Lintes boosts AI data center links with optical interconnect push

Lintes is a reminder that the race to AGI is as much about plumbing as it is about models. As hyperscalers push toward ever-larger AI clusters, the bottleneck increasingly lives in high-speed interconnects—how you move data between GPUs, racks, and data centers without melting down latency or power budgets. By leaning into 6.4T co-packaged copper and optics sockets and automating their production, Lintes is positioning itself as one of the quiet enablers of that scaled infrastructure.

This is also another data point in the ‘AI supply chain bifurcation’ story. A Taiwan-listed interconnect specialist deepening its optical capabilities and leveraging a precision manufacturing subsidiary strengthens the non‑US, non‑Chinese ecosystem of critical AI hardware vendors. For cloud providers and model labs worried about single‑supplier risk in networking, that diversification matters. It also hints at where margin pools may emerge as GPUs commoditize: in the specialized, high‑complexity components that bind AI systems together.

From an AGI perspective, better and cheaper optical interconnects lower the effective cost of scaling clusters and make it easier to experiment with more tightly coupled, high‑bandwidth training setups. That doesn’t guarantee qualitatively new capabilities, but it shaves friction off the path to larger, more interactive model swarms.