In a spare, drab office park in Sunnyvale, California, a bunch of two-by-fours and foamboard have been nailed together into a makeshift model of a shipping container. Inside, a bare, unlit Edison bulb hangs from a wire, over some simple IKEA furniture and a table with Lego blocks on it. The blocks are stand-ins for modules that might someday go into the Project Ara phone, which in theory will let users swap in different components on the fly instead of replacing the whole phone when it's time to upgrade.
The model is there because the people behind Project Ara are currently trying to think through potential retail experiences that would help people configure their phones. Not included inside the model are the non-invasive biometric monitoring tablets that measure galvanic skin response as a trigger to present the simplified configurator experience the team is looking into. Such technical jargon is par for the course inside Google's Advanced Technologies and Products (ATAP) group, which is in charge of Project Ara. As near as I can tell (without an engineering degree, at least), the only person in the room who needs such terms explained is me.
At ATAP, simple things like Lego blocks represent ridiculously complex ideas. This tiny group of engineers and designers has given itself the task of creating a phone with several unproven, next-generation technologies. They intend to make a phone cheap enough to be accessible to 5 billion people. To do so, they need to create an ecosystem of hardware manufacturers robust enough that it could literally challenge giant incumbents like Foxconn and even Samsung. The head of Project Ara, Paul Eremenko, says he is planning "the most custom mass-market product ever created by mankind" without a trace of irony in his voice.
He and his team have just one year left to do it.
The broad outlines of Project Ara have already been announced and had plenty of doubt cast on them. It's an attempt to encourage hardware manufacturers to build modules that will slot into a metal "endoskeleton," which serves as the basic core of a phone. The camera, screen, and any other feature that you'd traditionally associate with a smartphone would exist only as a modular tile — even the processor and the power jack would be removable.
The challenges that ATAP needs to overcome are formidable to say the least. Since well before the iPhone, mobile technology has been on an inexorable march towards integration. Putting as many parts as possible on a single circuit board — or even a single chip — has benefits that are impossible to ignore. Integration saves on battery life, weight, thinness, and cost, among other things.
"It's sort of analogous to Android for hardware, pushing the Android construct down to the hardware level."
For nearly every one of those concerns, ATAP has a new and surprising solution that's meant to address it. Many — if not most — of those innovations consist of technologies that have never been put into use in a mass consumer product. This week, ATAP is holding its first developer conference, where hardware makers can begin to learn about that technology and how to use it to make modules for Project Ara.
But first, you should know that Project Ara is not, technically, a phone. It's not even that accurate to call it a project. It's more like a mission. The end goal for ATAP is to hand off a viable product and stewardship of a hardware ecosystem to Google — Eremenko and his small team aren't just building a series of proof-of-concept prototypes; they're attempting to build an industry within an industry.
ATAP, which is also the home of Project Tango at Google, is tasked with coming up with big ideas and then actually executing them. It's not unlike the "moonshots" like self-driving cars or floating internet balloons that comprise the mysterious work of another division, Google X. Except that everybody within ATAP has a self-imposed deadline — two years — by which they need to actually create and sell real things to real people.

Paul Eremenko, head of Project Ara
It's a philosophy that's imported from the Defense Advanced Research Projects Agency (DARPA), where Eremenko and his boss, Regina Dugan, once worked. Dugan came to Google in 2012, where she was put in charge of ATAP, which was then meant to help reinvigorate the Motorola division. When Motorola was sold off to Lenovo, Google kept ATAP — and given the projects the team is working on, it’s not difficult to see why.
Within ATAP, there’s a lot of talk about the "DARPA model of innovation," which involves using cutting-edge science and technology to pursue real, practical goals rather than just pursuing so-called "pure research" or rehashing existing technologies in new forms. ATAP works with outside companies and universities to come up with surprising ways to solve problems — problems like the limitations you might run into with a modular phone.



