Smart Buildings at Billion-Square-Meter Scale: Inside China’s Massive Real Estate Digital Transformation

So Sean, welcome. Thanks for spending the time so late in your day. Thank you. Thank you for inviting me. All right. Let’s talk a little edge AI, but, but first set us up. Tell us [00:02:00] about space tech, the technology, the business model. And as I understand it, it’s a services unit of Vonka, if I’m pronouncing that correctly, which is a residential commercial real estate business.

Sean: Let me quickly introduction Wanky, or the Wanka the Wank Wanky is a hundred billion dollar real estate developer including the residential, including the commercial buildings developers. And VanWall is actually technology portion of the Wanky. And basically developing smart buildings, smart community, everything attached to that the, the, the, the let’s say.

8 million condos. Wow. Roughly 2000 commercial buildings, you know, 8 million condo plus 2000 commercial buildings. That’s, and including the IBMS intelligent building managing system, including the door keeper and the, the, the, the. [00:03:00] And also including surveillance and energy, you know, saving a lot of those things.

So that’s the one world space tech, what do we do? And, and on top of that technology, we also providing service. And because once you get an intelligent building managing systems, somebody need to operate it, somebody need to, you know, make using that. IBMS to run the buildings, run the electricity of the building, run the water supply of the building, run the, and the, the, the, the air conditioning, all that stuff that we provide as a service also, so provide a technology solution, the, that’s the left hand, right hand, also providing the service portion that we providing all the service.

Property service, including the, all the mechanics, including all the electronics, 2, 000 commercial buildings and including 8 million unit of [00:04:00] condos. That is an enormous housing stock. Huge. It’s a 1 billion square meter. Can you imagine that? I can’t, I can’t even picture that. So, so you, in China, basically everybody got a one square meter average for even for the population, which that’s huge.

Brian: So, so did the services, did Vanka create the services unit or was the services unit a separate company that eventually got acquired? How did that work? Vanka created a service unit. Well, I have to say that’s pretty visionary if you think about

Sean: it. Tremendous visionary. And I think they are learned from Sony.

That makes sense. Yeah. Sony also did the real estate developer in Japan.

Brian: Yeah. I can imagine, maybe not quite as big as,

Sean: no, the less than one tenth of one key.

Brian: So they create this unit. It’s [00:05:00] it’s vision is to bring smart technology to a billion square feet of space, square meters, sorry, billion square meters of space.

And then they need a smart guy to come in and work with the technology. A guy who’s been at Huawei mm-hmm . And who’s been at Alibaba and knows IO ot and voila, that’s you tell us, tell us how, how you got involved with this, this business and the, the then I

Sean: actually worked for Intel I OT division before also.

Oh, actually, okay. The iot division, it’s a mobile division and the, from 2010 to thousand thousand 14. And the Intel in the the, the, the closed down the mobile phone division. So I, yeah, it was the, the, the sensor hub portion of the mobile phone division. And the and I joined Huawei as [00:06:00] a chief scientist of the Huawei sensor application lab.

Basically doing the, the, the, the smart sensor in the mobile phone, smart sensor for the smart home and also smart sensor for the wearable. So, and, and, and I worked in Huawei for three years, then Alibaba IOTBO inside of the Alibaba cloud BG. So they hired me as the CTO of the IOTBO. So in the, the Alibaba cloud, I need a, you know, a team of 450 people and build the IoT platform, including the edge, including the terminals, including the, the IoT platform in the cloud.

So then. We also service different sectors, including smart home, smart communities, smart building and a smart city and a smart industry. So I worked for Alibaba four years and the plus three months and the, in the middle, [00:07:00] almost the end of the fourth year and the bank, he say, Hey, I’m doing the smart home, smart community, smart building.

So why not do you working for me for us? And I, you know, that time. I thought about it. So, so, so actually I basically, if you look at, at me, I leaving in the technology sector more and more close to the end user sector. So, you know, the intel is really similar to and then, you know, Alibaba is a little bit close to, you know, only clouds, a little bit too close to the end user and and further them the technology sector.

