Nokia Wants to Shrink the Internet's Busiest Highways to Router Size

Almost all of the internet lives underwater. Roughly 99 percent of intercontinental data travels through subsea fiber-optic cables, not satellites, and every video call, cloud backup, and AI training run eventually passes through glass threads laid on the ocean floor. This week, Nokia made a case for changing how those threads are lit. In a technical post published October 12, the company's optical networks marketing lead Serge Melle said coherent pluggable optics, small devices that already carry the bulk of land-based internet traffic, are now ready for subsea service after a trial on a live cable between Brazil and the United States.

It sounds like a footnote in telecom engineering. It is not. The box at the end of a subsea cable is one of the most expensive, power-hungry parts of the entire internet, and shrinking it changes the economics of every new ocean crossing.

What a coherent pluggable actually is

Forget the word "pluggable" and think of it as a laser with a brain. A coherent pluggable is a small transceiver module that slides directly into a router, the same kind of box that directs traffic inside a data center. Inside the module, a laser encodes data onto beams of light using a technique called coherent modulation, which reads both the phase and the amplitude of the light wave to pack more bits into every wavelength of fiber.

The reason this is news is the digital signal processor, the chip that makes sense of the light at the other end. Advances in DSP design have steadily raised the speed of these modules from 100 and 200 gigabits per second to 800 gigabits per second in current generations. Each leap means the same fiber carries more data with less equipment. On land, this was already a quiet revolution: operators moved capacity straight into routers, a setup called IP-over-DWDM, skipping racks of dedicated optical transport hardware.

The box at the end of a subsea cable is one of the most expensive, power-hungry parts of the entire internet, and shrinking it changes the economics of every new ocean crossing.

How a subsea cable carries your data

To see why pluggables matter at sea, it helps to picture the cable itself. A modern subsea cable contains several pairs of hair-thin glass fibers. Each fiber carries dozens of separate light wavelengths at once, a trick called dense wavelength-division multiplexing (DWDM). Think of each wavelength as a lane on a highway, with the fiber as the roadbed.

Light fades as it travels, so every 60 to 100 kilometers an underwater repeater amplifies the signals passing through it. The repeaters are powered by electrical current fed down the cable from landing stations on shore, where the cable comes ashore and the optical signals are converted back into packets that join the terrestrial internet. The landing station is where pluggable optics would live: instead of a room full of chassis converting wavelengths, a router with plug-in modules does the job.

That substitution saves two things operators care about most: power and floor space. Landing stations are expensive real estate in coastal cities, and the optical gear that terminates a cable can draw enormous amounts of electricity. Pluggables use less of both, which is exactly why, in Melle's phrasing, subsea operators are now asking whether the benefits proven on land can be leveraged underwater.

What the BRUSA trial proved

The answer, at least for medium distances, appears to be yes. Nokia ran its trial with Telxius, the infrastructure company that operates the BRUSA cable linking Brazil and the United States, and the numbers were striking. Using Nokia's ICE-X 800G technology, the pair pushed 600 gigabits per second on a single wavelength across 2,841 kilometers of live subsea fiber. They also reported a transmission record: 400 gigabits per second per wavelength across 5,862 kilometers.

Those distances matter. A transatlantic crossing is roughly 6,000 to 7,000 kilometers, so a pluggable that works reliably past 5,800 kilometers is suddenly relevant for the internet's most important routes. Nokia and Telxius had already announced in September that Telxius would deploy the same 800G coherent pluggable technology across its terrestrial networks in Europe, the United States, and Latin America, a sign the trial was not a one-off laboratory stunt.

The timing lines up with a broader market shift. Analyst firm Cignal AI has identified 800ZR/ZR+, the standard family these pluggables belong to, as one of the fastest-growing segments in optical networking, ramping roughly twice as fast as the previous generation. The driver is no mystery: AI workloads are pushing unprecedented volumes of traffic between hyperscale data centers, and operators are hunting for the cheapest way to add capacity.

What pluggables still cannot do

There is a ceiling, and Nokia is open about it. For the longest, most demanding routes, cables stretching 15,000 kilometers or more, the company says the big performance-optimized embedded optical systems remain essential. On those spans, amplifier noise accumulates and every fraction of spectral efficiency counts, which is territory where full-size transponders still beat compact plugs.

That is also why the industry is attacking the capacity problem from both ends. While Nokia pushes router-sized optics, Meta announced a very different approach in September: Petal, a planned transatlantic cable using two-core fiber, with two separate light-guiding channels in each strand, aiming for a headline aggregate capacity of one petabit per second. One project changes the light; the other changes the endpoints. Both are responses to the same pressure, which is that the internet's underwater highways are getting crowded.

The perspective to keep: the internet was never really wireless. It is a web of glass under the sea, and the most important upgrades of the next decade may be the smallest boxes anyone plugs into it.