Silicon IP for the zettabyte era

More data per die.

Sunalink develops latent-compression silicon IP that multiplies the effective capacity of storage and memory—up to 15:1, depending on modality.

Start a conversation
01

Increase effective density

02

Reduce bytes in flight

03

Deliver compute-ready data

The representation layer

The next density gain is representational.

Sunalink is developing silicon IP that changes how data is represented within storage and memory systems. The architecture is designed to preserve useful information in substantially less physical data—raising effective capacity while reducing movement across the system.

Built for integration into existing storage and memory roadmaps, Sunalink creates a new systems-level lever across capacity, bandwidth, and energy.

No new media. No new storage tier. More useful data from the same silicon.

Engineering approach

Built for integration.

Sunalink’s representation-layer architecture is designed as licensable IP for storage controllers, memory subsystems, and AI data infrastructure.

01Architecture

Hardware-efficient execution

Designed around the constraints of storage-class silicon.

02System

System-level impact

Less physical data to store, move, and prepare for compute.

03Modality

Modality-extensible

A common architectural foundation, with performance dependent on data type and operating target.

Integration model

IP that disappears into the data path.

Designed for integration across storage controllers, memory subsystems, and AI data infrastructure. Host systems retain familiar interfaces while the representation layer changes how much useful data each physical bit can carry.

Storage controllers · Memory subsystems · Computational storage · AI data infrastructure

Partner engagement

Start a conversation.

Currently engaging with a select group of storage, memory, controller, and data-infrastructure partners.

contact@sunalink.com