Unqualified Opinions: Amir Heleem and Helium's Future For IoT

Ryan Selkis interviews Amir Haleem, the Founder & CEO of Helium about the future of decentralized IOT.

Unqualified Opinions S2E3 Blog Graphic.png

Ryan Selkis
All right, everyone. This is Ryan Selkis and you're listening to Messari's Unqualified Opinions where each week I interview cryptos top builders, investors and personalities. Today I'm speaking with Amir Haleem, founder and CEO of Helium. It should be a very exciting conversation. Let's dive right in.

This podcast is presented by Block Works Group, one of the top blockchain events and media production companies I've worked with. For exclusive content and events that could help you with insight into the crypto and blockchain space, check them out at BlockworksGroup.io and you will not be disappointed.

All right, everyone. Welcome back. We've got a lot of excellence, unqualified opinions, live streams coming up with some pretty fantastic guests, including today's guest, founder and CEO of Helium and Amir Haleem. We're going to talk about the future of telecom. We're gonna talk about their new tokenized network that's helping to bring wireless networks, low impact wireless networks to cities across the world starting with Austin. We're gonna talk a little bit about that experiment and how they see crypto and telecom ultimately intersecting, where there's a obviously number of applications since the telecommunications industry in some respects could could put a bit of a stranglehold on Crypto if they were to ever censor transactions.

Timely conversation as Amir and his team just completed a fundraise a few weeks ago from the likes of Union Square and Multicoin. Disclosure, Multicoin is an investor. Sorry. But super excited to welcome Amir. Amir, I know that intro did not do justice to what you're working on but why don't we just start off with the basics. You know, how you kinda came to this project, what your crypto origin story was and how it's kind of intersected to bring, to bring Helium to market and how you guys have gotten off to such a fast start.

Amir Haleem
Cool. Thanks. Yeah, actually it was actually pretty good, pretty good intro. Helium has been around for a little while and we have always been trying to solve the same problem, which is how do we build ubiquitous low power wireless networks. Now we've sort of had this opinion for six or seven years, I co-founded the company with Shawn Fanning from Napster and our belief at the time, and is still very much true, is that low power things like sensors and tracking devices and stuff that look like this, you know, tiny little things just don't have a wireless network to use. It's just sort of that simple of a premise, right? It should be low power, it should be ubiquitous, it should be incredibly cheap and none of those, none of those, that combination of things just doesn't exist today.

That's always been the problem that we tried to solve. Crypto for us is a more recent additions. Probably two and a half years ago we, we started thinking about, you know, what if we could sort of flip this whole business model on its head, because one of the challenges building low power wireless networks is that the applications generally don't want to pay very much to use them. Building out, you know, expensive infrastructure in a way that cellular networks have been built today and trying to recoup your costs by charging people to use them is much harder in a low power space because people are sending, you know, like temperature readings and location data, stuff that is on the order of bytes rather than gigabytes. We sort of thought about it and said, what if we just flip this on its head and sort of just abandoned the whole idea of trying to make money on the network itself and trying to sort of create this incentive structure and this model that encouraged people to act as their own network operators.

We've been sort of nicknaming this thing, the people's network and that's, you know, sort of the approach that we've taken that we built a block chain from scratch and come up with a sort of novel consensus protocol that really tries to incentivize people to act as network operators by running these devices that we call hotspots. That's, you know, that's like a six years story attempted to be summarized in, you know, two to three minutes. I'm sure I did a mediocre job.

Ryan Selkis
Could you just elaborate a little bit on kind of Mesh networks in general and the impact that could have on something like the bitcoin network? Obviously if you're talking about 3G, 4G, 5G data and video streaming, it's much higher bandwidth and that's something that takes, you know, quite a bit of fiber and quite an expensive infrastructure to actually facilitate. That's probably gonna stay in the purview of the telecoms for the foreseeable future or maybe that'll change someday. But right now you are talking about relatively, you're talking about Internet of things devices, so low bandwidth devices that I would imagine include things like bitcoin nodes if they're in nodes and other crypto networks.