Then when I working for Banky, finally, I working for the consumer directly. So, right. So the, yeah, so that, then I, you know, last three years I learned a lot. How to service and using technology serve those and end user. So including, because I working for the chip company, because I’m [00:08:00] working for the cloud company, because I, and have understanding those technologies.

So now I really appreciate how the computing service, the, the, the, the consumer in the smart home, smart city, smarter building, et cetera. So then I. Build a completely, we call the IoT native platform. It’s more than cloud native. It’s very similar to cloud native, but you know, we want to build the edge native, edge cloud native, and terminal cloud native.

Because once we link the 8 million smart home, so every device, every computing device you can treat is a small little Computing part, just like a cloud, you’ve got 8 million clouds, right? That’s what we want to do, man. That’s what we want to do with Arm, because Arm is the most power efficient architect in the world.

Brian: Yeah. All right. So, so we’ll get, we’ll get to that in a second, [00:09:00] but so here you come with this, with this really nice experience chain into this environment, talk to us about building out. The vision, the architecture to achieve what you wanted to achieve. I can’t imagine that it was simple.

Sean: It’s not a simple, it’s a it take me almost you know, eight years in a building between, you know, the, the, I work for Intel and then working with Huawei.

Intel is a chip company. And the during Intel, we always know smart building, smart home and the cloud native, blah, blah, blah. Exactly. And I work for the Huawei. And we work for the hardware company. And the hardware they end up building the, the, you know, the edge node and building the sensors, et cetera.

And also I spent four years with Alibaba Cloud and that, and together is 10 years. So during that 10 year struggling are pretty struggle. And always try [00:10:00] to figure out how to, using the new technology to transform the traditional business like the real estate development. The literacy is huge in China, and it’s roughly we talk about the, the, it’s a 1 billion square meter in Wankee, but the total.

In China is 30 billion square meter. Wow. I worked for the U S for 15 years. Then when I get into China, I always think of the number is wrong. So maybe by a factor of 10 and also China using the Chinese number is always slightly different than the U S U S is like a thousand and a million and a billion, the China is using then.

And the then, then, you know, the four zero, then eight zero, then twelve zero. So when we always compare to eight zero and to nine zero, so we always [00:11:00] missing one zero. One billion, then one E. So we were thinking, okay, I’m wrong. You’re getting the number wrong? No. Actually, when we double check, the number is right.

Thirty billion square meter is like a 330 square feet of real estate in China, that’s the sheer size. And we think that was roughly, well, let me think about it. 200 trillion RMB, 200 trillion, roughly 200 trillion, lots of money. So technology, how we can enter that massive market? That’s what we, that’s taken me 10 years, including Intel, including Huawei, including Alibaba.

So we did a lot of smart home, a lot of smart community, a lot of smart cities during I was in Alibaba. [00:12:00]We think, we think about a solution. We think about the technology. We think about the DevOps, all those things, how we can enable that industry. So it’s, it took a long, long time. Then once we get into Vanki, I finally riding the horse, then now I understand what the horse looks like.

It’s a task, it’s a

Brian: task, it’s a task. Yeah, but, but you have that, so you come into this situation where there’s clearly a vision to transform property management, but yet you have 50 years of home and building automation history that is littered with. The carcasses of companies that had troubles with cost, standardization, reliability, right?

The Scotts started it with the X10 in the 1970s, and that was going to be great. But I think that was a power line solution. So that didn’t quite work. And then you [00:13:00] had Echelon some 30 years ago. So talk to us about. How you decided to ignore all that basically, and just create what must be sort of a bespoke architecture.

To tackle this problem.

Sean: I think the, you know, this is basically, if you look at a smart car, the same thing, you know, the traditional cars, if you look at the Ford, GM, Toyota, they have to talk to 150 vendors to change their software, make their software work together. Smart home, same thing. So of home, you know, smart building is same thing.

