How a Phone Connects to the Internet

TL;DR. Opening a map app and watching it load in under a second hides one of the longest journeys in this book: a radio handshake with a nearby cell tower, an authentication check against a carrier's database, a handoff into the carrier's core network, a jump onto the public internet through a gateway, a run across a fiber-optic backbone that may cross an ocean floor, and, more often than not, a stop at a server sitting much closer to home than any of that suggests. Almost none of the data that makes a modern phone feel instant travels by satellite. It travels through glass fibers thinner than a human hair, many of them lying on the seabed, tended by a global fleet of around sixty specialized ships, and standardized by committees most phone owners have never heard of.

Key takeaways

  • A phone doesn't just "get signal." Its SIM card holds a secret key that lets the carrier's network cryptographically confirm the phone is who it claims to be, every time it connects, before any data moves.
  • 4G and 5G aren't just "faster" versions of each other. 5G's biggest gains come from using much higher radio frequencies, which travel shorter distances, which is why 5G networks need vastly more, smaller antennas packed closer together than 4G ever did.
  • Once data leaves a cell tower, it stops being a "phone" problem entirely. It's routed through the carrier's core network to an internet gateway and then travels like any other internet traffic, often over the same fiber backbone your home Wi-Fi uses.
  • Somewhere between 95 and 99 percent of intercontinental internet and phone traffic travels through undersea fiber-optic cables, not satellites. Around 570 of these cables are in service worldwide, stretching more than 1.4 million kilometers.
  • Cell tower climbing has one of the highest fatality rates of any American job, roughly ten times the construction industry's rate during the years regulators studied it most closely. Undersea cable faults are repaired by a worldwide fleet of only about sixty specialized ships.
  • A single earthquake-triggered cable break off West Africa in March 2024 degraded or cut internet service in thirteen countries at once; suspected sabotage in the Baltic Sea in 2023 and 2024 has made undersea cables a live national security question, not just an engineering one.

The moment nobody thinks about

A thumb taps a map icon. Before the icon has finished its little bounce animation, streets, a blue dot, and a search bar are already on screen. Nothing about the moment suggests effort. But in the time it took to read this sentence, that phone has already done all of the following: proven its identity to a cellular carrier over the air, been handed a slice of radio spectrum on a nearby tower, sent a request through that carrier's private network and out onto the public internet, had that request routed across some combination of terrestrial fiber and undersea cable, arrived at a server that might be in the next city or on another continent, and received an answer that retraced the same path backward, all inside a window of time too short for a person to consciously notice. The map didn't load quickly. It loaded at the edge of what physics and engineering currently allow, and the entire point of the system underneath it is to make that edge feel unremarkable.

What a phone actually does to get on the air

The part of a phone that makes this possible isn't the screen or the processor. It's a fingernail-sized chip most people never look at directly: the SIM card (Subscriber Identity Module). A SIM doesn't just identify a phone number. It holds a unique subscriber identifier, the IMSI (International Mobile Subscriber Identity), and a secret cryptographic key that never leaves the card and is never transmitted in the open.

When a phone powers on, it scans for radio signals broadcast by nearby cell towers, more precisely called base stations or cell sites, and picks the strongest one operated by its own carrier (or, while roaming, a partner carrier). The phone sends its IMSI to that tower, which forwards it deeper into the network to the carrier's authentication center, a database that stores a matching copy of every SIM's secret key. Rather than sending the key itself, which would be easy to intercept, the network sends the phone a random number, and both the phone's SIM and the carrier's database independently run that number through the same cryptographic algorithm using the shared secret key. If the two results match, the network has proven the phone is genuine without either side ever transmitting the actual secret, and a session encryption key, generated fresh from that exchange, is used to scramble everything that follows. Only after this challenge-and-response check succeeds does the tower let the phone attach to the network and start moving data.