Amir Haleem
It depends, right? I mean like I think the bandwidth requirements for a typical, like a series of node for example, are pretty high, right? There's a lot of traffic there. A lot of syncing and the blockchain is big. That's not the kind of application that we would think of running on a network like this. Like this is a network that has data rates that are, you know, 1 to 20 kilobit per second. It's a very very small and so I don't think you would look to run a full node that way, although potentially clients and other stuff like that could exist this way. I know guys like goTenna are working on, you know, these sort of mesh networks that inter-operate with like being in Bitcoin in general. But our structure is less of a mesh and more like a traditional model.

More like a cell model where you have, you know, a base station and devices connect directly to that base station and then they don't on hop through each other. Mesh networks are, you know, where you have like a bunch of different end nodes and they communicate through those nodes effectively to get back to the Internet somehow. We don't do that. Everything in our case is a, is a single hub. You have a hotspot device and it can listen to, you know, tens of thousands of end devices as long as they are in range of that hotspot. There's no meshing in our, in our system per se.

Ryan Selkis
Can you talk a little bit about the Austin pilot and the economics to get a city up like that up and running. Then, you know, maybe some of the use cases for folks who might be considering purchasing one of these devices. Is this just something that a user would do as an experiment? Because I remember, you know, the 21 computer was, you know, the one line pitch that people kind of laugh at now was, you know, mine bitcoin from your toaster and obviously the economics of that were terrible, right? This is a little bit different because there is some type of consumptive use, but there also could be some type of payback period. One, the device isn't that expensive. You wouldn't necessarily call it a major investment, but it's enough where it'll give people pause. Is this going to be primarily for hobbyists early on do you think or, or is there, you know, kind of real models that can emerge around this?

Amir Haleem
There's, I mean, sort of two sides to the universe. I mean there's the supply side, right? Which is people building the network coverage, right? We'd ask them, we typically, so we have this wireless protocol that we call long Fi. It's extremely long range. A single hotspot could create anywhere from a mile to, you know, 25 to 30 miles of network coverage. The economics of building citywide coverage change a lot when you have that kind of capability, right? Where a single hotspot that's placed well with like a decent antenna could cover, you know, large parts of the city, you start to sort of realize that you could build this out in a reasonable way only needing 150 or 200 people per city. That's what we've been been targeting.

I think we have around 200 in Austin. I don't remember the exact number. We sort of hit the cap that we wanted to hit there. The economics are sort of different, interesting in this case, right? I think most crypto assets today are very speculative in nature, right? There's very, there's not that much utility underlying a lot of these. There is some, but not all of them. In our case, like we use crypto as a way to basically build a wireless network that we anticipate will be used by a pretty broad set of users. Some of those, you know, at the bottom end are makers and hobbyists and then at the top end are big enterprises, right. I don't know if you saw in our press, you know, we're working with companies like Lion and Nestle and…companies like that, that are building end products.

In the case of someone like Lion, it is the desire to be able to track scooters when they're stolen. You know, I'm sure you guys are both have all read, it happens often. In the case of someone like Nestle, it's all about like optimizing supply chains, you know, so it's like the got these water delivery business where they drop water off to like water coolers at businesses and offices and they have no idea how much water is being consumed there. Those are just like simple applications that that you feel like should be easily accomplished today but just aren't because it's either, you know, the service plans are too expensive or the hardware is too expensive or it's old and complicated or there isn't network coverage or you know there's always some combination of things that makes this hard. But this is truly a utilitarian network and we'll have some sort of announcements coming up in the next few months around the kinds of customers and deals that we're putting together for people that actually take advantage using it.

Ryan Selkis
When I'm, so walk through the roadmap. Austin, it sounds like you've got some of the infrastructure in place, but as you think about the test net main net kind of schedule, what are some of the key milestones over the course of the next year?

Amir Haleem
Austin is is sort of the main net launch for us and that's August 1st so that's a scary time because our blockchain I think is quite unique in the sense that each of these hotspots is an actual minor on the network and there's no sort of big cluster nodes that acts as sort of the main consensus of origin there. Each of these hotspots is sort of running in some way, running the blockchain itself. Our consensus protocol is, it's kind of interesting there. It's going to be an interesting time like deploying something like this in the wild. But that's August 1st is when we're doing that. We intend to ship to the rest of the country by October. We may accelerate some cities ahead of schedule, but we'll see about that.

