In the last couple of years, anticipating the launch of 5G, governments have been licensing new tracts of wave spectrum real estate at higher bands than in the past—in particular, a midband range between 3GHz and 6GHz, and highband slices that occupy the 24-52GHz patch, a level of spectrum known as “millimeter wave”—mmWave, for short. 2

Transmissions at the lower end of the wave spectrum can travel long distances and zip through thick building walls like they were laundry hanging from a line. These dynamics mean the signal coverage area is broad and reaches indoor users with ease. But these low-rolling wave bands carry relatively little data traffic. With overall bandwidth per user pretty low, it’s best suited for areas with few people—for instance, “it’s brilliant in rural areas,” says Nokia’s Twist. (Suburban areas, meanwhile, are better suited to midband coverage, at the 2.5-3.5 GHz frequency.)

These tradeoffs invert the higher up the spectrum you go. Since the shape of the wave means more signals can travel simultaneously, bandwidth abounds. For instance, the mmWave level covers something like 100,000 people in two Manhattan blocks. And connections are ultra-fast—up to 10 Gigabytes per second.

Sounds pretty great, right? So why, then, are we colonizing mmWave only now? The simplest answer is that, until 5G, we didn’t know how, at least, not at scale.

“The challenge has always been how to use mmWave spectrum in a mobile network,” says Alok Shah of Samsung. Handling these frequencies is tricky, he explains. Though highband can support more signals at once, those signals are way feebler. That makes their range abysmal: mmWave signals travel at most 500 meters from the base station, says Nokia’s Twist—and sometimes die out within 200 meters.

They’re also easily thwarted. For instance, mmWave signals can be blocked by obstacles as seemingly unimposing as tree leaves and human torsos. (Pre-war apartments and office buildings? Forget about it.) They’re also fair-weather signals: mmWave transmission dissipate in humid air, fading even more dramatically when it rains.

As a result, this frequency was better suited to transmissions sent by, say, oil and gas platforms back to processing facilities on land, says Shah. Even then, connecting these points required meticulously aligning antennas. Doing this at the scale required of a commercial service demands the computational power to process the signals emitted from zillions of small antennas. Until recently, says Shah, the processing power of chips simply couldn’t cut it.

Existing network setup wasn’t designed to work with these swaths of higher spectrum. Simply popping up more cell towers is expensive and, in the long term, won’t do the job. We need new technology to bend this unwieldy spectrum to our will and use it efficiently.

The genuinely transformative applications of mobile connectivity will be based on the new 5G architecture that uses spectrum above 6 GHz, according to IHS-Markit. The crowning glory of 5G, however, is its ability to use all those spectrum bands—high, mid, low, whatever—to target its services to customers’ needs. This is where the more awesome technological potential starts to come into focus.

LET’S TALK BEAMS

The revolution within 5G itself

Though inevitably kind of wonky, the innovations needed to get us there are useful to understand because they illustrate why 5G’s transformational potential is so much greater than that of past mobile wireless generations—and why it’s going to take a good few years to pull off.

Beamforming

When industry folks try to describe how mmWave works, talk inevitably turns to flashlights.

Picture a run-of-the-mill Eveready. Flick it on and its beam illuminates the 90-degree patch just in front of it, fading out within a few feet. Current signal transmission technology kind of works like that, by spraying a signal broadly but diffusely across the entire frequency channel. That’s well and good when there are few users. But the more activity on that frequency, the greater the noise—which makes users’ connections sluggish and unreliable.

Now recall what happens when you put your hand over most of the flashlight aperture. The beam that escapes is narrower and travels farther. If you turn on a bunch of other flashlights around you, narrowing the beam in the same way, they’re much less likely to overlap with each other.

This analogy is key to grasping “beamforming,” how 5G radio units extend the range of signals, says Heidi Hemmer, Verizon’s VP of technology.

“The thinner beam [comes from] 5G taking what used to be wide beam that talked to many people in the same place,” says Hemmer, “and dividing those into individual beams, one to one connections with a particular phone—by decreasing any interference.”

This beamforming is also intelligent. AI algorithms that locate and track users help antennas volley signals quickly and precisely. There’s less of a chance that they bang into each other and slow up each others’ signals. That ups the overall number of people that can use that frequency at once. It’s actually very similar to how a lot of smart speakers work.