For example, having different components physically far away from each other should mean that they're slower to communicate (since the bits need to travel farther), but ATAP is trying to push a communication standard called "UniPro" into widespread adoption a year or two ahead of schedule. Eremenko says that once UniPro is utilized by modules, an Ara phone should be fast enough to overcome that speed issue, thanks to "10 gigabits of throughput to most modules from the on-device network with a couple-microsecond latency." That's good enough for things like storage and cellular radios but not good enough for RAM, which will need to be on the same module as the processor.
Ara modules need to have a way to communicate with the rest of the phone, but physical contacts are often dirty and unreliable. So instead, the modules will use "capacitive interconnects," which are wireless and theoretically more reliable, especially at high speeds. The capacitive pads also will help save space on the modules, since they're smaller than physical pins.
When it comes to keeping the modules in place, physical latches are fiddly and can easily break. Instead, Ara phones will use electropermanent magnets to hold them in place. "It's kind of a cross between a permanent magnet and an electromagnet, in that it has an on state and an off state," Eremenko explains. "It uses an electrical pulse to switch between those two states, but it's a passive component, meaning it consumes no power in both the off state and the on state." An app on the phone will let you toggle the magnets on and off, and the 30 newtons of force in the on state should keep the modules from flying off when you drop the phone.
Project Ara is working with 3D Systems to develop a new kind of 3D printer that's capable of mass producing custom shells (the plastic pieces on the back of each module). Normally, Eremenko explains, a 3D-printer head needs to go back and forth, stopping on each side of the thing it's printing. The new system would place many tiles on a giant "racetrack," so the 3D-printer head could simply zip around the oval nonstop, without waiting for parts to set or dealing with the inertia of starting and stopping. If that weren't enough, Eremenko says the system would be "a first in terms of surface finish and color, 600 DPI, full color, and sub-micron RMS surface finish," the latter of which basically means that the system can print glossy finishes.

While it all sounds great in theory, it also sounds like the sort of project that could languish in a lab for years while the engineers work out the bugs. But that's where another piece of DARPA philosophy comes in. Eremenko says that ATAP projects "can’t just culminate in theory or PowerPoint or a lab demo. So the DARPA mantra is that we do demonstrations at convincing scale."
"We have to hit critical mass in the developer ecosystem before this is an interesting product for consumers."
For Project Ara, a "demonstration at convincing scale" means much more than producing a compelling prototype. It means getting a critical mass of module developers actually making hardware; it means actually producing the endoskeleton; and it even means selling all of the above to real consumers in real markets. Only at that point will Ara be considered a success and be ready to hand off to Google, which will handle the marketplace for selling modules going forward. Anything less will fail to "retire all the key technical business and market risks," to use the terminology DARPA adopts when trying to convince the military to adopt its technologies.
Eremenko says that the mantra was hard-won at DARPA, because just creating a prototype or a lab demonstration wasn't enough to get the military on board. "Saying 'Yeah, look, we demonstrated stealth. Air Force, here, you take it.’ The Air Force says 'not interested.'" Perhaps not coincidentally, while at DARPA Eremenko worked on projects that seem remarkably close to the work he's doing with Project Ara. The Adaptive Vehicle Make program tried to rethink how war vehicles were designed. The fractionated spacecraft program considered ways to allow spacecraft to be made up of multiple, modular, disconnected pieces — each communicating wirelessly. Fractionated phones seem relatively tame by comparison.
"There's no free lunch."
Similarly for ATAP, every technical innovation has a practical purpose. Eremenko contends that the benefits of a modular phone will become compelling to consumers once ATAP can get issues of weight, battery life, and size reduced down to a certain point, which he hopes will happen with the next prototype, sometime later this year. "We think the crossover point is somewhere at the one-third overhead point," he argues, summing up all those issues into a single number. "We think we'll come in at about one-quarter, about 25 percent overhead." Still, he admits, "There's no free lunch."
Even so, 25 percent worse battery life, weight, and thickness doesn't sound like a great deal, but ATAP is betting that the benefits of a modular phone will make them worth it. Eremenko cites the example of batteries that might offer double the usual runtime at the expense of fewer life cycles. That's the kind of thing that would never make it into a mainstream phone, because a "very large, risk-averse, fairly slow (comparatively speaking) OEM" wouldn't be willing to take a chance on it. Small-time hardware manufacturers, using open source design tools created by ATAP, could develop and sell modules at a fraction of the cost of developing a fully integrated product.