None of this is a one-time event. As a phone moves, physically walking down a street or riding in a car, it continuously measures the strength of the signal from its current tower and from nearby towers it can also hear, and reports those measurements back to the network. When a neighboring tower would serve the connection better, the network coordinates a handoff (also called a handover), shifting the phone's connection to the new tower. Cellular networks generally try to do this as a "soft" handoff, briefly overlapping the old and new connections so nothing drops in between, rather than a "hard" handoff that disconnects first and reconnects second. In a dense city, with towers spaced closely together, this can happen many times during a single walk; in open countryside, where one tower's coverage area might span many miles, it happens rarely.

Don't be confused: a handoff is not the same thing as roaming. A handoff moves a phone's connection between two towers that belong to the same network, invisibly, usually mid-call or mid-download, with no change to the phone's plan or billing. Roaming happens when a phone connects to a different carrier's towers entirely, typically because it has left its home carrier's coverage area (crossing into another country, for example). Roaming requires the two carriers to have a prior agreement letting each other's subscribers use their network, and it's the reason an international phone bill can look different from a domestic one even though, from the phone's perspective, both just look like "having signal."

4G, 5G, and what the "G" actually changes

Marketing tends to compress "5G" down to a single idea: faster. The real differences are more specific, and they explain both 5G's advantages and its limits.

Every generation up through 4G LTE (Long-Term Evolution, the technology that actually delivers what phones display as "4G") operated mostly on radio frequencies below 6 gigahertz. Those frequencies travel relatively far and pass through walls and foliage reasonably well, which is why a single 4G tower can cover several miles. 5G keeps using that same lower range for broad coverage, but adds access to much higher frequencies as well, including millimeter wave spectrum in the 24 to 100 gigahertz range. Millimeter wave can carry enormous amounts of data very quickly, but it barely penetrates walls and its useful range is often measured in a few hundred meters, sometimes closer to a thousand feet. That physical limitation is precisely why 5G networks depend so heavily on small cells, compact antennas mounted on lampposts, buildings, and utility poles rather than tall towers, deployed in far greater density than 4G required. Industry estimates suggest a small-cell-heavy 5G network can support on the order of a million connected devices per square kilometer, versus roughly 4,000 for typical 4G deployments, simply because so many more, smaller cells share the same area.

The other headline difference is latency, the delay between sending a request and getting the first byte of a response. 4G networks typically run in the 30 to 50 millisecond range; 5G is designed, under ideal conditions, to get as low as 1 millisecond, though most real-world 5G connections today land somewhere around 10 to 20 milliseconds rather than that theoretical floor. Lower latency matters far more for things like cloud gaming, remote-controlled machinery, or augmented reality than it does for loading a static map tile, which is one reason 5G's biggest practical benefit for an ordinary map app is usually higher throughput and better performance in crowded areas, not the headline millisecond figures used in advertising.

Don't be confused: not all "5G" on a phone's status bar is the same 5G. Carriers commonly deploy three different tiers under one label: low-band (long range, only modestly faster than good 4G), mid-band (a middle ground most people actually experience day to day), and millimeter-wave high-band (the dramatic speeds used in demonstrations, available only within a short distance of a small cell, often just on a single city block). A phone showing "5G" in a suburb and a phone showing "5G" outside a stadium may be using entirely different frequency ranges with very different real-world performance.

From a cell tower to "the internet"

Once a phone's data leaves the tower, it stops being a phone problem. The tower is connected, usually by its own fiber-optic or microwave link called backhaul, to the carrier's core network, the part of the system responsible for authentication, mobility management (keeping track of which tower a phone is near as it moves), billing, and policy enforcement. The core network doesn't just relay traffic; it also assigns each connected phone an IP address, the same kind of numeric address any computer needs to communicate on the internet.

At the edge of that core network sits a gateway, a piece of infrastructure whose entire job is to be the boundary between the carrier's private mobile network and the public internet. In 4G networks this function is typically called a packet gateway; in 5G's redesigned core, a similar role is handled by what's called a user plane function. Whatever it's named, this is the exact point where a request typed into a map app stops being "cellular traffic" and becomes ordinary internet traffic, subject to the same routing rules as a request from a laptop on home Wi-Fi. From here on, the cellular part of the story is over. What happens next is shared by every device on the internet, wired or wireless.

