July 29, 2026, (Inside AI) — Chip interconnect startup Eliyan has secured $145 million in Series C funding at a $1 billion valuation, the company announced Wednesday, aiming to solve the data transfer logjam that leaves AI processors idle. The round was led by Seligman Ventures with strategic investments from Cisco Systems and Lumentum, alongside existing backers.
The capital injection positions Eliyan to ship its first chiplets this year, targeting a market where custom AI chip designers from hyperscalers to startups struggle with the same bottleneck: processors that compute faster than they can communicate. The problem leaves expensive silicon waiting for data, dragging down utilization rates.
"People have tried to build faster compute, faster processors and GPUs," Ramin Farjadrad, Eliyan CEO and co-founder, told Reuters. "Today we are at the point that we are maybe 30-40% of the GPUs is being used, and is being mainly limited by how fast they can receive the data to process. We're solving that problem."
The Mellanox Playbook, Reimagined #
Eliyan's approach echoes Nvidia's $6.8 billion acquisition of Mellanox in 2019, a deal that gave the GPU giant in-house networking technology to link thousands of chips efficiently. But where Nvidia's solution is proprietary, Eliyan pitches an independent alternative for the growing roster of firms designing their own AI processors to compete with Nvidia and Advanced Micro Devices.
The startup licenses its interconnect technology and sells physical chiplets—small, specialized silicon blocks that custom chip designers integrate into larger packages. This modular approach mirrors industry trends toward heterogeneous integration, where different functions are mixed on a single substrate. A recent paper on chiplet-based architectures highlights how such designs can improve yield and performance but require advanced interconnect standards.
Eliyan's timing capitalizes on a market shift. Alphabet's Google and Amazon's cloud unit already work with Broadcom and Marvell Technology for networking technology, but Eliyan claims its approach delivers higher bandwidth and lower latency. Patrick Soheili, chief strategy and business officer, said revenue would climb from "low millions" of dollars in 2025 to "hundreds of millions" by 2027, with initial chiplet shipments starting this year.
Boardroom Signal and Market Validation #
The round includes a notable board appointment: Umesh Padval, managing partner at Seligman Ventures and a former Mellanox board member during its sale to Nvidia, will join Eliyan's board. His presence signals deep connectivity expertise and investor confidence in Eliyan's trajectory.
Eliyan's technology addresses a fundamental scaling challenge. As AI models grow, they require more chips working in parallel, but traditional interconnects create bottlenecks that leave processors underutilized. The company's proprietary interconnect architecture promises to reduce this idle time, directly improving total cost of ownership for data centers.
The funding arrives amid a surge in custom AI chip development. Companies from Meta to Microsoft are designing their own silicon to optimize workloads and reduce reliance on Nvidia. Eliyan's independent model could appeal to these players, offering a networking layer without vendor lock-in. The startup's ability to deliver on its 2027 revenue forecast will depend on execution and the pace of adoption for chiplet-based designs.