If you follow the dollars from major technology companies and cellular carriers, it’s clear the consumer electronics world is betting augmented reality will be one of the next big things: Google has invested in everything from Glass to Magic Leap, Microsoft put most of its chips on HoloLens, Facebook is planning an all-day AR wearable, and Apple’s AR hardware project has been an open secret for years. But each company has struggled to come up with an actual product consumers want to use and wear in public — and can afford.

Nreal doesn’t have those problems, at least not to the extent of its rivals. Instead of making a $3,500 visor-slash-computer for enterprises or a $2,300+ standalone computing platform, the company created Light, a $499 consumer AR headset that looks like a pair of sunglasses and relies on an Android smartphone for most of its processing. Rather than trying to solve every possible AR problem in a first-generation device, CEO Chi Xu seems genuinely happy and excited to be polishing Light for a second-quarter 2020 consumer release — all while looking forward to inevitable future improvements. The goal, he told VentureBeat in an interview at the start of CES, is to get first-generation hardware into consumers’ hands and start collecting real-world feedback that can be used to continually improve the experience.

Xu’s thoughts on the current state of consumer AR are particularly important given how many hardware players (led by Qualcomm) and cellular carriers (such as China Unicom, KDDI, and Deutsche Telekom) are already supporting the company’s vision for a smartphone-based AR solution. And some major announcements are coming very soon. Here’s our discussion, lightly edited for clarity.

VentureBeat: Nreal went in a different — and arguably smarter — direction from rivals that decided to build complete AR computer solutions from the ground up and sell them for thousands of dollars. Why did you choose lightweight glasses backed by a smartphone? Was it internally your thought that this was the right thing to do? Or was it Qualcomm that pushed that as the direction?

Chi Xu: Google Glass is actually the first product that really tried to reach the domestic market. I think all of them have the wrong product definition in some way. For Google Glass, they were trying to get lightweight, but functionality-wise, they were too weak. On the other hand, HoloLens and Magic Leap were too ambitious; they want to replace a cell phone, replace the laptop or TV right away, and it’s not going to happen.

So why should we try so hard to do that? For our first generation, why don’t we just try something easier — complement a cell phone, not [replace it]. Suddenly it gets a lot easier. And then you realize, okay, if it’s going to be a cell phone companion, you have to make sure that it looks good so you’re willing to really carry it around with you. It has to be very easy to use with the phone, so when you plug it into the phone, boom, you have a much better experience. It doesn’t have to be everything.

Say we have the top 10 use cases that actually feel much, much better with glasses on, compared to watching them or using them on a cell phone, and then it gets bigger and bigger, until later on you realize, “Why do we need a phone anymore?” This kind of transition is what we think is going to happen in maybe five years.

VentureBeat: At CES, you’re showing a variety of different user interfaces for AR. Where do you think that the overall user interface is going? Will it be 100% the glasses looking at your hands and looking at your eyes, or external controllers?

Xu: That’s a great question. And to be honest, we don’t know yet. We have a lot of good thinking about it, but I think we have to let the consumer tell us what they really want. That’s why we want to give all the options there.

I think the first thing to do is really get our glasses to the final consumer so they can use them on a daily basis and we can collect some feedback from them, and then we can know what’s the best to do. But the interaction for the AR paradigm is going to be more complicated; it’s not going to be a single UI dictating the whole thing. It might be a combination of eye-tracking/gaze, gestures, and maybe the cell phone as well.

VentureBeat: One of the clear AR paradigms is work. Do you foresee an interaction between Light and Bluetooth accessories like keyboards such that someone walks up to a real keyboard and then interacts with it using a virtual computer screen?

Xu: At this moment, the real keyboard is definitely much, much better [than virtual typing]. We’ve actually been contacted by some companies that want to use glasses to replace a computer. Take a look at customer service, for example. Now, you have a computer, a monitor in front of you, and you have your keyboard. Later, you can just have the cell phone, glasses, and a Bluetooth keyboard.

VentureBeat: If you wanted to make a version right now that was completely wireless and didn’t require that USB cable at all, but could still communicate with either a phone or a computer, could you do it? And if so, how much would it add to the cost?

Xu: Well, we could do it. But that’s the debate we’re having internally — what is the best trade-off? To be honest, there’s no optimal solution at this moment. You have to make trade-offs. So that’s why we ask “What is the advantage, versus having the biggest weight you can put on your head?” Having only one device for this, it’s not ready, but we can definitely do that. Maybe we will do that for an enterprise solution.

In the future, you’ll actually offload most of the compute to the 5G network. Most of the computing and most of the rendering will happen in that part of the system. AR glasses can be ready to play that game.

VentureBeat: I’m somewhat skeptical that the 5G network will be ready for AR glasses in the immediate future.

Xu: China and Korea will probably be ahead of the curve in adoption. But on the other hand, we don’t know yet what we’re testing. We’re working with different partners testing different kinds of technologies in different countries, including carriers. I’m hoping we can talk about those in the near future.

VentureBeat: Hopefully, AT&T will be among them in the states. But they seem to be aligned with Magic Leap. What’s the story with that? They accused Nreal of copying them.

Xu: The only thing they have [said] is I used to work there, like three and half years ago. So I do realize that they think we are a big threat. We demonstrate that in real life it doesn’t really have to be that kind of “magic.” I think they need to convince their investors why the smaller company using so little to do this has a better experience. It’s hard for them to justify that. And then it’s sad because initially, both companies were thinking innovation is the key to driving the whole industry.

VentureBeat: Is it true you have funding from the Chinese government?

Xu: That’s the impression that they try to put on us, that we’re backed by the Chinese government, or we’re stealing the technology. It’s not always the case that all Chinese engineers are stealing technology. In our case, they don’t have any proof and real data. They’re saying, “You guys are doing this so fast. You’re so small.” There are no specifics. The latest update is we just filed a motion to dismiss all of their complaints.

VentureBeat: Your display technology is different, right? You’re using a “light guide” projection system instead of waveguide, which is capable of displaying imagery at multiple focal depths?

Xu: For waveguide, the main reason to make it is not to get multiple depths, but to try to create a [conventional] glasses-type form factor product. In my opinion, I think that the ultimate solution is really to make a process like this, but it has a long way to go.

The whole community, what they’re trying to do is put semiconductor-like technology into the optics. That would make sure for the next 30 years this industry is going to explode, because look at the semiconductor — Moore’s Law … You can bring Moore’s law into the tech, but it’s really hard. If you look at the whole semiconductor supply chain, it’s very long and it takes a long time to really get there.

Three or five years ago, when people first got into this industry, they were really optimistic about waveguide. They thought they’d figured out the whole process, and the whole chain was ready, the only thing that was not ready was the design.

It’s really not going to be anytime soon. Three years ago when I left Magic Leap, the general assumption from the industry was, “It’s going to be ready in five years.” Nowadays, I think it’s going to take five to 10 years. Look at what Facebook has been saying about their solution. Five to 10 years to get to 100%. Because they’re waiting on waveguide to really mature to that level.

Nreal will be making some major software announcements this evening, and after an early hands-on, we’re hugely impressed by how its consumer AR offering is shaping up. Stay tuned for the news.