The long way around: fiber, oceans, and the servers that are secretly close

Between a phone and a map app's servers lies a physical network almost nobody sees. Traffic leaving a carrier's gateway generally rides onto a backbone network, high-capacity fiber-optic lines that carry the overwhelming bulk of long-distance internet traffic between major cities. In North America, these routes link hubs like New York, Chicago, and Los Angeles; in Europe, cities like London, Frankfurt, and Amsterdam; in Asia, Tokyo, Shanghai, and Singapore. A handful of companies, often called Tier 1 providers, own most of this long-haul fiber, and it's genuinely boring-looking infrastructure: buried conduit running along railway lines, highway easements, and utility corridors, carrying glass fibers that move data as pulses of light.

When that traffic needs to cross an ocean, it almost never goes by satellite. It goes through a submarine cable, a fiber-optic cable, often not much thicker than a garden hose once armored for protection, laid directly on the seabed. As of 2025, roughly 570 of these cables are in active service worldwide, with dozens more planned, together stretching more than 1.4 million kilometers, enough to circle the Earth over thirty times. Depending on how the math is done, somewhere between 95 and 99 percent of intercontinental internet and phone traffic travels through these cables rather than through satellites, a fact that surprises most people who assume "global" data mostly beams through space. It doesn't. Satellites remain useful for remote coverage and specific applications, but their total data capacity is still a small fraction of what the undersea cable network moves every second.

Cables come ashore at a cable landing station, a secure facility where the undersea cable connects into a country's terrestrial fiber network, and from there the data keeps moving through the same kind of backbone fiber described above. Along the way, at critical junctions in major cities, networks operated by different companies, carriers, and content providers meet at an internet exchange point (IXP): essentially a shared facility where many networks plug into the same switching equipment and exchange traffic directly with each other, rather than each paying a third party to carry it between them. An IXP in Frankfurt, for instance, routinely handles traffic peaks above ten terabits per second, all of it different companies' networks handing data directly to one another in the same building.

None of this explains, though, why a map loads in under a second rather than the several hundred milliseconds a genuine round trip to a distant data center would take. That's the job of a content delivery network (CDN): a set of servers positioned in many cities around the world that keep cached copies of frequently requested content, including map tiles, close to where people actually are. Instead of every phone in, say, Denver fetching a map tile from a data center in another country each time someone opens the app, a CDN server much closer, sometimes in the same metro area, already has a copy ready to serve. The request from the phone still travels through the tower, the core network, the gateway, and some stretch of backbone fiber, but it usually doesn't have to cross an ocean or reach the app's origin servers at all. That shortcut, quietly built by companies most users have never heard of, is a large part of why the whole experience feels instantaneous.

Who keeps it running

A map app loading is the visible end of a chain of work that spans several distinct, mostly invisible professions. Radio frequency engineers design where towers and small cells go and which frequencies each one uses so neighboring cells don't interfere with each other. Cell tower technicians physically climb towers, sometimes several hundred feet up, to install, inspect, and repair antennas and equipment, a job that has carried one of the worst fatality rates of any occupation tracked in the United States; a ProPublica investigation found that during the years OSHA studied the industry most closely, tower climbing had a death rate roughly ten times that of construction, and nearly 100 climbers died on communications towers of all kinds between 2003 and 2011 alone, about half of them on cell sites specifically. Network operations center (NOC) staff, working in shifts around the clock, watch dashboards tracking traffic, equipment health, and outages across an entire carrier's network, ready to respond the moment something looks wrong. On the ocean side, specialized cable-laying and repair ships, crewed by mariners and cable engineers and operated by a handful of firms worldwide (the global fleet capable of this work numbers only about five dozen vessels), spend weeks at sea installing new cables and racing to fix broken ones. And behind all of it, spectrum regulators, like the Federal Communications Commission in the United States, decide who is legally allowed to transmit on which radio frequencies in the first place, without which every carrier's towers would simply drown each other out.

Where this came from

Cellular technology and undersea cables developed on separate historical timelines that only merged once phones needed to reach the wider internet, not just each other.