In practical terms, forming these beams requires many, many little antennas. (The higher the wave frequency, the smaller the antenna). They’ll be sprouting from places they’ve never been before—“city furniture” like lampposts and streetlights. (To start with, anyway. Engineers are working on incorporating miniature antennas into a number of less obvious objects, including manhole covers and bus stop shelters.)

To get started, operators are blending 5G antenna technology with existing 4G. “What we can see is that the evolution of 5G is actually taking 4G with it,” says Nokia’s Twist. “They will coexist for probably the next 5-10 years. It will be a few years before 5G becomes dominant for data carriers.”

And carriers still have a lot to work out, as a Wall Street Journal survey of 5G service in the US recently detailed (paywall). “Everybody is talking about theoretical speed [of 10 Gigabits per second], but this is what you get when you’re talking very close to [the radio antenna],” says Stephane Teral, telecom analyst at IHS-Markit. “If you move just one foot away the experience already changes dramatically.”

Beamforming isn’t the only critical technology that needs to get the kinks worked out, he says. “It’s going to take some time to get our act together. We have lots of moving parts in brand new 5G architecture.”

What, exactly? By replacing the hardware-based core of cellular network architecture with software, 5G will eventually be dramatically more elastic, agile, and powerful than anything we’ve ever seen. As with the case of beamforming, the key to spiffy new hardware actually lies a good deal in the software—the processing power and algorithms to manage the data.

Virtualizing the network

When the 5G buildout is complete, we’ll stand back and find ourselves looking at an entirely new network architecture—made up mostly of software. That makes running a network cheaper, of course. But it also lets carriers take network functions that are currently geographically fixed and shuffle those functions around.

5G’s predecessors didn’t really need that kind of flexibility. “In the past, we were predominantly doing one thing: accessing apps on our phones. All of us had a data plan,” says Ericsson’s Linder. “Everybody got the same thing—a bit of data they paid $50 per month for, and they accessed applications in the cloud. Everyone did similar things.”

But 5G’s goal is to support a much richer diversity of customers and types of connections. To build a network that supports dense thickets of simultaneous connections, service operators are taking many of the jobs that used to be done by hardware—usually located in little rooms at the bottom of towers you see dotting the skyline—and assigning those instead to new software that will run off the cloud. “As we now move to the world where performance and delays are more critical, then we have to start moving out functionality closer to users,” says Linder.

Edge computing

Right now, the physics of 4G means that there’s a slight delay in the amount of time it takes for a device to talk with the network. You’ve undoubtedly noticed this. “If you go to the concert you’ll see that what’s on the screen is happening a little behind what’s on the stage,” says Nokia’s Twist. Known as “latency,” this delay, the time it takes for your device to send a signal and receive a response, depends largely on how long it takes for the signal to travel from your device to a cell tower, where the signal is processed, and back. The further your signal has to travel, the less reliable the response. Sometimes this delay is perceptible—for instance, when the video you’re watching stops to buffer. But even imperceptible delays are a problem, making it almost impossible to do anything that’s fast-moving and interactive. Right now, under optimal 4G conditions, the lowest latency is 10 milliseconds (1/100th of a second), but 40 milliseconds is more typical. That’s too great of a delay for our brains to process. In fact, using VR headsets via a wireless connection where the latency is too high is known to make people seasick.

It also prevents us from using wireless connections for what businesses deem “mission critical” activities. For instance, you obviously wouldn’t want your remote surgeon to suddenly pause to “buffer” in the middle of an appendectomy, or for your robot driver to lose connection while doing 65 down the interstate. Or consider a factory. Automakers can’t run the risk that a robot fritzes while tightening a brake bolt. Even with the fastest 4G connection, anything that requires a seamless experience provided by an undisrupted flow of data is off the table. 5G aims to change that.

A big hope for guaranteeing low latency lies in what’s termed “mobile edge computing.” The new 5G architecture will dispatch heavy-duty processing power that used to sit at the core out to base stations at the frontiers of the network. That will put computational heft and applications that can run off that processing much closer to users, so devices and the network can swap signals far faster. Experts expect 5G to deliver latency of between four milliseconds, for regular connections, and even as low as one millisecond for “mission critical” applications.

Mobile edge computing will also let users to store a lot more data locally and run applications closer to home, sending data over the network only when necessary. That will support things like high-resolution surveillance cameras, which collect super-huge volumes of data but only occasionally need to send out alerts or otherwise interact with the network.

The innovation that brings all this together—harnessing the combined benefits of higher spectrum and flexibility-enhancing software—is perhaps the most important, and least technically proven, part of the 5G technological arsenal: something called “network slicing.”

