An international team led by Japan's NICT, with Sumitomo Electric Industries, transmitted data at 1.02 petabits per second over 1,808 kilometres. That is a capacity-distance product of 1.86 exabits per second × km, the highest ever recorded, and the fastest long-distance transmission achieved in any optical fibre.
Headline capacity records in this field arrive every year or two and most of them do not matter. This one does, for a reason buried in the specification.
The number that is not in the headline
The fibre has 19 cores — nineteen separate light paths in one strand, instead of the single core in ordinary fibre. That part is expected; multi-core fibre is how these records are set.
The significant figure is the cladding diameter: 0.125 mm. That is the standard dimension, the same as the single-core fibre already buried under streets and laid across ocean floors worldwide.
Previous multi-core records were often set on fatter fibre. A thicker strand means new cable designs, new splicing equipment, new connectors, new handling procedures and new manufacturing lines — which means the record is a laboratory result that cannot be deployed without replacing an industry's worth of tooling.
Keeping to 0.125 mm means the fibre fits existing cable architecture, existing conduits and existing installation practice. That is the difference between a demonstration and a product.
How the distance was achieved
Capacity over a short span is comparatively easy. Capacity over 1,808 km is hard, because signals weaken and need periodic amplification, and amplifying nineteen cores separately would be impractical.
The work split accordingly. Sumitomo Electric designed and manufactured the coupled 19-core fibre, reducing optical losses across both the C-band and L-band by optimising the structure and arrangement of the cores. NICT built the transmission system and — the harder part — developed an optical amplification relay that amplifies all 19 cores simultaneously.
That amplifier is what converts a short-distance capacity figure into a long-haul one, so the capacity-distance product is the metric the researchers lead with rather than raw speed.
What it does not mean
It does not mean faster home internet, and the gap is worth being clear about.
A petabit per second is backbone capacity — the long-haul links between cities, countries and continents. Nothing about it changes the connection into a building, which is limited by the last mile, the equipment at each end and what a provider is willing to sell.
Nor is this close to deployment. It is a research demonstration using laboratory transmission and amplification equipment, not a product anyone can order. The standard cladding means it could be deployed eventually without rebuilding the industry — which is a claim about the path, not the timeline.
Why backbone capacity is the thing to watch anyway
Demand on long-haul links is rising faster than consumer demand, and for a reason that has nothing to do with consumers: AI training and inference move enormous volumes between data centres. That traffic is machine-to-machine, concentrated on a small number of routes, and growing in a way video streaming never did.
A record that can be installed in existing conduit is directly relevant to that problem. A record that requires new cable everywhere is not.
Sources: NICT's announcement; Sumitomo Electric; the Fiber Broadband Association; TechSpot; and Prothom Alo's original report.