Mobile phone networks are usually described in "generations." 1G, the first commercial cellular service, launched in Japan in 1979 and spread through the early 1980s; it was purely analog, carrying voice calls the same way an old radio carries a broadcast, with no real data capability and notoriously easy-to-intercept calls. 2G, launched on the GSM standard in Finland in 1991, went digital, which made calls more secure and efficient and, almost as an afterthought, introduced text messaging. 3G, first deployed commercially in Japan in 2001, was the generation that made mobile internet access genuinely usable, delivering data speeds around a few megabits per second instead of a few kilobits. 4G, and specifically the LTE standard that came to define it, launched commercially in Scandinavia at the end of 2009, and delivered the kind of speed that made streaming video and modern map apps practical on a phone for the first time. 5G began commercial rollout in 2019, led by South Korea, bringing the higher-frequency, higher-capacity, lower-latency architecture described earlier in this chapter.

Undersea cables predate all of this by more than a century, and they didn't start out carrying data at all. They carried telegraph signals. The first transatlantic telegraph cable was completed in August 1858, and Queen Victoria and President James Buchanan exchanged a congratulatory message over it within days of its completion. The celebration was premature: engineers pushing for faster signal speed applied far too much voltage to the line, and the cable failed within about three weeks, destroyed by the very people trying to improve it. It took until 1866, after the American Civil War had ended and after engineer William Thomson (later Lord Kelvin) helped refine the cable's design and signaling methods, for a second transatlantic cable to succeed where the first had failed, this time reliably and for years. That 1866 cable is the direct conceptual ancestor of every fiber-optic line now lying on the ocean floor: the material changed from copper wire carrying electrical pulses to glass fiber carrying light, but the basic proposition, a physical line laid across an ocean floor to connect two continents, has never been replaced by anything else at scale.

Standards that make it all interoperable

None of this works without agreements that let equipment from different manufacturers, in different countries, talk to each other reliably. Cellular technology itself is defined by 3GPP, the 3rd Generation Partnership Project, an international collaboration of telecommunications standards bodies founded in 1998 that writes and maintains the technical specifications for GSM, UMTS (3G), LTE (4G), and 5G NR. When a phone bought in one country connects to a tower built by an entirely different manufacturer in another country, 3GPP's specifications are the reason the two sides agree on how to talk to each other at all.

Radio spectrum itself is a limited public resource, and national regulators decide who gets to use which frequencies. In the United States, the Federal Communications Commission (FCC) allocates spectrum and, since 1994, has mostly done so through competitive auctions rather than administrative assignment, having raised tens of billions of dollars for the U.S. Treasury across dozens of auctions while granting carriers exclusive rights to specific frequency bands.

Undersea cables answer to a messier layer of international law. The United Nations Convention on the Law of the Sea (UNCLOS) guarantees every nation the right to lay submarine cables on the high seas, but it only weakly obligates countries to protect cables from damage, leaving most real enforcement to individual national laws. Where a cable comes ashore, national regulators step back in: in the United States, the FCC licenses every cable landing station connecting the country to another nation, and cable operators must separately negotiate rights of way and landing permits with each coastal government whose waters or territory the cable crosses or touches.

Keeping it working

Reliability here is an ongoing maintenance program, not a one-time installation. Cell towers and small cells go through scheduled physical inspections, and technicians periodically swap out or add radio equipment as carriers roll out new spectrum bands or additional 5G gear onto existing structures rather than always building new ones. Fiber-optic backbone networks are monitored for physical cuts, most commonly caused by construction crews accidentally digging through buried lines, and repair crews are dispatched to splice a break back together, a delicate process since a single fiber strand is thinner than a human hair.

Undersea cable maintenance is its own specialized discipline. When a cable faults, engineers first locate the break using a technique that resembles sonar for light: a pulse of light is sent down the fiber, and where the cable is broken, part of that pulse bounces back, letting engineers calculate almost exactly how far down the cable the fault sits. A cable ship, several of which are kept on standby around the world specifically for this purpose, is then dispatched to the location. In deep water, the ship uses a grappling hook to snag the cable from the seabed and haul the damaged section to the surface; in shallow water, a remotely operated underwater vehicle handles the retrieval instead. Once the damaged section is on deck, engineers splice in a length of new cable in an onboard repair room, test the new joint, and lower the repaired line back to the seabed, a process that can take anywhere from several days to a few weeks depending on weather, water depth, and how much spare cable and permitting the operation requires. Carriers and network operators also run continuous capacity planning, projecting how much data demand will grow in a given city or region years ahead, because building new towers, laying new backbone fiber, or commissioning a new undersea cable can take years from planning to completion, far longer than any single spike in demand takes to arrive.

