Kove's Software-Defined Memory Boosts AI Inference Workloads

Sep 11, 2025
Kove has announced benchmark results showing that its software-defined memory solution can handle AI inference workloads up to five times larger than local DRAM, with reduced latency.

Kove has announced benchmark results demonstrating that its software-defined memory solution, Kove:SDM™, can handle AI inference workloads up to five times larger than local DRAM, with reduced latency. This announcement was made during CEO John Overton's keynote at the AI Infra Summit 2025, announced in a press release.

Kove:SDM™ allows for dynamic memory allocation across servers, creating larger elastic memory pools that perform like local DRAM. This innovation addresses the memory bottleneck in AI inference, enabling faster processing and reducing energy consumption. Benchmarks conducted on Oracle Cloud Infrastructure showed significant performance improvements for Redis and Valkey, two widely used engines in AI inference.

The results indicate that Kove:SDM™ can accelerate AI inference by eliminating key-value cache evictions and redundant GPU computations, potentially saving enterprises millions annually. The solution is available now and can be deployed without changes to existing applications or code, running on any x86 hardware supported by Linux.

We hope you enjoyed this article

Consider subscribing to one of our newsletters like Silicon Brief or Daily AI Brief.

Also, consider following us on social media:

AI Chips & Datacenters AI Brief AI Brief (X)
Free newsletter

Silicon Brief

Weekly coverage of AI hardware developments including chips, GPUs, cloud platforms, and data center technology.

AI Token Webinar

Whitepaper

Tensordyne Napier: What If One Rack Could Do the Work of Nine?

Tensordyne

This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.

Read more

You may also like

Free, six days a week

Daily AI Brief: the AI news that matters, in your inbox.