So Brian, just like you said. There are so many different siloed systems and the obsolete technology, right? So they don’t talk to each other [00:14:00] in one building even. The lighting system do not talk to the air conditioning system. Do not talk to the air refresh system so that Right. So the, you know, if you feel, you know, air conditioning is the violating either pumping a lot of air conditioning but you still don’t feel fresh because they don’t work together.

So that’s tremendously. And I’m building need a 10 to 15 guys sitting on the 10 to 10 to 15 guys on a different terminal to control different system. You need a lighting control, ma’am. A security control man, the firefighting control man, air conditioning control man, then the elevator control man. That’s, and each one, you need a 15 person cost a lot of money.

Yep. And they don’t talk to each other. Then make people feel really [00:15:00] uncomfortable. The software doesn’t work each other. So then we figure, hmm, we shouldn’t using the continuity technology. Running on one server, make the 15 software on 15 containers running on one server. Then the server, then we connecting the different application, the different container together into one data lake.

We need one man to operate the entire building. And we get a Data to talk to each systems. You talk to me, I talk to you. So, and the then the AI can really maximize all the data petition.

Brian: So from the edge to the cloud is your basic domain. Did you have to write your own software to consolidate all these various applications?

How, how did that work?

Sean: Basically what we need to do is using the [00:16:00] cloud technology and moving that technology into. The Edge virtualize the edge server and creating the containers. Then the Edge operates like a mini, mini cloud, you know, 15 application running, one server. Then you can on, on top of that application, you need to can using one piece of software and the, the cut through all those 15 applications or links, all the 15 applications.

So we call the common object, the common data object. Okay.

Brian: All right. So talk to us about the hardware approach. You’re using Arm at the edge. What was the motivation behind that decision to go with Arm versus Intel versus whoever? Yes. I think of the, the,

Sean: the, the, the advancing really important things. And I think of the number one, that used to be the, the ash, the.[00:17:00]

In the buildings, there are two different Edge. One is Gateway. The Gateway is already using Arm. And one is Edge Server. Edge Server is used to be dominated by x86, the servers, server chips. And why? Because x86 server is more Compatible with cloud technology, including the container technology and the virtualization technology.

So today, why are we using Arm, adopt Arm? So we think two things or three things are really important. Arm is better than x86. The number one, we have better, better and a bigger development community. Right. That’s really important. Almost, almost in China, we have. Three to 5 million embedded developer using Arm.

That’s a big number. That’s educated by Arm and STM. Right. [00:18:00] So the ST micro wrote all the textbook for China, all the universities. So because of all their controller line. All the controllers. So that’s, that’s, everybody is being trained by either microcontroller or AP. Application process. That’s a large number of development.

That’s really important. Number two, of course, and the power and the price is way better. Okay. But one thing is we have to point out is the power because in the edge room, there is not like the server farm. Server farm is nobody there. But in the edge room, there are people sitting right next to the cloud and edge server.

And if you have a huge fan and the people will be annoying to death. Basically, people only can resist the annoying noise less than five minutes. Yeah, that’s longest limit. No normal person [00:19:00] can resist. If you have people working in the the edge server room and you have to control the security camera, you have to control the firefighting.

Then, then the edge server running very noisy, just like cloud server. You, you basically, tomorrow, you know, the, you will resign and don’t

Brian: show up anymore, right? So, so, so no fans or, or fewer fans or less noisy fans, but also energy consumption. Is a problem, the world over. And you, your, your properties must consume a lot of energies.

Any savings are go straight to the bottom line, maybe. Yes.

Sean: I think the, the, the, the, the low power is very critical. The fan less is or less fa. We need to control the edge room less than 50 db, 50 db. So that’s the most important. So that’s why the edge room have to be low power. That’s the number [00:20:00] one critical reason.