We want to start getting into Europe next year. I mean Europe I think is a huge market for both crypto and IoT in general. I think there's a massive opportunity there. In terms of long long term road mapping, something you said there, I think it's sort of interesting to think about which is that we view this as a mechanism to like deploy wireless infrastructure in general. Like we happen to view low power networks as the least well-served today. Like there isn't really good infrastructure for those kinds of things. But you could imagine how something like this could work in the realm of something like 5G especially when you think about how many base stations are required to deploy a 5G networks. And there's a lot more complexity there. There's like back-haul questions and you know, there's a lot more complexity with how you do that. But ultimately if this succeeds, like we would view it as a roadmap to like deploy other wireless networks this way that are sort of beyond just low power applications.

Ryan Selkis
As you think about scaling, so you know, you said it's a couple of hundred of these devices to cover a city. How many IoT or, or other low bandwidth connected devices can those support? Do you have to scale the network as the number of increases that leverage? That two hundred? Or does that stay pretty static based more on kind of redundancies and just geographic coverage?

Amir Haleem
Yeah, I mean I think eventually you would probably need a little bit more capacity than that as the network gets really on full tilt. I mean, if I remember, and I'll probably butcher these numbers, I think each hotspot can support around 300,000 devices connecting per hour. You know, if you multiply that out by 200 you're supporting like whatever 60 million devices per hour. That's, you know, probably pretty good for the time being. But we'll see. You know, it depends on the kinds of applications. Like if you have, you know, an application that is constantly sending data all the time or like you know, like pinging locations every second or something like that, then you could sort of quickly need more capacity than the 200 hotspots can supply. But for the most part, like we feel pretty good about 150, 200 being a number that gets you pretty strong coverage. Of course like augmenting it is not that complicated, right? I mean the hotspots today we're selling for 495 but you will also be able to DIY your own hotspots for less than that. Adding more hotspot capacity is relatively inexpensive. When you think about the cost of like deploying cell towers for example, to try and do the same thing, which is, you know, a hundred grand plus per site.

Ryan Selkis
Could you give a sense for the potential cost savings, you know, just rereading the Multicoin guys' posts, you know, they mentioned things you mentioned Lion. Maybe we'll go with that, but you know, they mentioned pet tracking and you mentioned the scooters. Just how much cheaper is this and how much more or less reliable would it be to use a network like Helium versus 3G or the telecoms? Because it seems like there might be a trade off in performance. But I'd love to know how you think through actually scaling that up and getting to the performance that you need to work with someone like a Lion more with a large, you know, national pet chip tracking company.

Amir Haleem
Let's start with the cost. I mean, in some settings this is something like a thousand times cheaper than the way you would do it with cellular. This is constantly in flux, right? I mean, the, the guys like Twilio and the Saw companies are constantly sort of beating that price down. One way of thinking about it is just, you know, what does it cost versus cellular? Then the other way of thinking about it, which sort of gets to your other question too, is that this is unlike any cellular technology that's available today. There's a new thing called NB IoT for example, that the carriers are rolling out, which is a low power and low cost version of cellular, but even then it is, you know, anywhere from 10 to a 100 times less battery attrition than what we're doing with long…

It becomes difficult to compare the two in a way because there isn't really a sort of direct equivalent, like there is no low cost version or low power version of cellular that sort of matches what we're talking about. But that's what we think. I mean if you looked at it compared to some of the pricing out there that we've seen today, which is you know, 20 or 30 bucks a month or something, it's, you know, thousands of times cheaper than that to use our network. In terms of like the reliability and redundancy, I mean, one of the interesting characteristics of this network compared to cellular is that we don't really have this like notion of being attached to a hotspot. If you have a scooter for example, that's just the zipping around Austin and it transmits a GPS location or or you know, some other reading, like a battery reading or whatever, you know, multiple hotspots, will probably hear that at the same time because the sensor isn't associated with a particular hotspot.

It just transmits. We noticed this in San Francisco in our testing, like several hotspots receive it at the same time because the range is so large. You end up with pretty good redundancy this way. I mean we have, I think now about 30 hotspots in San Francisco and it is very rare that we go anywhere in the city with a transmitter and are only heard by a single hotspot. You get good redundancy even with you know, 20 or 30 in a city like San Francisco, which is, you know, complicated by, by hills and stuff like that. We expect that the redundancy created that way in Austin will be pretty good. One of our investors is Munich Re, which is one of the world's largest reinsurance companies and you know, we've had conversations with them around, can you offer, you know, SLAs effectively via an insurance policy on the network? Those sort of interesting conversations that are happening there that I think could be really, really interesting but fundamentally like just giving the way the protocol works, given the range, we don't think that we're going to run into a a too much of a reliability concern just because there'll be so much redundancy with that many hot spots in the city.