When it breaks

The clearest illustration of how much intercontinental connectivity depends on a small number of physical cables came on March 14, 2024, when several major submarine cables along the west coast of Africa, including WACS, MainOne, SAT-3, and ACE, were damaged in the same general area, apparently by underwater seismic activity. Because so many separate cables happened to converge at a similar point offshore, the damage disrupted or cut internet connectivity across thirteen West African countries simultaneously, in some cases for weeks; Nigeria alone was estimated to have lost more than $590 million in economic activity over just the first four days of the outage before service was gradually restored.

Not every incident is accidental, or at least not conclusively so. Starting in 2023 and continuing through 2024, a string of undersea cable and pipeline faults in the Baltic Sea drew intense scrutiny after investigators traced several of them to ships whose anchors appeared to have dragged for miles along the seabed directly beneath the damaged lines. The most closely examined case involved the tanker Eagle S, linked to Russia's so-called "shadow fleet" of vessels used to evade Western sanctions, which Finnish investigators say dragged its anchor across the Baltic seabed for dozens of miles on Christmas Day 2024, severing a subsea power cable between Finland and Estonia along the way. Russia has denied any role. Whether or not each individual incident proves to be deliberate sabotage, the pattern has pushed NATO governments to treat undersea cable security as a live military and intelligence concern rather than a purely commercial maintenance issue.

Cellular networks fail in more familiar ways closer to home. During wildfires, utilities sometimes deliberately cut electrical power across whole regions to avoid sparking new fires, and cell towers without power quickly exhaust their backup batteries or generators, some of which are rated to last only a few hours; California wildfire seasons have seen days when more than a fifth of cell towers in some counties went dark at once. Hurricanes cause a more direct kind of damage: high winds and flooding sever fiber backhaul lines and physically damage towers, which is why Hurricane Helene in 2024 knocked out cellular service across parts of several U.S. states and prompted carriers to deploy emergency mobile equipment, including trucks nicknamed COWs (Cells on Wheels) and COLTs (Cells on Light Trucks), to restore temporary coverage in the hardest-hit areas.

The scale of it

Globally, roughly 5.8 billion people, about 70 percent of the world's population, held a mobile subscription in 2025, and about 4.7 billion people were actively using mobile internet on their own device. In the United States alone, counts vary depending on definition: industry group figures put traditional free-standing cell towers at roughly 142,000, while broader counts that include smaller cell sites and both outdoor and indoor small-cell nodes push the total number of individual cellular access points toward 1.4 million. Underpinning nearly all of the long-distance traffic those towers ultimately connect to, the roughly 570 undersea cables in service worldwide, and the fleet of only about sixty ships capable of laying or repairing them, form a physically small set of infrastructure carrying a wildly disproportionate share of global communications, financial, and even military traffic. Cables connected to just the United States alone are estimated to underpin more than $12 trillion in financial transactions every single day.

Trade-offs and what's next

The most visible near-term challenger to this system is satellite internet, led by SpaceX's Starlink constellation, which grew to more than 9 million active subscribers across over 150 countries by late 2025, roughly double the number it had a year earlier. Starlink and similar low-earth-orbit satellite networks are genuinely useful where laying fiber or building towers isn't economical: remote rural areas, ships at sea, disaster zones with damaged ground infrastructure. But they still carry only a small fraction of the data volume undersea cables move, and nothing suggests satellites will overtake fiber for bulk intercontinental traffic soon. Their real role is filling gaps the cable-and-tower system doesn't reach, not replacing it.