And of course, you know, because we are in the edge server room, so we still can much better than the battery powered device. And we can, using better, still we can using little bit more power, but then we still, you know, saving more power is better. Yep.

Brian: So. You have a PhD in AI, did I get that right? Okay.

Let’s talk about, let’s talk about AI. How are you guys integrating AI workloads into this network, into the edge servers, into the edge devices to improve the user experience?

Sean: Yeah, well, I think the, each building we have maybe 2000 cameras. So just like in my office, there is a camera outside. And we were using the camera, we watch if somebody, nobody in the office room, then we turn off the light and turn off the air conditioning automatically because the air conditioning system already in the edge [00:21:00] server, the lighting system is in the air conditioning server, then we look at using the analyze the camera, they see, okay, nobody moving, nothing here, everything, nobody there, then we turn off the lighting, saving quite a lot of the energy.

And we also, and and the, the, the, the, the running AI in for the air conditioning systems, we, we know. And most of the people, and we use the history, we know, say, most of the people show up at 8. 30, then we turn on the, the, the air conditioning at 8. 15, so make sure everybody coming in is comfortable.

We know most of the people leaving at 6, then we turn off the air conditioning at 6. 15, then saving air conditioning, all that. And we know During the weekend, the only people to show up floor seven and floor eight, then we only turn on the floor eight and floor seven. That’s the ai how learn, the history then to helping optimize the air.

Then that tremendously [00:22:00] saving power. Actually what we, what I just said is actually easily can be implemented by one people. If we send one facility man and go there and searching around and turn around, turn off the air conditioning easily. But nobody can remember as no one person so disciplined can do that.

Oh, the human memory, the fallibility. Yeah, today you are disciplined. Tomorrow you are disciplined. Sunday you forget. Yeah, right. But the AI can discipline forever.

Brian: Absolutely. Let’s, let’s do a scenario here. I’m, I’m a, I’m a customer, I’m a renter, I’m an owner of some few meters of your 1 billion meters, and I’m outside my building.

Talk me through how I’m going to interact with the building as I [00:23:00] enter it, go up the elevator or park my car, go upstairs, go into my domain. Talk about the digital experience.

Sean: So I think the, you sort of, almost everything is digitized when you walk into the door and then the two things you can do, either you can, you’re using the, the iMetrix you, for example, the fingerprint or the palm print, palm print.

Or, or either if you want to, you can use face recognition, but most people should choose using either fingerprint or the badge. So once you come in and the door will automatically recognize you with your badge, then the door automatically open up, then automatically call the elevator, tell you you which floor you are.

Once you walk into, walk across the doorkeeper, then you punch your card automatically. So basically, you don’t need to punch your card anymore, then you punch your card automatically. [00:24:00] Then, you are, the office, you know, the Qubic is automatically turned on the light, turned on the air conditioning, turned on the fresh air.

So once, it take you maybe 3, minutes, take the elevator and walk to your, the Qubic, you automatically sitting in a, Pretty good lighting place and a visit, you know, fresh air. So that’s automatic and connect. Also, if and your chair is yours, you always, you’re sitting there, they automatically adjust to the, you know, the heights and the thing, the table and automatically adjust the heights to yourself.

So just like the car. Remember how you, you, you, you turn on the chairs and, and the, the, the, you know, mirrors, all that stuff. Right. Then you’re ready to go. You’re ready to work. Then. Yeah. And also you, you, you could give you some reminding, say you, what’s your calendar looks like today. [00:25:00] Right. I think.

Right. But then mostly a PC can do that.

Brian: So. Well, let me sort of circle back to something we talked about a little bit earlier. So when it comes to edge servers, have you made the shift to Arm or are you, are you considering that shift? Made.

Sean: You made it. Okay. The, the, the, the, we already, so I just illustrated it.

There’s two, two edge. The first edge we already made. The second edge we are still working on. The, the, the first edge we, it’s called the gateway. The connecting to the air condition, connecting to the elevator is already using container one, the cloud version one. We are still trying with. Okay. Yeah. We need, you’ll get there.

