7,194 km · 4 Landing Points · 3 Countries · Ready for Service: 2026
| Length | 7,194 km |
|---|---|
| Status | In Service |
| Ready for Service | 2026 |
| Landing Points | 4 |
| Countries | 3 |
| Location |
|---|
| Annie's Bay, Bermuda |
| Myrtle Beach, SC, United States |
| São Miguel, Portugal |
| Sines, Portugal |
Nuvem is the newest entry on our monitoring list. The 7,194 km submarine cable was designed by Google to cross the North Atlantic, and it does something uncommon among recent transatlantic systems: it lands in four places, not two. There are cable houses on both sides of the ocean and on two mid-Atlantic archipelagos — Annie's Bay in Bermuda and São Miguel in the Azores — wired directly into the same trunk. The US endpoint is Myrtle Beach, South Carolina. The European endpoint is Sines, Portugal. Between them, Nuvem quietly connects two islands that used to live on the latency frontier.
The cable is listed as ready for service in 2026, and our probe has been collecting ping samples from Myrtle Beach toward a Portuguese target next to the Sines landing station. Over the past 14 days, 30-plus clean samples have arrived. Minimum round-trip: 137.8 ms. Average: 148.4 ms. Standard deviation over the full window: about 8 ms. That is a well-engineered transatlantic — quiet, tight, and almost boring.
"Almost" is the operative word, because Nuvem's design is anything but ordinary.
Most new transatlantic cables are straightforward: one cable house in Europe, one in North America, one big fibre trunk spanning the ocean. Equiano, Amitié, Marea, Dunant — all follow that pattern. They are optimised for a single job: carry as much capacity as possible between two continents.
Nuvem does not. It picks up two islands that sit almost exactly on the transatlantic route — Bermuda and the Azores — and gives them landing stations on the same cable body as the US and Portuguese mainlands. This matters for three reasons.
First, it changes the economics of island connectivity. Bermuda and the Azores have historically been served by dedicated cable systems (SOL, Gemini, Atlantis-2, ARBRE, BUGIO, AFOS) that carry only their own island's traffic. A cable like Nuvem, with an existing transatlantic customer base paying for the majority of capacity, can hand those islands a fibre pair at marginal cost. It turns a stop on the way into a primary site instead of a spur.
Second, it creates real redundancy. Bermuda and the Azores now have a direct path not just back home to Europe or the US, but to each other — all on a single cable body. Before Nuvem, an Azores-to-Bermuda packet would traverse Europe and then the US, tens of thousands of kilometres to cover what is geographically a short Atlantic hop.
Third, it changes the failure domain. A single cable with four landings is more resilient than a single cable with two. When a fishing trawler severs the US-to-Bermuda segment, traffic can still reach Bermuda via the Bermuda-to-Azores or Bermuda-to-Portugal sections. A two-point cable has no such fallback.
| Landing | Country | Coordinates |
|---|---|---|
| Myrtle Beach, SC | United States | 33.69° N, 78.88° W |
| Annie's Bay | Bermuda | 32.36° N, 64.66° W |
| São Miguel | Portugal (Azores) | 37.74° N, 25.68° W |
| Sines | Portugal | 37.96° N, 8.87° W |
Our monitor sits near the Myrtle Beach landing, pinging an IP in the neighbourhood of the Sines cable station. The samples are clean and consistent:
| Window | Samples | Min RTT | Average RTT | Max RTT | StdDev |
|---|---|---|---|---|---|
| Last 14 days | 30+ | 137.8 ms | 148.4 ms | 197.1 ms | ≈ 8 ms |
A curiosity in the raw data: for most of March the ping target was 193.136.0.1 (a Portuguese academic network router) with hop count 12 and a typical RTT of 149–150 ms. Starting around April 10, the target shifted to 204.27.68.152 with hop count 15 and RTT dropped to 137.8 ms. That is about 12 ms less latency on what is ostensibly the same cable.
