Cloud Manage Luster is a persistent storage offering powered by DDN Xcaler [1] - The product launched four and a half months ago and has good customer traction [1] - It is fully integrated with Google Cloud, including Google Cloud Storage (GCS) and Google Kubernetes Engine (GKE) [2] - The product is compliant and has a 999% SLA [2] Performance and Scalability - The product runs up to 1 terabyte per second of throughput with sub-millisecond latency and millions of IOPS [3] - It scales from 9 terabytes to 8 petabytes [3] Technology - The company partnered with EDM for a mature and reliable product with a rich feature set, choosing Exosscaler as the foundation [1]

St Jude's Mission and Goals - St Jude's mission is to advance cures and means of prevention for pediatric catastrophic diseases [3] - St Jude aims to save lives and improve the lives of children globally, focusing on areas with lower survival rates [6] - St Jude is in the business of knowledge discovery through research [5] Data Growth and Infrastructure - St Jude's storage utilization has doubled in the past four years, reaching 100 petabytes [13][14] - Bio image data is the fastest growing data source, increasing from 12% to 28% of research storage in the last four years [15] - The institution has expanded from three storage systems to eight to address different regulations and improve resilience [17] - A second data center with a live, asynchronously replicated copy of almost all data has been added for disaster recovery and business continuity [17][18] Future Focus Areas - St Jude anticipates doubling its data utilization again in the next four years and is exploring improved tiering strategies [20][21] - The institution is focused on improved sharing and collaboration tools for its global initiative [21][22] - St Jude is exploring the integration of quantum computing with its current systems [23] - The organization is addressing the challenges of balancing data accessibility for artificial intelligence with cost-effective tiering techniques [24][25]

Science at full speed, from single GPU to full data center scale 🦾 Ken Hester, NVIDIA Solution Architects Director, shares how NVIDIA’s full-stack approach, spanning hardware and a deep software stack, powers reference architectures that scale. With MagnumIO and RDMA creating a direct GPU-to-data path, NVIDIA and DDN align tightly to cut overhead and accelerate real research outcomes. Watch to learn how this NVIDIA and DDN collaboration boosts AI Infrastructure performance and take away best practices you c ...

I don't think you can do AI unless you have flexibility on prem in the cloud and cloud the consumption and the utilization needs to be flexible. I have to be able to today. So you know DDN we're offering exa through Google. We have a onep offering with Google.Well if a customer wants to consume and it makes sense for them to consume on Google. If they want to consume through core wave so be it. If they want to do it on prem so be it.So the flexibility I think is very very important. uh and unlike the migrat ...

The final thing that was just announced at GTC is our hyper pod solution. So our hyper pod solution is a core data center building block. So it is something where the GPU servers and the storage servers are integrated and delivered at different scaling sizes.So basically t-shirt sizing that you can get from super micro that's a combination of DDN storage infinia storage a cluster of GPUs and this is and this is available in t-arch sizing so that you could go from four GPU servers to 32 GPU servers and we wo ...

Company Growth & Strategy - Super Micro has experienced significant growth, evolving from a sub $1 billion company to a much larger entity due to the rise of AI [1] - Super Micro differentiates itself from tier one vendors by being 100% software-defined, aligning with the trend of startups in the server and storage industries [4] - Super Micro utilizes a building block approach, integrating from motherboard to rack level, and now data center building blocks for AI solutions [5][6][7] Product & Technology Focus - Super Micro offers a diverse storage server portfolio, including large capacity 3.5" drives and high-performance flash storage solutions [9][10] - Super Micro is developing advanced cooling solutions like DLC2, using warm water to improve power efficiency for high TDP components [13] - Super Micro emphasizes the importance of proper cabling and integration for delivering optimal performance at scale, addressing challenges with building block solutions [11][12] Partnership with DDN - Super Micro partners with DDN to deliver solutions, reselling DDN's Exoscale for traditional HPC and integrating with DDN's Infinia platform, a software-defined solution [14][15] - Super Micro and DDN have a strong relationship, cemented by a large AI cluster project [7] - Super Micro and DDN offer a hyper-pod solution, a core data center building block with integrated GPU and Infinia storage, available in various sizes from 4 to 32 GPU servers [16][17]