Lead with the surprising fact: for long-haul routes, a low-Earth-orbit constellation can beat terrestrial fiber on latency, because light travels faster through vacuum than through glass and the path can be more direct. But that advantage is only real if the inter-satellite links themselves are not adding delay. Latency in space is won or lost on the backbone.
Intel's grant US12278690B2 (inventors including Stephen T. Palermo and Valerie J. Parker), classified in H04B 7/18584, H04B 7/18519, and H04B 7/18521 (satellite-communications inter-satellite links and routing), claims ultra-low-latency inter-satellite links. That a major silicon and networking company is patenting here signals the inter-satellite layer is now serious networking infrastructure.
The mechanism targets the delay budget end to end: how links are formed, how traffic is routed across the mesh, and how processing and queuing are minimized so packets cross the constellation with as little added latency as possible. Shaving microseconds is the kind of optimization that matters only once the basic links work, which is itself a marker of maturity.
It is the latest layer atop the inter-satellite progression we traced through 2023, Huawei's link acquisition, Ciena's express optical mesh, and now Intel pushing latency down. The subfield matured in stages: first see and acquire the link, then network it efficiently, now make it fast enough for the most demanding applications.
The competitive stakes are real. Financial trading, real-time control, and interactive applications care about milliseconds, and a constellation that can credibly offer lower latency than fiber on certain routes has a genuine commercial wedge. The patent tells you Intel is engineering for that wedge. Whether the deployed links hit the latency targets is, as always, settled in orbit, not in the claim.