Yeah. We need to solving all those binary compatible problem. You’ll get

Brian: there. They assure me you’ll get there. So that billion number just rings in my head. You have [00:26:00] the scale to really bring this Technology to a vast space at a vast level. Do you see your technology becoming a broader standard outside of Vanka?

Sean: Well, actually it’s a. It’s very important. It is very important and we are actually working with Harmony Os. Okay. Yeah, harmony Os is actually the standard outside of the bank. So, so we also work with DBM. We also work with the many other Os. It is important to be the technology actually already and adopt by several companies.

Just like a bank there is maybe one or two company provided with similar technology like One World Service, One World Cloud. So it’s not only us do that, say the, the, the, the [00:27:00] technology, but several other, at least two competitor is doing the same thing I can imagine. So in the one company called the another company called the phone, though they also working in the same place and everybody actually that only I name a few, there is a lot.

Maybe it’s. Okay. Maybe tens of smaller guys that didn’t show up on the radars. So everybody understand that’s a, you know, 30 billion square meter of real estate need to be digitized. Right. So that’s, that’s one thing there it’s obvious, it’s not us to see it. You know, also there’s a couple of guys can see that same thing and also everybody to figure out how to digitize it and digitize operate everybody, you know, if they’re all competitive the same, so, and the, the, also the government sees it, the, the China government sees it, [00:28:00] because.

The, there is a lot of the hazard happening and firefighting happening, everything. So they want to digitize and monitor it. Then they can, you know, save a lot of life and save a lot of power. The 26%, 26 percent of the power consumption is burned by the buildings. Okay. That makes sense. That makes sense.

Yeah. So even 1 percent of China power consumption is huge. If we can lower the 25, 6 percent to 20%, that’s massive. We are contributing tremendously to the carbon neutral. What’s

Brian: the monetization strategy? around this technology, or is it simply a service that has to be delivered when you’re selling real estate space these days?

Sean: Today, actually, is a value added [00:29:00] service. It’s not a must have service. Yes, one thing. And, but, you know, the energy consumption actually is right now become a mandate of the China government, slowly become a mandate. And it is right now is adopted by several advanced cities like Shanghai, Shenzhen. It will slowly penetrate into the other cities and ask them to have to be green building and a carbon neutral building, etc.

Right now, several. Advanced company like the Alibaba, like Tencent, they want to do a green building because they want to say, I’m a technology company. I want to show example of the world, how I contribute into the the green of the world. That’s, that’s how they want to do. So right now we are only catering those advanced, the early adopt company customers.

And I will see the early majority will come in [00:30:00] next three to five years, you know, later majority maybe take seven, eight years.

Brian: But three to five years, that’s, that’s a very tight timeline. So that’s good.

Sean: Three to five years actually in China is not very fast. China three to five years, one generation of technology.

Brian: Yeah, three to five years in the United States is, is pretty fast, even though we want to move faster. The reality is that implementation isn’t always as fast as we’d like, or at least as fast as.

Sean: You actually take 10 years to iterate a one generation of technology, but the China is five years. Let’s just iterate

Brian: faster.

That’s a nice pivot point. And thanks for saying that. Speaking of time. So you talked earlier about your experience, how you’ve gotten to this point, you have a front row seat at IOT technology, at edge server [00:31:00] technology, at the adoption of new AI workloads. What do you see going forward? How are we going to evolve all this technology that we’re implementing today?

Sean: I see the two things. I think the one is adoption of those buildings. So we see massive buildings will, I think the penetration rate from 1 percent to 30 percent in the next Five, 30 to five years. That’s the penetration rate. The second, the, the computing, there will be the change in the game. IOT is become a simple technology.

And AI and the big model will be, become a major technology disrupt. So if you build the, the, the LLM into the buildings, so you just need to talk to it. So, and help me, you know, turn off the air condition. So, and optimize the air condition for me. So then basically we’re running that optimization and the implemented strategy.