Ryan Selkis
We got a question from, from Tom Shaughnessy at Delphi. He asked, is there any way that the ISPs can block the reselling of their data? Because it seems like at some point you're gonna have to be on a collision course with some of the local telecom and ISP providers.

Amir Haleem
Yeah, I mean there's this sort of, there's two, three answers to this. I mean, one is like we think you should be able to do whatever you want with your own Internet connection. That's fundamentally, that's our first position is that, you know, screw those guys. But from a more like practical point of view, you know, the kinds of data that gets sent and received here are so small. You know, like this as an example, like a GPS packet is in our network of something like 35 bytes, right? It's absolutely a absolutely tiny, right? At some volume, maybe that becomes significant. But I think identifying that kind of traffic, you know, identifying that it's a hot spot then and blocking it is just a lot of work for ISPs. I don't know that they'll ever bother with it. If they do, you know, we can always look at alternate tactics like using tour and other things to obscure this. But I stand by the first part. I find it obnoxious that you can't just do whatever you want with your own connection. That's what we're going with.

Ryan Selkis
Well certainly it certainly makes some sense. You know there's been a couple of other companies, projects early on that were focused on, you know, kind of creating these networks for IoT devices. Filament comes to mind, a project based out of out of Colorado. Why do you view the time as right now and what are some of the things that you learned or borrowed from some of the other major blockchain protocols that are kind of the most integral design components of your own network? Because you are spinning out something from scratch. There's a pretty unique consensus mechanism that you're implementing. It makes intuitive sense. But you know, I'd be curious if you can kind of draw on some other parallels or what some other projects have already built to help people better understand this network and how it operates.

Amir Haleem
Yeah, that's a hard one. We certainly gained a lot of inspiration from looking at other projects. The simplest way of thinking about it is the network is asynchronous BFT protocol at the end of the day. We use this thing called Honeybadger BFT, which was designed by this guy Andrew Miller who's a professor. I can't remember exactly where, but he's been involved in the crypto space for a very long time. I think he's involved in Z cash. We liked Honeybadger because it's asynchronous so messages can sort of arrive at any particular time. There's no, you know, timing guarantees required, which we thought was important in a network like this, because people are going to be connecting hotspots over all sorts of different internet connections. Whether they're not going to be a data centers, they're going to be on people's home networks and wherever they may be.

The way that, what's interesting about our system is that usually I think in the sort of BFT type setups, like if I thought of something like Hash Graph or, iOS or you know, anything that looked like that or Stellar even to some degree, like they pick their consensus members and that sort of it, right? iOS this block producer thing and you know, there's 21 of them. Those are the elite few and Hash Graph has the council or whatever it is. We never really liked that model because it's sort of very centralized in nature at some level. You know, like someone picked that group and it's unclear how you would ever, you know, replace them without completely forking the network. What we wanted to do was, was have this BFT consensus group, but cycle the members constantly.

The way we do that is we have another, we have a civil resistance system that we call proof of coverage and it's, you know, difficult to describe well, at least I don't think I'm very good at it, but it's a challenge response protocol that happens over the wireless network. Basically hotspots that are sort of close together, you know, within miles of each other, send and receive these encrypted onion packets to each other that are layered and can only be decrypted by the specific hotspot that was supposed to receive it. We use this system and it's, you know, more complicated than that I can probably get into in a reasonable a period of time, and I'm pretty bad at describing it. But what ends up happening through that challenge response system is that hotspots gain and lose score.

Over time, you know, hotspots are losing score. The default state is that you are losing score and that you're dishonest. You increase your score by actively participating in the proof of coverage protocol and submitting proofs to the blockchain. As you do that, your score increases. Every 30 or so blocks, we elect the best scoring hotspots in the network to become part of the consensus group. Constantly cycling out census group we think is a very unique characteristic of the network and also very scary because no one is really in control of it at any time, right. It's just constantly electing the next best hundred hotspots to to be the consensus group for that period of time. It's a very interesting system and we think it's probably the most distributed type of of block chain network that we've seen just by nature of the fact that the mining power can already be consolidated in any way though there's no benefit to like buying a 100 hotspots and putting them in your office.

