Eliyan announced $50 million in strategic investment on Jan. 28 from AMD, Arm, Coherent and Meta, with Samsung Catalyst Fund and Intel Capital also participating. The company said the funding would support commercialization of its interconnect technology for AI and high-performance computing systems.

The important part of the story is not a newly disclosed unicorn valuation; Eliyan’s primary announcement did not provide one. It is the list of strategic backers and the technical problem they are funding. Faster AI processors are useful only when data can move to, from and between them quickly enough.

The system is larger than the accelerator

Modern AI hardware increasingly combines specialized chiplets, memory and networking components rather than relying on one monolithic processor. That approach can improve manufacturing economics and allow designers to mix the best components for a task. It also makes the connections among those components central to overall performance.

Eliyan develops physical-interface technology and chiplets for die-to-die, chip-to-chip and rack-scale connectivity. The company says its NuLink and NuGear products are designed to address memory and input-output limits while using less power. Those performance claims require customer validation, but the bottleneck is well established: a processor can sit underused when memory or interconnect bandwidth cannot keep it supplied.

The company previously announced a $40 million Series A in 2022 and a $60 million Series B in 2024. Together with the 2026 strategic financing, its publicly announced rounds total at least $150 million. That cumulative history may be the source of late reports describing a single $145 million raise.

Infrastructure competition is spreading outward

The AI-chip race is often reduced to GPUs, yet system performance depends on a wider supply chain: high-bandwidth memory, advanced packaging, optical links, switches, power delivery, cooling and software. Improvements in any one layer can expose the next constraint.

For enterprise buyers, that means infrastructure comparisons should move beyond the model of processor and its theoretical speed. Useful measures include performance on the actual workload, power per result, latency, data-movement cost, software support and how easily capacity can scale across a rack.

Eliyan is a small company in a market shaped by large semiconductor vendors and emerging standards. Its strategic investor list nevertheless shows that the connections between chips are becoming important enough for the industry’s biggest participants to fund directly.

Sources for editorial review

Editorial note: The supplied brief described a new $145 million round at a $1 billion valuation. Eliyan’s primary materials support a $50 million Jan. 28 round and do not disclose that valuation. This draft corrects the amount and date and should not be backdated to the unsupported late-July aggregation.

Drafting note: This draft was prepared with AI assistance from the linked source material and requires author review, independent fact-checking and final editorial approval before publication.