Two interpretations are consistent with the shift. Either the newer target is behind a cleaner peering path — one less intermediate hop costs roughly a fraction of a millisecond but cuts out routers that can add variable processing time — or the underlying transit operator has migrated the route to take advantage of Nuvem itself, which may not have been in their default path until recently. Both would show exactly this signature.
The great-circle distance from Myrtle Beach to Sines is approximately 6,100 km. Light in submarine glass travels at about 204,500 km/s (the refractive index of a single-mode fibre core is close to 1.467). A round-trip through 6,100 km of fibre therefore costs a minimum of about 59.7 ms.
We measure 137.8 ms. That is 2.30× the physics floor — a multiplier that sits squarely in the normal range for transatlantic systems. For comparison, our piece on Marea documented a 1.95× ratio on Bilbao to Virginia Beach (best-in-class for long-haul), and Equiano logged 2.5× on Sesimbra to Melkbosstrand. Nuvem's Myrtle Beach to Sines route sits comfortably between the two.
The extra 78 ms over the theoretical floor is distributed across: landside backhaul at each end (a probe in Myrtle Beach is not in the cable hut itself), serialization delay through 15 IP hops of routers and switches, and optical regeneration at submarine amplifiers. None of this is waste — it is the operational overhead of running a real-world packet network.
Cables designed by hyperscalers tend to look unusual on the map, and Nuvem is not an exception. The branching architecture is not generic — it is tuned to three specific corporate goals.
Latency to Europe. Google operates data centres in Portugal (near Sines, in fact) and needs the lowest-latency possible path from US east coast customers to European workloads. Nuvem's Myrtle Beach to Sines segment is among the shortest transatlantic routes available — most competitor cables land in Lisbon, Bilbao, Marseille, or the UK, which are all marginally further.
Mid-Atlantic coverage. Bermuda and the Azores are strategic for both content delivery and disaster recovery. A Google edge site or CDN pop on either island can serve regional traffic at single-digit millisecond RTT instead of 50+ ms over to the continent.
Redundancy through geography. A single cable with four landings carves its own fault-tolerance. Most transatlantic "diversity" today is achieved by running two physically separate cables. Nuvem adds a third axis: four-corner redundancy on a single cable body, so that traffic between any two landings has at least three geographically distinct failure paths available from the same conductor.
Because Nuvem is new, our monitoring picture is incomplete. We only have one direction of measurement today — Myrtle Beach toward Sines. The reverse path (Sines toward Myrtle Beach) is not yet populated in our cable_monitor dataset, meaning we have no way to cross-check whether the route is symmetric. Nor do we have samples on any of the three spur segments (US to Bermuda, Bermuda to Azores, Azores to Portugal) — those require probes at the respective landings.
The 137.8 ms minimum tells us one thing with certainty: when traffic is routed across Nuvem, it performs as a well-built 2026-era transatlantic cable should. The newer 15-hop path, with ~12 ms less latency than the 12-hop path used earlier, is a small sign that transit operators are in the process of adopting Nuvem as a preferred route — but we do not yet have enough data to draw a strong conclusion about long-term routing patterns.
Three design trends visible in Nuvem are converging across new submarine systems:
Nuvem is still settling into its routing slot on the global internet. The measurements we have are from a narrow window and one direction. We will update this article as more samples accumulate and as reverse-direction probes come online.
Live data we collect on this cable is on the Nuvem cable page. Compare with our coverage of other recent transatlantic cables on Marea, Equiano, and the Baltic-focused COBRAcable.
| Status | ✓ Normal |
|---|---|
| RTT | 134.52 ms / base 138.46 ms |
| Last checked | 2026-04-17 20:31 |
Monitored using RIPE Atlas probes. Open monitoring →
| Min | Avg | Max | # | |
|---|---|---|---|---|
| 7 days | 134.5 | 138.2 | 141.7 | 9 |
| 30 days | 134.5 | 147.6 | 197.1 | 48 |
| 60 days | 134.5 | 148.0 | 197.1 | 61 |
Find the actual cable routing distance between any two cities
Open Calculator →