JUNO

Based on 15 RIPE Atlas measurements from GeoCables monitoring infrastructure, April 2026.

JUNO is a 11,710-kilometre transpacific submarine cable that connects Hermosa Beach, California, on the United States Pacific coast to Shima, on the Pacific side of Japan's Mie Prefecture. The cable came into service in 2025 and is among the newest transpacific systems in operation. It is a single-purpose, two-landing point-to-point trunk built to the latest specification: twenty fibre pairs, a design capacity of 360 terabits per second across the full system, and a route plan optimised for the shortest workable subsea path between the two coastlines. The system is owned and operated by Seren Juno, a project company organised around the cable, with NTT-affiliated capital and engineering on the Japanese side.

JUNO sits at one of the busiest data corridors on the planet. The link between Tokyo-area data centres and the Los Angeles peering region carries financial traffic, public-cloud east-west replication between AWS and Azure regions in us-west-2/ap-northeast-1, content delivery for Asian markets, and a growing share of generative-AI inference traffic that returns model outputs from US compute back to Japanese end users. JUNO joins a small set of high-capacity cables that share this load — FASTER, JUPITER, PLCN, and the older but still heavily used Unity and TPC systems among them — and adds a new physical route to the diversity pool. For the cloud operators that fund this generation of cables, additional physical paths between US and Japan are not optional features. They are how single-cable cuts stop becoming regional outages.

One percent above the physics floor

The minimum round-trip we observe between Grover Beach (the closest probe to the Hermosa Beach landing) and Shima is 115.77 ms. The physics floor for an 11,710-km path — the smallest possible round-trip if a photon travelled the cable's full geographic length at the speed of light in fibre — is 114.60 ms. JUNO measures at 1.010× of that floor. One percent above the theoretical minimum.

That number describes a cable that is doing almost nothing wrong. The 1.17 ms gap between observation and physics is small enough to be accounted for entirely by the route's small over-length above great-circle distance, plus a few microseconds per kilometre of regenerator and amplifier latency along the run. There is no detectable detour, no mid-cable hop into a third country, no routing into a regional hub before crossing. The packet enters the wet plant at one end, traverses the fibre, and exits at the other end, with the link adding effectively no overhead beyond the unavoidable cost of crossing the Pacific.

This is structurally the opposite of what we measure on most multi-landing trunks. EXA North and South across the Atlantic measures at 0.716× of its system floor — well below the floor — because the Southport-to-Lynn corridor is a chord across the cable's branched topology rather than a walk along its full length. ARCOS-1 in the Caribbean reads 0.613× for the same reason: traffic across the basin uses a chord, not the full ring. JUNO is structurally simpler. There are no other landings, no branching units, no alternative chords. The only path between Hermosa Beach and Shima that the cable offers is the entire cable. The 1.010× number tells us the engineering succeeded.

Stable to within two milliseconds

Across the 15 measurements we have collected on the Grover Beach → Shima direction in April 2026, the round-trip averages 118.24 ms, with a maximum of 119.32 ms and a standard deviation of 1.33 ms. Every observation falls inside a four-millisecond window. The traceroute path is consistent at 17 hops median, with the same intermediate AS-level transit each time. For a cable in its first full year of revenue service, carrying transpacific cloud-replication traffic that has zero tolerance for jitter, this is the consistency profile that the operators were building toward.

It is also the consistency profile that the transpacific corridor as a whole demands. The gap between US-West and AP-Northeast cloud regions is one of the most aggressively measured and tuned latency budgets in commercial networking, alongside the New York-London corridor on the Atlantic side. Carriers, hyperscalers, and trading firms watch this number to the third decimal. A new cable that arrives at 1.33 ms standard deviation on day one is a cable that has been operationally accepted by the customers paying for it.

The Shima return path

One observation worth flagging concerns the reverse direction. Probes located near Shima that target the Hermosa Beach landing return measurements averaging in the low double digits — far below the physics floor of 114.60 ms. This does not contradict the forward number; it tells us that the return packet does not retrace the cable. The traffic from the Japanese side of our test pair exits through a regional Tokyo or Osaka peering point and reaches its US-coast target via a different route — most likely a shorter logical hop into a CDN edge or anycast endpoint that happens to share the destination's address space. JUNO is the path we measure when we drive traffic from California to Shima. The reverse leg in this corridor uses a different physical solution. This kind of asymmetric routing is normal on the Pacific corridor, where multiple cable systems and large peering fabrics compete for the same flows, and it is one reason single-cable measurements need to be read directionally.

Shima as a transpacific hub

The choice of Shima as JUNO's Japanese landing is not incidental. Japan's 70-plus submarine cable landing stations are concentrated along three coastal clusters, and Shima is the principal Pacific-facing one — a quiet stretch of Mie coast that already hosts FASTER, Unity, AAG, and several older transpacific systems. Cables landing at Shima reach Tokyo's data-centre cluster through diverse terrestrial backhaul that has been hardened against the seismic risk profile of the region. The same backhaul carries JUNO's traffic the last few hundred kilometres into the metro. On the California side, the Hermosa Beach landing is part of the Los Angeles peering complex, which puts JUNO into the same terrestrial fabric as Matrix Cable System and the other recent Pacific entrants once their traffic reaches the coast.

What 1.010× means in the corridor

For users and operators, the practical effect of JUNO's near-floor measurement is straightforward. Any transpacific traffic that is path-engineered onto this cable inherits the lowest end-to-end latency the route currently offers — the first millisecond above the physical limit. For high-frequency replication between cloud regions, this matters in absolute terms; for content delivery, the difference is measured in single-digit milliseconds against neighbouring routes but compounds across millions of daily flows. A 2025-RFS cable arriving at the floor on day one means that the corridor's latency frontier has moved by another small amount, in the same direction it has been moving for two decades. Gondwana-1 between New Caledonia and Sydney sits at 1.081× of its own floor on the same kind of single-trunk geometry; JUNO at 1.010× is doing the same thing on a route an order of magnitude longer.

What we measure on JUNO — 118 ms to within a millisecond and a third, hugging the physical limit — is what a brand-new transpacific cable looks like when the engineering brief has been executed cleanly. The cable is one year into its service life. The current measurement window will continue, and the position relative to the floor will be one of the things we watch for as the corridor's traffic patterns shift around it.