And into the [00:32:00] new buildings using the existing maybe 1, 000 buildings and then transport to the 11 plus 1 buildings. So I think that technology evolving next 10 years, then I think the adoption rate evolving next 10 years. I think the computing power become cheaper and cheaper. In the next two years, and binary compatible of Arm is becoming better and better.

I will say that, you know, the cross point and tipping point in the next three, maybe three to five years. I think that all the edge server will be adopting Arm and with large language model. Running into different buildings and enable the AI to optimize different buildings. And I think that’s my view is coming and it will in the next three, five years is show us total different ballgame, I think.

Brian: Yeah, I [00:33:00] agree. It’s, it’s an amazing time to be alive and, and the good thing about AI is. It sounds like you have a fantastic hardware architectural foundation on which to build, right? A lot of AI workloads run on a CPU right now. You don’t need racks and racks of GPUs necessarily. But as you mentioned, more compute power, more compute efficiency going forward.

Is always better, but it sounds like we have a nice runway to start on. It’s

Sean: my view, it’s you know, for example, we just think about it, you know, you know, we, right now it was 1 percent of the building and be able to adopt the technology. You can afford a higher price, higher price. What are you for the rest of the 9 percent and the rest of the 90%?

So once we expand from 1% to 10%, the price have to, and the power has to come in down at least half the, the flop per dollar or dollar per flock has to [00:34:00] half, then another 90%. You have to continue to driving the, the computing cost down. And the, and, and that’s one thing. Another second things is how you help in the.

To integrate into that the different buildings, because you have to integrate a six, 15 to 16 different mechanics and the electronic system together. That’s cost a lot of money for the software guys is then we really need an arm to come in to helping integrate the software ahead of time on the each building.

So you basically connecting all the different systems already. You already work with the Honeywell. You already working with it. The, the, the Siemens, you already worked with Schneider. Then we have 1, 000 buildings and we deployed plug and play. Immediately, that saved tremendous money of the manpower.

That, that money even costs more money probably than the hardware alone itself. Yeah, I agree. [00:35:00] It’s a, it’s a, it’s a, my view today is I’m working on that. I will see next three to five years, it will totally transform the game. So once I, you know, and work with, I’m already, you know, working with the Honeywell Snyder.

Next three years, I basically connect all the system. Then once a year, I can plug and play much faster with this, you know, penetration rate.

Brian: Well, we look forward to watching it with you. So Sean, as a consumer, what’s your favorite piece of home automation? Siemens, huh? Which application? Which application dishwasher and why is that because it’s automatic you don’t really have to push buttons It just knows

Sean: I think I have to push button, but I don’t need to you know, rub it So when I sit at home, I already have the meal.

I feel [00:36:00] relaxed I don’t want to work.

Brian: Well right now right now. I wish you lived in my neighborhood because my Sonos based home entertainment system is not working upstairs for some reason and updating software is non trivial. So I wish you were here. You could help me figure it out.

Sean: If we work with you know, if we already work with them, then we always can easily OTA.

Right now it’s not connected. So, so you have to buy your own gateway, plug own 4G SIM card. Then you can ask your vendor to upload, upgrade your firmwares. It take forever. If you have Arm edge server, you, you know, the Arm and AV, Honeywell, Sonos, everybody can upload, you know, upgrade for use. Yeah. So that’s basically Arm.

And us together, we paved the highway for the smart buildings. [00:37:00] Then, in the

Brian: application, you can’t lie to them. Well, one of these days, my home will be that much smarter. Thanks to you. And thanks, Tom. Yeah. Well, thanks to Arm. Yeah. Sean, you’ve been very generous with your time late in the evening on a Friday.

So very much. It was an excellent conversation and I wish you and your colleagues the best going forward. We’ll we’ll keep an eye on you. Okay. Thank you very [00:38:00] [00:39:00] much.