Case Study

Sterling Federal SI: Gaudi 3 Reference Architecture

"500,000 pages processed in 8.5 hours -- 48x faster than Xeon."

How Sterling deployed Blockify on Intel Gaudi 3 to create a reference architecture for federal AI operations with classification marking preservation.

500K
Pages in 8.5 Hours
48X
Faster vs Xeon
17.8M
Pages/Month
78X
Accuracy Improvement
Back to Case Studies

Federal AI Challenges

Sterling needed to demonstrate enterprise-scale AI processing for classified environments where data governance, classification markings, and air-gapped operation are non-negotiable.

Blockify on Intel Gaudi 3 accelerators transformed millions of pages into AI-optimized knowledge bases with full RBAC and classification marking preservation (SECRET, TS/SCI, NATO).

Implementation Areas

Data Processing

  • 500K pages in 8.5 hours
  • Classification marking preservation
  • RBAC integration
  • Multi-format ingestion

Security

  • SECRET/TS/SCI support
  • NATO-restricted content
  • 100% on-premise
  • Red Hat OpenShift AI

Gaudi 3 Performance

500,000 pages were processed in 8.5 hours on Gaudi 3 -- a 48x throughput improvement versus Xeon, which would have taken 17+ days for the same workload.

Key Outcomes

  • Throughput: 48x throughput improvement vs Xeon
  • Scale: 17.8 million pages/month on single Gaudi 3 core
  • Classification: Classification markings preserved throughout AI pipeline
  • Accuracy: 78x accuracy improvement for intelligence analysis
Free Case Study

Download the Full Sterling Federal SI Case Study

Get complete details on Gaudi 3 reference architecture for federal AI.

  • 500K pages in 8.5 hours
  • 48x faster than Xeon
  • Classification marking preservation
  • SECRET/TS/SCI/NATO support
  • Red Hat OpenShift AI deployment
Case Study 12 min read 678 KB

Get Your Free Copy

We respect your privacy. No spam, ever.

Enterprise AI for Federal Operations

See how Blockify on Gaudi 3 can transform your operations.

Request a Demo