You won't increase your score that way. Like they have to be distributed out into the wild. The sort of miners in our network are very broadly distributed. The consensus group that runs the network is constantly changing. I like to, among the sort of highest scoring nodes. We got some of these tricks from, I forget the exact network that we drew inspiration from for this but it was the first time that we had seen one system be used to sort of bootstrap another. We use proof of coverage to bootstrap, you know, our Honeybadger BFT implementation. I can't remember where we got that, but we thought that was very clever and that sort of inspired us to do this. Then we were inspired by the guided tour protocol, which was a paper that we read around around preventing DDOS attacks by requiring a similar kind of mechanism to gain access to the server.

It wasn't applicable to what we're doing because our…only has a limited range of it. No one can sort of target anyone else, like you have to be at least within miles of each other. We drew some inspiration from that and sort of modified it along the way. But you know, there's a lot of really interesting projects out there that we looked at to think about this. But in terms of, you know, building a wireless network, like we were not really aware of many others that have tried to do this. Nodal is one that comes to mind, but they're using Bluetooth and you know, we started to think the range of Bluetooth is feasible for building citywide networks. As we've seen with things like Tile in the past. You know, it's a little bit of a knowing what Filament is doing, but I don't think they're in the sort of wireless network building business. I think they're probably on a slightly different different tangent.

Ryan Selkis
Are these city specific networks? Because you're starting in Austin, right? What happens when you move into New York? You'll have New York and Austin. Will those be two separate networks with two separate consensus, two separate block chains? How do you think about consensus in a multi-city system so that you don't have a situation where in New York you're going to have maybe as large a population of devices, users that are participating in the network and maybe that's larger than the next five cities combined. Does New York then become the largest stakeholder in the ecosystem that's actually providing proof of coverage and if you extend that out, does that create any geopolitical challenges is as you scaled other larger cities internationally?

Amir Haleem
We don't think so. We spent a lot of time on like how the scoring system works and I, and I think it differs pretty substantially to what's in the white paper at this point because we wrote that so long ago, but we've tried to, the proof of current challenges are rate limited. They can only be submitted and requested at a reasonable interval. The scoring system, as I said, sort of penalizes people being clustered together too much. The real issue is that you're going to have hotspots that are sort of lone wolves that are sort of sitting out there on their own. They'll get some mining rewards for participating in the proof of coverage as a challenger as as we call it in our system.

But those guys I think will on average, you know, earn a lot less than people that are in a densely populated cities like New York. But we don't think for example that New York would dominate Denver or something just because I haven't been more people there. The network is a single sort of large networks. We've talked about sort of future scaling where you may have separate consensus groups that sort of bubble up to a singular main consensus group. But today it's a single group. One of the interesting characteristics of the wireless network that we are able to create this way is that we sort of created this like gigantic VPN, right? Any device that uses the network can talk to any hot spot, but have its data, you know, routed in an encrypted way to the correct endpoint, right?

If I'm Lion, you know, and I have a bunch of these little tracking devices in my scooters, it doesn't matter what hotspot I talked to, the hotspot will always inspect the blockchain and look at the ID of the device, sending the data and then route the data to the correct end point. You know, that's how this works. That's actually one of the major issues that we've seen in low tower network deployments so far today, is that they're very fragmented, right? Like there's a bunch of different companies that have a bunch of different low power networks and they don't talk to each other and there's nothing that resembles a roaming standard or roaming specification that really works. If you have, you know, like a…device today or something, there's no way you really sort of taking that up between networks and often these networks like overlapping the same city.

The whole roaming problem is a really a big one for an end user because they don't really want to have to think about that. Right? They just want the device to just work. One of the benefits that we get from having this distributed ledger that everything could inspect is that we can route data to the right end point, you know, as a result of like looking up device IDs and things like that. It's a, we think, a relatively clever system that leverages a lot of what we learned over the years, which is, you know, which is things like, you know, enterprises don't want their data to go to a start up server and you know, stuff that in hindsight is relatively obvious, but it's difficult to implement from a technical point of view or at least implement well.

