Venture Capital Influx and Interconnect Technology Development for Artificial Intelligence Semiconductors

A successful Series C venture capital funding round yielding $145 million at a corporate valuation of $1 billion was announced on Wednesday by Eliyan, a Santa Clara, California-based semiconductor startup dedicated to mitigating data transfer bottlenecks between artificial intelligence processing units within data center environments. Strategic backing for the financing round was provided by prominent technology entities, including Cisco Systems, optical technology developer Lumentum, and early institutional investors associated with networking pioneer Mellanox prior to its acquisition by Nvidia.

The advanced interconnect technology under development by the startup is targeted toward a expanding cohort of technology enterprises engaged in designing custom artificial intelligence microprocessors aimed at competing against dominant graphics processing units produced by Nvidia and Advanced Micro Devices. Across the semiconductor industry, a fundamental operational bottleneck has been encountered, wherein data processing speeds within modern computational units significantly outpace the rates at which data can be transmitted or received, thereby causing high-cost processing hardware to remain idle during intensive computing workloads.

The operational challenges confronting modern data centers were contextualized by Eliyan Chief Executive Officer and co-founder Ramin Farjadrad, by whom it was observed that despite historical industry efforts to engineer faster microprocessors and graphics units, current computational utilization rates remain restricted to approximately 30 to 40 percent. This inefficiency was attributed primarily to severe data transfer constraints, a critical structural limitation that the company’s proprietary interconnect architecture was designed to resolve.

Within the broader market landscape, data transmission bottlenecks were previously addressed by Nvidia through its $6.8 billion acquisition of Mellanox in 2019. Similarly, custom hardware divisions at major hyperscale cloud providers, including Alphabet’s Google and Amazon’s cloud computing division, have mitigated connectivity constraints through strategic partnerships with specialized networking firms such as Broadcom and Marvell Technology to secure access to proprietary transmission protocols.

An independent technological alternative is intended to be provided by Eliyan through a dual commercial strategy involving the licensing of its intellectual property alongside the production of modular semiconductor components known as chiplets. These modular components are designed to be directly integrated into custom processor architectures by third-party chip designers seeking enhanced data bandwidth.

Commercialization timelines and financial expectations were outlined by Patrick Soheili, chief strategy and business officer at Eliyan, by whom it was projected that initial shipments of specialized chiplets would commence within the current year. Following initial revenues in the low millions of dollars recorded in 2025, commercial sales are forecasted by executive management to reach several hundred million dollars by the conclusion of 2027.

The $145 million Series C financing was led by Seligman Ventures, with participation from both new and returning institutional investors alongside strategic partners Cisco and Lumentum. As a condition of the investment agreement, a seat on Eliyan’s board of directors will be assumed by Umesh Padval, managing partner at Seligman Ventures, who previously served on the board of directors at Mellanox prior to its sale to Nvidia.

The substantial capital injection secured by Eliyan highlights the growing importance of high-bandwidth memory and interconnect technologies within the global semiconductor supply chain. As artificial intelligence workloads become increasingly complex and data-intensive, the ability to rapidly transfer information between distributed processing chips has emerged as a primary determinant of system efficiency. By offering scalable chiplet solutions and licensable interconnect intellectual property, specialized hardware startups are positioning themselves to address critical infrastructure constraints while enabling broader competition in the rapidly expanding artificial intelligence hardware market.