Eliyan, which aims to license tech and make physical “chiplets” to ease AI chip data transfer bottlenecks, raised a $145M Series C at a $1B valuation
Eliyan, a startup that aims to ease the bottlenecks in transferring data between AI chips in data centers …
Context & Ripple Effects
Eliyan has progressed from a $40M Series A for its NuLink interconnect technology to a $60M Series B focused on improving AI-chip performance. The new round is a larger capital commitment to the same chiplet-connectivity thesis.
The story sits alongside funding for other AI-hardware specialists, including a space- and power-saving AI-server chip developer. It matters because data movement between components is becoming a distinct target for AI infrastructure investment, rather than an incidental part of processor design.
First-order effects
- Eliyan receives $145M to advance and license its interconnect technology and produce physical chiplets, while the $1B valuation gives the company a clearer financial platform for that strategy.
- Customers and prospective partners evaluating chiplet-based AI systems gain a better-capitalized specialist supplier focused on data-transfer bottlenecks.
Second-order effects
- Other AI-hardware startups pursuing performance, power, or server-density gains face a higher funding and execution benchmark as capital continues to reach component-level infrastructure companies.
- Chip designers and system builders may place greater emphasis on interconnect choices when assessing AI hardware, because faster compute components alone do not resolve data-transfer constraints.
Third-order effects
- If such funding continues, AI-chip competition could shift further from monolithic processors toward differentiated component ecosystems in which interconnect IP and chiplets are strategic layers.
- That shift would broaden the AI infrastructure capital cycle beyond headline accelerators, distributing value across the suppliers that address bottlenecks in moving data between compute components.
The trend: AI infrastructure investment is moving deeper into the hardware stack, funding specialized technologies that address system-level bottlenecks around compute, memory, power, and data movement.