DDN

Sovereign AI is not about shutting the world out. It is about staying in control. At RAISE Summit, Mohsen Moazami, Vice Chair at DDN, drew on his experience leading international efforts to show how countries can maintain control of their AI infrastructure while still participating in the global AI economy. From Saudi Arabia exporting AI tokens to building sovereign compute stacks, his message was clear: sovereignty and openness can coexist. Hear Mohsen’s perspective on what real sovereignty means for the future of AI Infrastructure → bit.ly/4wE1Zwh

Everyone is optimizing the watt. Almost nobody is auditing the wait. NVIDIA's 800 VDC architecture exists to strip conversion stages between the grid and the accelerator. At gigawatt scale, a few percentage points of loss is worth redesigning an entire industry to recover. We are fighting that hard for single digits in the power path. Inside the same building, GPUs sit waiting for data. A stalled GPU draws nearly full power, occupies capacity, depreciates on schedule, and produces nothing. That loss is not measured in single digits. It is also not measured at all. The GPU shows as allocated. The invoice shows as paid. Nothing on any report says the work never happened. Ask your infrastructure team what percentage of last month's GPU-hours produced computation and what percentage was spent waiting. If nobody can answer, that is the answer.

Enterprise AI moves faster when your data is ready to keep pace with the AI factory. Join Andrew Murphy, Sr. Director Product Management at DDN, Pompey Nagra, Product & Ecosystem Marketing at Solidigm, and Michael Ang, Director, Storage Solutions at Supermicro, for a discussion on building AI data platforms that can scale with enterprise demands. They’ll cover how modern AI Infrastructure can streamline ingestion, normalization, vectorization, RAG, visual search, and document processing. 🗓️ August 27 ⏰ 10am PT / 1pm ET ➡️ Register now: bit.ly/4gsvEUa #AI #AIInfrastructure #EnterpriseAI

When your job makes it to the SLURM queue and your data platform didn’t. Unlock your potential with DDN.

Massive CapEx may build the models, but insight is what creates the value. On the Mainstage at RAISE Summit, Alex Bouzari, CEO and Co-founder of DDN, shared why powerful foundation models will continue to demand significant CapEx, while the real opportunity lies in transforming data into intelligence. That intelligence can drive better products, stronger services, and greater security for enterprises, consumers, and nations. Hear more from Alex Bouzari on what will define the next phase of AI innovation → bit.ly/3SBfkHE

GPU idle time is often invisible until it shows up in your cloud bill. Curious how many teams are tracking it proactively. 

Real-world agentic AI doesn’t behave like a clean benchmark. Your inference testing shouldn’t either. 🚫 Eric Wright, Consulting Solutions Architect at DDN, explores how “golden traces” can recreate the interleaved requests, shifting prefix-cache behavior, and varied context lengths found in production agentic and RAG workloads. By testing AI infrastructure against realistic, repeatable traces, teams can better understand GPU utilization, KV cache performance, token throughput, and cost per token. Read the blog to learn more 👇

Enterprise AI moves faster when your data is ready to keep pace with the AI factory. Join Andrew Murphy, Sr. Director Product Management at DDN, Pompey Nagra, Product & Ecosystem Marketing at Solidigm, and Michael Ang, Director, Storage Solutions at Supermicro, for a discussion on building AI data platforms that can scale with enterprise demands. They’ll cover how modern AI Infrastructure can streamline ingestion, normalization, vectorization, RAG, visual search, and document processing. 🗓️ August 27 ⏰ 10am PT / 1pm ET ➡️ Register now: bit.ly/4gsvEUa #AI #AIInfrastructure #EnterpriseAI

Most GPUs in the world are waiting. Waiting on data. Waiting on an agent, a loop, a graph. Waiting in a queue behind a job that shouldn't be running. The industry's answer has been to buy more GPUs. It is the most expensive way to solve an idle-GPU problem ever created. We measure this across more than a million GPUs. The gap between what an AI factory can do and what it actually does is 10x to 100x more. Same silicon. Same power. Same capital. Nobody publishes their utilization. That should tell you something. The next gigawatt will not fix what the last gigawatt is wasting. And no single layer fixes it, because the chip, the data path, the scheduler and the workload all have to agree. We've worked alongside NVIDIA on this for a decade. If you're building at the model layer, the cloud layer, or the energy layer and you think utilization is the real bottleneck, I want to talk. Comment and I'll reach out.

Saturating your GPUs is only the first step toward real AI efficiency. At PEARC, Morris Skupinsky, Solutions Architect at DDN, explained how repeated prompts and workloads can force GPUs to recompute the same tokens again and again. By caching reusable tokens with LM cache, AI Infrastructure can reduce redundant work and keep valuable GPU cycles focused on new computation. Learn how smarter caching can help teams improve GPU efficiency, accelerate AI workloads, and get more value from their AI Infrastructure. → bit.ly/4xwI96T