DDN Boosts AI Inference Speed, Cuts Costs with New Data Platform Upgrades
Event summary
- DDN announced new AI data platform releases to accelerate inference and lower cost per token.
- Infinia platform achieves up to 27x faster KV cache loading with NVIDIA Dynamo integration.
- EXAScaler enables multi-tenant training infrastructure with API-driven management.
- Solutions run on NVIDIA BlueField-4 DPUs, reducing latency and power consumption.
- General availability expected by summer 2026.
The big picture
As AI transitions from experimentation to production, data infrastructure becomes a critical revenue driver. DDN's upgrades position it to capitalize on the growing demand for efficient, scalable AI factories. The company's focus on cost reduction and multi-tenancy aligns with industry trends toward shared GPU infrastructure and operational efficiency. The partnership with NVIDIA underscores the strategic importance of DPU-accelerated data services in modern AI deployments.
What we're watching
- Adoption Pace
- How quickly AI factories will integrate DDN's multi-tenant solutions into production environments.
- Cost Competitiveness
- Whether DDN can sustain its cost-per-token advantages against competitors.
- Partnership Synergies
- The extent to which NVIDIA and Oracle integrations drive customer adoption.
Related topics