On the cellular side, standards bodies have already begun work on 6G, targeting technical specifications around 2028 and commercial deployment around 2030, aiming at even higher-frequency spectrum, tighter integration between communication and sensing, and architectures designed from the start around artificial intelligence workloads rather than having AI added later. Whether 6G changes daily use as much as the 3G-to-4G jump did, or turns out to be as incremental for ordinary users as 4G-to-5G has been, remains unsettled.

Undersea cables, meanwhile, are becoming a more openly strategic asset than they have ever been treated as before. Governments in the United States, Europe, and Asia are now explicitly discussing cable security in the same conversations as pipeline and power grid security, funding redundant routes, tightening landing station permitting, and, in several cases, restricting which countries' companies are allowed to build or maintain cables that touch their territory. A piece of infrastructure that spent a century and a half being thought of, when it was thought of at all, as a boring engineering problem is now, increasingly, a national security one.

Back to the map

The map loaded. The dot found its position, the streets filled in, and none of the radio negotiation, the authentication challenge, the handoff logic, the core network routing, the ocean crossing, or the cached tile sitting on a nearby server ever became visible or necessary to think about. That's not an accident or a coincidence of good engineering; it's the entire design goal of every layer described in this chapter, pursued for more than a century, since a fragile copper telegraph wire first survived a trip across the Atlantic Ocean in 1866. The wire is now glass. The message is now a map tile instead of a few words in Morse-adjacent code. The ambition underneath it, to make distance disappear, hasn't changed at all.

The leap: what it replaced, and the work behind it

For nearly all of human history, distance meant delay, and delay meant silence. A message from New York to San Francisco in the 1850s took about 45 days by steamship, or 20-some days overland by stagecoach, which is to say a parent might not learn a child had arrived safely, or died, until a month after the fact. The telegraph compressed that to minutes but charged by the word (Western Union billed 5 cents a word in 1920), so people stripped their news to bones, "ARRIVED CHICAGO STOP," and paid dearly for anything longer. Even the telephone, once it came, did not connect you to anyone directly. Into the 1920s a long-distance call was placed by a human operator patching cords by hand, and in 1918 the average time just to complete that connection was about 15 minutes. And most homes had no privacy on the line at all: by the 1950s, roughly 75 percent of US residential customers were on party lines shared with neighbors, any of whom could quietly lift a receiver and listen to your whole conversation.

The leap is that a message which once took a month now takes milliseconds, and it feels free and private, and it works almost every time. That reliability is not automatic; it is maintained. Cell tower climbing has carried one of the worst fatality rates of any American job, roughly ten times construction's during the years regulators studied it most closely, and the undersea cables that carry 95 to 99 percent of intercontinental traffic are tended by a worldwide fleet of only about sixty specialized ships. On land, telecommunications technicians in the US numbered around 268,500 in 2024, splicing fiber, swapping radios, and watching network dashboards in shifts so that the gap between the map tap and the map load stays too short to notice.

You feel the leap constantly and mostly do not register it: a video call to another continent, a text that arrives before you have put the phone down, a map that fills in as you walk. The morning it fails, you get the old isolation back, undiluted. When several submarine cables off West Africa were damaged around March 14, 2024, internet service degraded or cut out across thirteen countries at once, in some places for weeks, and Nigeria alone was estimated to have lost more than $590 million in economic activity in just the first four days. When a hurricane like Helene in 2024 tears down fiber and towers, whole regions go quiet in a way that would have been ordinary in 1850 and is now an emergency. That the quiet is now the exception is the work of a climber several hundred feet up a tower and a cable crew splicing glass on a pitching deck at sea.

Real-world examples and recent developments

Beyond the companies and incidents already covered, a small number of other named players build, own, and occasionally endanger the physical network this chapter describes.