Network slicing

The way things work now, each customer is assigned their own data channel, regardless of how much bandwidth they use. That lack of flexibility means lots of bandwidth lays idle. Network slicing lets operators partition a single network into virtual mini-networks, allowing them to accommodate users’ diverse data needs—whether that’s lots of bandwidth, for instance, or perhaps a super-reliable signal.

“Up until now [we’ve had] gigantic highways in air with no paint whatsoever to separate different lanes,” says Linder. To follow the metaphor, 5G technology lets operators paint lanes onto that spectral tarmac, reserving lanes for critical applications the way you might an express bus lane, while allotting a “bike lane” for traffic that doesn’t require extreme speed and reliability, like mobile internet browsing. What’s more, the technology is dynamic, says Linder—meaning operators “can change the paint” as needed.

In practice, this will mean that the data of, say, a tween streaming HD video will flow alongside the transmission from a team of wireless assembly robots and a fleet of driverless buses, all through the same network.

NOW FOR THE MAGIC

Bringing it all together

Getting all these pieces of technology to fit right will take time. And there are plenty of logistical hurdles—banal things like getting site permits and locating power sources. Paying for all this will also be hugely tricky. Installing the dense network of small cells that 5G signals ping around on will demand heavy spending on fiber optic cable. And at the same time as carriers are shelling out on that—and other 5G infrastructure—they will have to keep buying new spectrum while continuing to upgrade their existing 4G networks, note UBS analysts. Estimates of the amount of global investment 5G will entail top $1 trillion. That expense means a lot depends on how quickly operators are able to sell their services.

But when all that comes together, it should achieve three things: boost bandwidth, allow low latency connections, and expand the capacity to support hordes of devices connecting at once. That process also depends hugely on when entrepreneurs pioneer models that transform different industries. As that discovery starts unfolding—probably about five years from now—we’ll find ourselves entering the second phase of the 5G transformation, the one where the reality starts to overlap a bit more with the hype. And while all that’s nice to have for consumers, it opens up a range of heretofore impossible ways for businesses, utilities, and local governments to automate the way they operate.

Okay, but how?

5G technology is designed to serve three major “use cases” that form something approaching the holy trinity of 5G marketing hype and jargon: enhanced mobile broadband (eMBB), massive machines type communications (mMTC), and ultra-reliable low-latency communications (URLLC).

The holy trinity of 5G

The launch of 5G will happen in two main phases that loosely correspond to the difficulty of each corner of the holy trinity. Generally speaking the eMBB will be enabled in the first phase, which harnesses 5G frequencies via 5G-specific smartphones while still using 4G infrastructure as the core network.

The initial “use case,” eMBB, will be only slightly more exciting than it sounds—at first, anyway. Consumers will get faster data connections on their mobile devices. Of the consumer applications that will be possible, the coolest involve VR/AR and gaming. 5G makes possible “higher speeds and a totally new high resolution experience,” says Samsung’s Shah. “Your car becomes connected and you can take advantage of high data speeds to do next generation navigation and do 4K video in the backseat.”

Both Verizon and AT&T have gone whole hog on sprucing up football stadiums with 5G. (Verizon has linked up 13 stadiums and counting.) Right now it means spectators (at least, those with 5G handsets) won’t suffer from the normal mass gathering network congestion, and can enjoy regular streaming speeds. But more is planned, says Verizon’s Hemmer. By next year, spectators will be able to watch plays from multiple camera angles on their phone, and see a player’s real-time stats as he runs down field. They’re also working on integrating AR/VR technology so that someone sitting in the stands watching the game can perceive a player sitting next to them in the stands talking to them, perhaps like a sort of pre-kickoff Jiminy Cricket.

The far less sexy part of Verizon’s 5G offering is home wireless connections for people in areas where cable or broadband internet companies can’t go (or simply haven’t gone). More generally, 5G experts say better spectrum management—and deployment of mmWave technologies—will make fixed wireless access faster and more appealing.

Still, these improvements are unlikely to be so dramatic that the vast majority of people’s behavior will change all that much. And eMBB is the relatively easy part of the holy trinity—what can be offered in the early years, while still building on the 4G network.

It’s what comes after, with the more sophisticated elements of 5G technology, that takes 5G beyond the consumer-directed benefits of 4G and empowers businesses to radically overhaul automation and cities to harness intelligent machines to manage our daily lives.

