Edgecore Networks Unveils Nexvec Solution for Enterprise AI

Edgecore Networks Corp. a wholly owned subsidiary of Accton Technology Corporation, announced the launch of Nexvec, a turnkey solution purpose-built for Enterprise AI. Featuring disaggregated networking and composable compute and storage, Nexvec delivers an integrated user experience and simplified lifecycle management that accelerates AI adoption for enterprise.

Nexvec represents Edgecore’s vision of bringing the spirit of openness and fast-paced innovation into Enterprise AI infrastructure, enabling organizations to drive tangible business outcomes through AI technologies.

Click to enlarge

Solution brings integrated user experiences with certified offerings:

• Edgecore’s open networking portfolio
• Software Defined Composable Compute architecture powered by our partner Liqid, Inc., enabling dynamic allocation of GPU, Memory and Storage
• Newly introduced fabric controller, Nous, delivering full Day 0 through Day 2 lifecycle management across the stack

“We’ve been a pioneer in open networking for over a decade,” said Jun Shi, CEO, Accton Technology Group, parent company of Edgecore Networks. “Now, with Nexvec, we’re moving beyond networking to deliver a full-stack solution, integrating disaggregated networking and composable compute to simplify Enterprise AI adoption. This effort reflects our ongoing commitment to open architecture and ecosystem collaboration.”

Nexvec is designed to be deployable, scalable, manageable, and programmable, meeting the infrastructure demands of next-gen Enterprise AI, particularly for inference, reasoning, and emerging Agentic AI use cases. These technologies aim to retool enterprise workflows, automate cross-functional processes, and enable contextual reasoning by unlocking the value of enterprise data.

Click to enlarge

From a scale-up compute perspective, Nexvec supports open standards such as PCIe and CXL to create a compute fabric for GPU pooling and memory pooling, enabling on-demand dynamic resource allocation for multi-tenant enterprise deployments. The solution integrates Liqid Matrix orchestration, supporting leading frameworks including VMware, SLURM, and Kubernetes for efficient AI workload placement. By scaling up fewer compute engines through hardware resource abstraction, Nexvec solution reduces the space and power overhead typically associated with multiple high-end servers for performant utilization.

“Edgecore’s adoption of Liqid’s software-defined composable solution is a powerful validation of the market’s need to dynamically manage and allocate critical resources like GPUs and memory,” said Sumit Puri, co-founder and president, Liqid. “Together, we’re delivering next-generation AI infrastructure as a total solution to help enterprises accelerate AI adoption.”

For scale-out architectures, Nexvec leverages the company’s portfolio of open networking switches, battle-tested across Hyperscalers, Neocloud, and Enterprise AI deployments. It features rich Ethernet-based AI fabrics built on Broadcom’s Tomahawk (e.g. TH4, TH5) and Trident lines, as well as VoQ-based Jericho3/Ramon3 architectures with congestion-free routing for AI workloads.

Edgecore’s recently announced Pure-Play SONiC strategy further reinforces our ongoing commitment to open ecosystems and continued support for SONiC.

As part of the Nexvec solution, the company continues to support 3rd-party NOS options and offer a certification and integration program to ensure interoperability for quality assurance. Additionally, the newly introduced Nous fabric controller provides an integrated, managed profile for the entire solution — delivering Day 0 through Day 2 lifecycle management through automation for the AI fabric.

Nexvec will launch with limited availability, with full general availability targeted by the end of this year.

Resource:
Blog: From Building Blocks to Solutions: A Turnkey Open Infrastructure Solution for Enterprise AI