But while custom enclosures and super-charged batteries may appeal to the high end of the market, the wealthy are far from the only — or even primary — target. "What we don't want to create is the consummate nerd toy that doesn't have a market outside of Silicon Valley," Eremenko says. The other, more important targets are the 5 billion people who don't yet use a smartphone.
Next year, ATAP hopes to produce what it's calling a "gray phone," a bare-bones device with little more than a processor, Wi-Fi module, and screen. The target Bill of Materials cost would only be $50 (though the retail cost could be higher or lower, depending on Google's largesse or desire for profit). A consumer could buy and use the gray phone, or use an on-phone app to buy module upgrades or custom shells.



"No monolithic phone will ever become a 5-billion-person phone," Eremenko says, but he believes that devices based on the Ara platform could reach that market. That's where the shipping container comes in – it's just one of several different retail experiences the Ara team is considering.
"When this goes to market, this will be the most custom, mass-market product ever created by mankind," Eremenko argues. The closest thing to that level of customization is "maybe Chipotle burritos," he jokes. But the key problem is that when it comes to building out an Ara phone, people will "have to make a huge number of choices that are quite difficult to make, both aesthetic and functional."
Like the modular phone itself, choice is much better in theory than it is in practice. "When presented with choice, if it's not appropriately curated ... or presented to them, they freak out. They tend to get stressed out, they frequently seize up."
To solve the paradox of choice, the Project Ara team is looking into an on-device app, social experiences, and "friend modes" for phones, where users could hand their phone to a friend and let them "clone" their setup as a starting point. They’re also investigating non-invasive biometric scanners that measure your pupil dilation and sweat to see if you're stressed by figuring out all your options (though whether such a system ever becomes reality seems unlikely).
Google will host the store that sells the modules and also will sell the plastic, 3D-printed shells (the latter is because, Eremenko says, the antennas for various modules will have to go into the shell and Google will be responsible for proving to the FCC that they're safe and don't interfere with anything). It will not, however, sell the modules, leaving that to hardware partners. If Project Ara is successful, Google could have a business on its hands that rivals — or even exceeds — Android.
Just as importantly, though, Project Ara could have a ripple effect on the entire mobile industry. One of the goals is to "democratize the hardware ecosystem, break it wide open, basically disintermediate the OEMs," Eremenko says, "so that component developers can now have privy with the consumer."
"Democratize the hardware ecosystem, break it wide open, basically disintermediate the OEMs."
When a component maker, say a camera company or a battery company, wants to get its part into a smartphone, it needs to convince both the factory and the actual smartphone company to include it. But if they can sell the part directly to the consumer — with the help of Ara's design tools to build the module and Google's help to market it — then the Ara ecoystem wouldn’t need giant manufacturers like Foxconn, Pegatron, or even Samsung and HTC to build Android phones. "We want to empower the consumer to make those decisions," Eremenko says, "rather than having the component developer go through an OEM to do that."
It almost goes without saying that Project Ara is wildly ambitious. It’s the sort of thing that typically would involve billions of dollars and multiple years of research and development. Instead ATAP is trying to pull it off with three full-time Google employees, a handful of contractors, and outside companies contributing key parts. And it’s trying to do it all in two years, a seemingly arbitrary deadline that’s just as crazy as the technology itself. "Innovation under time pressure is higher quality innovation," Eremenko argues. "It tends to get rid of red tape, it tends to get rid of dithering, and an inability to make decisions. And it tends to take away risk aversion."
Project Ara — both the proposed phone and the organization trying to create it — is a complicated system. All the pieces have to fit, they have to talk to each other, and they have to all work as promised. That's assuming the modules get produced in the first place, which isn't anything close to a sure bet. But even if all that happens, expect the result to be somewhat inefficient, a little confusing, a tad bulky, and still the most intriguing thing to happen to phones in years.