The shift from evolution to “revolution”

What 5G marketers call mMTC—”massive machine type communications”—is better known as the Internet of Things, the menagerie of objects that 5G’s ultra-zippy connection will empower to communicate with each other. The wide new frontiers of capacity that 5G opens up will allow a mindblowing number of these devices to be linked at once. The general expectation is that the 5G network will eventually connect as many as 1 million devices per square kilometer. (To get a sense of that density, imagine 3.5 million devices connected to each other in an area the size of Central Park.) That compares with the 10,000 per square km maximum possible on 4G.

Health care could change radically. Sensors will be able to monitor heart rates, blood pressure, and other bodily functions, as presaged by smart watches. With AI, they’ll be able to predict heart attacks and other emergencies.

Homes will become much “smarter.” But the revolutionary potential probably won’t come from our homes. The real potential lies in smart cities and automated factories and the like. 5G experts say this will turn cities “smart,” by powering super-efficient policing and traffic control, among other things. Ambulances will be able to whiz to hospitals, thanks to smart traffic lights—perhaps with an ER doctor helping the patient via holograph en route, says Twist. Eventually, 5G networks will allow surveillance cameras to gather high-resolution video feeds, analyze them on the spot, and dispatch alerts about unusual activity.

No one knows exactly what this will look like, of course. Really, anything you can pop a 5G chip into can join the ranks of networked things. For instance, forecasts of the number of devices that will be online by 2023 range from 500 million to 20 billion.

Most of these devices probably won’t be consuming huge amounts of data. So why do we need 5G? To support an internet with a massive scale of things, networks will need way more capacity. The flood of things onto wireless networks will also require software platforms to manage and coordinate them. The smarter and denser these networks become, the more we’ll need sophisticated analytics powered by machine learning to manage their interactions.

Most of the futuristic use-cases of 5G hype involve “mission-critical” latency—a.k.a. ultra-reliable low-latency communications (URLLC) services. This final element of the 5G holy trinity will eliminate these disruptions by shrinking the buffering speed of data transmission to less than one one-thousandth of a second, down from the current rate of between 10 and 40 milliseconds—as Nokia’s Twist puts it, “effectively, almost instantaneous.”

That will allow things like autonomous vehicles, drone-controlled firefighting, hologram interactions, AR/VR technologies, advanced industrial automation, robotic gloves, and a touchable internet. Interactive retail becomes possible—something Verizon and its competitors are already working on. Hospitals will benefit from ultra-reliable connections, says Twist. So will industries that involve dangerous chemicals or inhospitable environments (mining, for example). But it will also empower more mundane things like utilities and critical public infrastructure to become smart.

The possibilities here are truly myriad. Some are already happening. Nokia’s says that at one of its factories in Oulu, Finland, wireless robots running off its private 5G network (or something very close) are already producing 5G base stations. Verizon is working on using AI and facial recognition to help it predict the location of abducted children and get images to nearby police officers, says Hemmer, the company’s technology VP.

There’s something else to keep in mind here: The aftershock of 5G’s impact—the innovations and applications that companies build on top of these new capabilities—will almost certainly be bigger and more transformative than the arrival of 5G itself.

For instance, the oft-cited example is how Uber took advantage of the continuous connectivity made possible by 4G/LTE to completely shake up urban transportation networks. But again, thinking within the parameters of what’s possible with a smartphone will miss the real potential.

“What was really stunning about 4G, and I don’t think anyone could have predicted, was the idea that the smartphone became a platform for just massive amounts of innovation and transformed the way we live our lives. On demand services, OTT video [“over the top” video, like Netflix or Hulu], all these different things took off thanks to 4G connections,” says Shah. “In 5G what’s really interesting is that the innovation ecosystem extends beyond the smartphone into the rest of our lives. We think it’ll bring innovation around connected cars, smarter cities. It will allow us to connect everything—and understand how to use that in a way that improves people’s lives.”

HOLD ON A MINUTE

Transformation not guaranteed?

Or… not. Amidst all this heady hype, it’s easy to lose sight of the fact that the 5G transformation depends largely on whether businesses and cities seize on the technology—and pay for it. Given how expensive the buildout will be, 5G could be a big gamble, says Tim Wu, professor at Columbia University Law School and a thought leader on technology policy. “I think the question is not ‘Are there desirable big changes?’ but ‘Are they 5G dependent or not?’” he says. “That’s the big question.”

Take driverless cars, for example. The technology is probably pretty commercially important, says Wu. “But I’m not sure the case has been proven that you need to spend $1 trillion on 5G to have driverless cars,” he says. “In fact, there [already] are driverless cars. Their problem is the AI is not good.”