Ryan Selkis
Last question I have is just on kind of size and scale of the opportunity because, and I'll make the disclaimer for you. This is not investment advice this is for information only. I'm just trying to do kind of back in the napkin like market sizing, right? Obviously one of the theses in building out this type of network has to be that there's going to be an explosion of IoT devices. Something like Helium could potentially unlock the usefulness of of many other low bandwidth devices. There's kind of a pretty clear kind of macro IoT play here, but if you just kind of look at the current snapshot of the world, what type of value could be captured by users in a network like Helium's if you, you know, replaced or were able to take, you know, just like Lion internationally to begin using this as the communications protocol versus going through their local wireless carrier.

Amir Haleem
I think the aggregate opportunity size is extraordinarily large. I mean, depending on who you talk to about this and which analysts you like, you'll hear numbers thrown around that are, you know, devices in the tens of billions range over the next several years and it's always a little bit, you know, difficult to take those things seriously because I think they've been making the same claim for, you know, about a decade…

Ryan Selkis
That's the number of devices, right? I'm talking about like what type of economic value, you know, at the communication layer where, where you are, what would potentially sit there.

Amir Haleem
It's difficult to say. I mean it's difficult to say because there isn't a perfect parallel today that you can draw them. There isn't a good example of a network like this that you can look at. I think number of devices is the only reasonable way to start thinking about what the opportunity size could be and then sort of backing that out into like, you know, how much could a single hotspot capture in a single city for example, it gets, you know, significantly more complicated. But in terms of, you know, just the way I think about this or a way to think about it is that we expect the number of total IoT type or low power IoT devices to be in the tens of billions of the next several years. We know that cellular carriers are spending tens of billions or more dollars building out narrow band IoT networks that aim to solve the same kind of problem.

If you put at least those two things together and you get some sense of the scale of the opportunity here, which is that if a network like this succeeds and gains any kind of market share, and if you know, you end up in 4 or 500,000 Lion scooters as an example. We have another customer that we'll announce soon that is in a similar size, like 3 or 400,000 units over the next two or three years. You know, you start to imagine that the potential of sort of owning a piece of that network or being your own network operator is potentially very lucrative. That's, you know, the best way that we know how to estimate the size of this now. We have various sort of, you know, token economic models and things like that.

But yeah, I'm sure you'll understand with the regulatory climate, we're not really going to share them with anyone, but we think that the opportunities is substantial to participate. That's sort of the rationale for already doing any of this. I mean that's, that's why the company exists. That's why we thought that this was a good opportunity in the first place and we've sort of been encouraged over the years that like no one has actually figured it out very well. But because of the economics that I mentioned at the start, you know, building the network yourself, you know, the way that Sig Fox has done it over the years, for example, it's just extremely expensive. Like there's a very difficult path to like monetizing the network back when you were hundreds of millions or billions of dollars into it. We think distributing the economics like this is the right way to go from sort of a privacy and security perspective.

Like we think is the only way to do it that it makes any sense. We hope that over time that we will start to be able to this strategy into other spaces that you know are in dire need. You know, I continue to be frustrated with by only having Comcast as my broadband option at home. I would love to have alternatives. I hope that 5G and other technology like it sort of bring about that change and that things like Helium can participate in that and sort of distribute that a little bit more or evenly.

Ryan Selkis
Well, it's certainly an exciting vision. We're going to be on the lookout for how things go beginning of August. Best of luck. In the meantime, where can people find out more information, get their devices or started playing around with the code?

Amir Haleem
Yes, Helium.com is our our page. The store has been online for a few weeks now. If you go to Helium.com/store you can buy devices now and so of course if you're early in the mining process you were sort of disproportionately earning tokens versus being later on in the process. If you're interested in participating now is definitely the right time. In terms of code, we have a get…or or an organization gethub.com/Helium. Some of the code is open source. I think the periods of P libraries and some distributed key generation stuff and some of the crypto primitives that are there will be open sourcing and pretty much all of the rest of it very, very, very soon. Which includes, you know, instructions and guidelines on how to build your own hotspots and participate that way too.

Ryan Selkis
Well, Amir, thank you so much for joining us. Until next time. Thanks guys. Peace.

That's a wrap. Thanks for listening. New episodes of Unqualified Opinions go live weekdays at noon eastern time. Can you follow me in the meantime on Twitter @TwoBitIdiot, if you want to continue the conversation or troll me. Otherwise, I'll see you next week.