Designing & Building Data Center AI Infrastructure at Scale (original) (raw)

89,000+

GPUs Deployed & Managed

3.3+ Billion

Hours of GPU Runtime

Penguin Solutions designed, built, deployed, and now manages one of Korea’s largest GPU clusters, consisting of over 1,000 NVIDIA Blackwell GPUs integrated into a single cluster.
Shell powers its sustainable high-performance data centers with Penguin’s high-performance computing (HPC) solutions, including immersion cooling.
Penguin Solutions designed, built, and deployed the infrastructure to support the Georgia Tech AI Makerspace.
Penguin Solutions deploys NextSilicon accelerator technology as part of the Vanguard program at Sandia National Labs.
Accelerate time to value by basing system architectures on a proven set of designs that have been validated at scale in numerous production deployments.
Achieve high rates of system stability with our in-factory experts who integrate and validate all components of the compute cluster including rack integration, network configuration, and burn-in testing.
Drive on-site installations with coordination of data center staff, data storage partners, and infrastructure cooling providers—and utilize ICE ClusterWare software to validate production readiness.
Assure production readiness and change management by working with a certified NVIDIA DGX Managed Services provider, the offers a full set of end-to-end services.

“Penguin Solutions demonstrated a deep understanding of our technical requirements, translating them into a sophisticated infrastructure environment that meets and exceeds expectations.”

“It takes a village to do AI well, it takes an infrastructure, it takes a data center, and it takes experts. And, I think in that regard, having Georgia Tech, NVIDIA, and Penguin—that’s what it takes.”

“After a thorough RFP process, it was clear early on that Penguin was the right partner for us. Not only do they have the technical expertise and decades of experience, but they’re able to move very fast.”

Woman in data center with tablet

OriginAI®

OriginAI® is an AI factory infrastructure solution built on proven, pre-defined AI architectures that can scale from hundreds to over 16,000 GPU clusters.

OriginAI integrates these validated technologies with Penguin’s intelligent, intuitive cluster management software and expert services for designing, building, deploying, and managing AI infrastructure at scale.

ClusterWare on laptop screen on desk

ICE ClusterWare™

Simplify the deployment and management of AI clusters to realize greater productivity at speed.

With ICE ClusterWare™, bare-metal hardware, network, and software resources are transformed into high-performance cluster environments, reducing administration complexity and optimizing resource availability.

Data center room aisle

Delivering NVIDIA DGX-Ready Managed Services

Penguin Solutions has designed and deployed large NVIDIA DGX clusters with high-speed NVIDIA InfiniBand networking and optimized storage.

We have deep expertise and relationships with most storage vendors which allows us to provide bespoke solutions for every customer.

ztc Endurance

Stratus ztC Endurance®

Stratus ztC Endurance® is an innovative family of computing platforms that enables intelligent, predictive fault tolerance and 99.99999% compute platform availability.

The platform combines built-in fault tolerance, proactive health monitoring, and serviceability by OT or IT, all while meeting your cybersecurity requirements.

ztc Edge

Stratus ztC Edge®

Stratus ztC Edge® is a secure, rugged, highly automated computing platform that improves productivity, increases operational efficiency, and reduces downtime risk at the edge of corporate networks.

Its self-protecting and self-monitoring features drastically reduce unplanned downtime and ensure continuous availability of business-critical applications.

everRun

Stratus everRun®

Stratus everRun® is a software solution that pairs two servers via virtualization to create protected and replicated virtual machines (VMs) within a single operating environment, ensuring your applications run without interruption or data loss.

Stratus everRun accelerates time to revenue by transforming your applications into continuously available solutions with customized availability.

CXL expansion memory

Introducing the New Family of CXL® Add-in-Cards (AICs)

Compute Express Link (CXL) enables data centers, cloud services, and HPC providers to expand memory for intensive computing easily and cost-effectively.

ZDIMM