He sees the risk of focusing too heavily on 5G as similar to Japan’s obsession with the supercomputer in the 1980s.

“What makes it similar to me is there was this sort of cognitive error at the center of it, which was this idea that since we have computers, we all know the next obvious next step is a supercomputer. So we need to double down and invest everything we can in getting to the supercomputer first,” says Wu. “Japan took that idea pretty seriously. They went all in on the supercomputer. And they set themselves back 20 years because they never developed personal computers, which ended up being the actual future.”

The track record of the previous “G”s doesn’t instill a lot of confidence. Then again, their impact fell mainly on consumers and how they used their leisure time—think Instagram, shopping on Amazon, playing Candy Crush or Pokémon Go.

With the benefits it offers to businesses in particular, 5G could finally translate leaps in wireless connectivity into commensurate economic returns. Tom Wheeler, former head of the FCC under president Barack Obama, anticipates something even bigger. The world has seen two major network revolutions, he says, “the first being Gutenberg [the invention of movable type], the second being the railroad and [the parallel spread of] the telegraph, which evolved to give us telephone and beginnings of internet.”

As Wheeler argues, we’re currently on the cusp of the third.

After the “revolution”

Once you grasp what sets 5G apart, the basic economic logic of “revolution” isn’t hard to see. When you put sensors in everything, the scale of data these devices observe and record are the inputs that AI transforms into something new entirely: intelligence. Instead of merely transporting information, the new network will coordinate the machines that apply this intelligence. The potential to harness that will spur corporate spending, boosting workers’ output—something much needed after decades of anemic investment and stagnating productivity growth. Within a decade or two, the changes set in motion could rival the scale that occurred in the second half of the 19th century.

Perhaps not surprisingly, the 5G optimists tend to gesture toward something even grander. Take, for instance, Twist’s prediction that “5G is going to shift the balance of power from technology back to people.”

“At the moment you are constrained by having mobile phone and wifi coverage—you live your life according to the availability of technology,” he says. “5G will reverse that. No matter what you’re doing, what you’re trying to do, instead of people adapting to tech, tech adapts to people.”

It could be that digital connections equalize the growing rifts in prosperity between urban and rural areas—the world-flattener that globalization proved not to be. Ericsson’s Linder reflects on where he grew up, a Swedish town of 5,000 or so people, too small to merit its own high school or even a traffic light. “I wanted to get a university degree so I had to move away for four or five years. When I was done with that, then I had to move away again—to Stockholm, [to get a job], and then around the globe. I was lost in space from a resource that could be meaningful for my home town,” he says. “I think what we’ll see is that these new ways of learning [enabled by 5G] will open up ways for education to move to where people are, not people moving to where education takes place.”

“We talk a lot about smart cities. But we don’t talk too much about clever countryside,” Linder adds. “This is what it takes to make that happen.”

Is that naive? Maybe. Then again, throughout history, big technological advances in how we disseminated information have always massively shaken up society, says Wheeler, who analyzes this process in his recent book, From Gutenberg to Google: The History of Our Future.

Of course, the upheaval in question wouldn’t be good for all—or even, necessarily, most. When we talk about wireless robots and autonomous cars, we’re talking about replacing workers and drivers. That’s the point. Economists will assure you that automation merely entails shifting workers into different jobs. But conventional economic logic won’t necessarily apply.  The intensity of automation that 5G optimists are describing is on a scale bigger than anything society has ever experienced. The social upheaval caused by mass unemployment could be similarly unprecedented.

Compounding that could be how technology changes the nature of the jobs that remain. 5G could challenge the very notion of what we think of as human skills and the value we put in know-how, and even, maybe, learning itself. “Today we’re moving into a world that’s more and more learning intensive. The challenge for any business is keeping staff up to date as the world changes around them,” says Ericsson’s Linder. He hypothesizes that AR will allow workers to master tasks as they do them—kind of like how you might watch a YouTube to install a door knob now, but faster and more fluid. That could, in theory, obviate the need for firms to invest in worker training or even, maybe, the value of advanced education.

In a way, the vagueness of 5G hype reflects this gaping uncertainty about 5G’s impact on our future. And something else, too: The only sure thing that 5G’s arrival heralds is that the technology to radically reshape mobile networks has arrived. That’s a reality worth grappling with. For, as Wheeler puts it, “The networks that connect us are the forces that define us.”