  • Alcatel Submarine Networks (ASN): one of only a handful of companies, alongside SubCom, NEC, and HMN Technologies, that together manufacture and install nearly all of the world's undersea cables. ASN alone supplied roughly a third of all new cable systems built worldwide between 2020 and 2024. Wikipedia, Alcatel Submarine Networks
  • 2Africa (core system completed November 2025): a Meta-backed submarine cable, built by Alcatel Submarine Networks together with carriers including Orange and Vodafone, that circles Africa and reaches into Europe and Asia, connecting 33 countries and more than 3 billion people. It's the longest subsea cable system ever built. Meta Engineering, Announcing the Completion of the Core 2Africa System
  • Red Sea cable cuts near Yemen (February to March 2024): three cables, Seacom, AAE-1, and EIG, carrying roughly a quarter of the region's internet traffic between Europe and Asia, were damaged after the anchor of a cargo ship sunk by Houthi forces dragged across the seabed as the vessel went down. A reminder that cable risk isn't confined to the Atlantic incidents already described in this chapter. CBS News, Ship sunk by Houthis likely responsible for damaging 3 telecommunications cables under Red Sea
  • Amazon Leo, formerly Project Kuiper (first satellites launched April 28, 2025): Amazon's own low-earth-orbit satellite internet constellation, the most direct competitor to Starlink, planning more than 3,200 satellites and aiming for commercial service in several countries, including the US, Canada, France, Germany, and the UK, by 2026. About Amazon, Amazon Leo mission updates

Recent developments

  • France's nationalization of Alcatel Submarine Networks (completed November 5, 2024): the French state took an 80 percent stake in the world's leading cable manufacturer and installer, folding a company that lays roughly a third of new global cable systems directly into government hands as cable security became a stated national priority. Wikipedia, Alcatel Submarine Networks
  • Meta's Waterworth project (announced late 2024, ongoing through 2025): a planned $10 billion subsea cable that Meta would own outright rather than share with a consortium, deliberately routed to skip Europe and China and connect the United States directly to South America, Africa, and India. TechCrunch, Meta plans to build a $10B subsea cable spanning the world

Glossary

SIM card. A small chip in a phone that stores a subscriber's unique identifier and a secret cryptographic key used to authenticate the phone to a carrier's network.

IMSI (International Mobile Subscriber Identity). The unique number stored on a SIM card that identifies a specific subscriber to a cellular network.

Cell site / base station. The tower, rooftop antenna, or small cell that a phone connects to over radio spectrum to access a cellular network.

Handoff (handover). The process of transferring an active connection from one cell site to another as a phone moves, without dropping the connection.

Roaming. Using a cellular network operated by a carrier other than a subscriber's home carrier, typically requiring a prior agreement between the two carriers.

Small cell. A compact, short-range cellular antenna, often mounted on a lamppost or building, used heavily in 5G networks to compensate for the limited range of higher radio frequencies.

Core network. The part of a cellular carrier's infrastructure responsible for authentication, mobility management, billing, and routing traffic toward the internet.

Gateway. The point in a cellular core network where traffic crosses from the carrier's private mobile network onto the public internet.

Backbone network. High-capacity, long-distance fiber-optic infrastructure that carries the bulk of internet traffic between major cities and regions.

Submarine cable. A fiber-optic cable laid on the ocean floor to carry internet and telephone traffic between continents.

Cable landing station. A secure facility where an undersea cable comes ashore and connects into a country's terrestrial fiber network.

Internet exchange point (IXP). A facility where multiple networks connect to exchange internet traffic directly with each other rather than through a third-party provider.

Content delivery network (CDN). A distributed set of servers that cache copies of frequently requested content in many locations, so users receive it from a nearby server instead of a distant origin server.

3GPP. The 3rd Generation Partnership Project, the international standards body that defines the technical specifications for GSM, LTE, 5G, and related cellular technologies.

Sources and notes

Open questions

  • Exact 5G latency and small-cell density figures vary considerably by country, carrier, and spectrum band; treat the numbers here as representative ranges rather than fixed values for any specific network.
  • Global mobile subscriber counts, Starlink subscriber totals, and the in-service submarine cable count all change quickly; the figures above are a 2025-era snapshot, not a stable long-term count.
  • Whether specific Baltic Sea cable incidents were deliberate sabotage or accidental anchor drags remains, in several cases, an open question under active investigation at the time of writing.

Knowing which tower a phone is talking to is one thing; figuring out exactly where the phone itself is standing is a different system entirely. How GPS determines a location. 👉