George West conducted interviews of industry leaders. These were published in the WTRS Newsletter on a regular basis.
| Robert LeFort – CEO of Ember Corporation George: Can you tell us a bit about Ember and how it’s changed over the last 8 months, since the time we spoke with Bob Metcalfe? Bob: If you look at what has changed, there are two things that have happened – one with my coming on board, and one independent of me. The independent piece is that we had started down a path of being an independent hardware as well as a software company back in January, but now with the EM250 and EM260 chips in volume production I think there is a lot more substantiation and a lot more credibility that Ember brings in terms of saying, “Hey, we’re not just a software company borrowing someone else’s hardware. We are a full-fledged hardware company with actually, from what we can tell, the best solution available today in providing a working node and a working network for our customers”. That is important for a variety of reasons. In this space we are seeing that we can use software as a differentiator but customers really expect to get software and hardware as a package. The idea of differentiating those two is becoming harder and harder and what I call the ‘mainstream players’, people like Freescale have gone and acquired their own software stack now. And people like TI have their own software stack. The companies that are trying to get in with hardware that is independent of the software, and those that have been around with software that is independent of the hardware, are becoming more and more marginalized. George: That’s exactly what we are seeing too. Bob: It’s very hard to do that. In these embedded type of applications, on an end-application item, if you are Adobe or if you are Lotus/IBM, you can survive being independent of the hardware because the hardware is very standardized. But when you are doing embedded software it is very hard to do that effectively. It’s very nice when you are in the research mode and people are playing around, but it’s very hard when you get into the production mode. George: We basically agree on that and I think one of the things that we have seen is that some of the companies that were around early as stack providers are being marginalized and falling by the wayside. Bob: That’s right. I can’t take credit for it, because it was my predecessors who made the decision, but the idea of getting into the hardware business, was a critical one for the survival of Ember. We have watched other start-ups that were either in the software or in some level of the system side who are clearly not enjoying the success that Ember is. George: I’ll raise my hand. I thought at the time that going down the hardware path was fraught with all kinds of risk and at the time I thought “What are those guys doing? They own the world over there and now they are going down this hardware path.” But you guys were right. Bob: What I will say is that the value of software is getting harder and harder to do and as you see the volumes going up, it will be challenging. So what we have always said is that we are going to be a very competitive chip provider. We have a lead now, but that is a transition period because TI is still integrating Chipcon and Freescale is going down the 802.15.4 radio side and is just now integrating into a broader offering. So we are in a fortunate position, but these are world-class companies and they will get there. So that will not be a sustainable competitive advantage for us. But, we will be equally capable there. Our great software will be our sustainable competitive advantage because that’s what Ember does better than anyone. That is how we started and I know the guys at TI and I know the guys at Freescale and I did this at Infineon. While they may have a fair amount of software resources, it’s never a core competence for them. Its always one of those things they have to do to keep their chip business going. George: I completely agree. The focus there is basically selling silicon and to the extent that you need the software it has to be there. But they do it begrudgingly and kicking and screaming all the way. Bob: That’s right. What turns out to be helpful from a practical matter for us, even though it creates some challenges, is that we have our software team in Boston and we have our hardware team in Cambridge, England. While that means that we have to work harder to make sure that the two are tightly coupled, it also means that we keep that expertise pure. The software guys don’t get absorbed by the hardware guys or don’t get pushed off to the side and minimized by them. The software guys have their excellence and the hardware guys have their excellence and so I do think that our edge there is very sustainable. And we will always be able to differentiate in the sense of having not just better software that is more compact and more robust, but also more features. George: If you were going to pat yourself on the back here, what do you think is the most significant change that you have brought to Ember over the last 6 or 8 months? Bob: It’s not really done yet, so I can’t say that I can pat myself on the back, but what I am trying to bring is more focus and more customer orientation. Ember has always had the technical excellence. You already talked about that. I think I don’t need to validate that with you, but they have not had strong sales and marketing leadership in the organization. So that is what I am trying to bring: a customer focus, letting customers drive our priorities, and picking markets that we think we can be successful in. We focus on those, and then find ways to have a “low touch” model for customers that are not in our core focus. George: How have you developed your distribution channel, and what role does STMicro play? Bob: The distribution channel is a pretty broad brush for us. You have got the traditional distribution channel, not so much in the U.S. but over in Europe and Asia where we have distribution partners. We also have STMicro, who has very broad market coverage and a strong application engineering organization around the world. So it may be that they become part of that channel. Our strategy is that we have our core competency, we know where we want to go, we know what products we want to develop, and we know what the market requires to be successful. We will use our ecosystem to bridge any gaps between what our customers’ needs are to succeed, and Ember’s core competencies. We want to make sure that our ecosystem is very strategic for our business, so its not a given that anybody who talks to us is going to be a good fit. It’s also not a given that anybody who was a good fit in the past will be a good fit in the future. George: We spoke with someone, who is not really one of your competitors because their business model is much different. Their model was that they offered a very strong services and integration-oriented approach to the market. They were saying that if you are a technology-only player its probably not a good place to be these days. The conclusion I drew was that you don’t really need to build one single company that can do both services and products. You can basically be opportunistic where, in the case of Ember, you have a very strong technology, if there are people doing integration and other kinds of services in a market that you want to attack, you just simply go out and make an alliance rather than trying to build a specific company to go into that market. Bob: I would say that model was very good a year ago, its still very good today for the industrial side of the business, its becoming less attractive in the home and building automation. The reason for that is pretty simple: the services model is great when there is very little volume, but there is a lot of benefit from having the capacity…big companies that are not yet ready to totally integrate their entire environment into a ZigBee network, but want to get key pieces of data to manage their facility better. So there you can’t make money on the Ember side by selling 20 nodes and then putting in 2,000 hours to make the nodes get all the data that they want. You can make a very nice service proposal to do that but as the market migrates to higher volume, more standardized features/functions, then the willingness and ability for these customers to pay the service is much less and the hardware implementation is where all of the action is. I think we are just seeing the building automation sector migrate and we will be offering services, but those services enable our customers to accelerate their implementation into the market. It’s a means to an end, not the end in itself. That’s why we like the idea of an alliance with these integrators, who are then free to move from service to service to service. If these integrators can transition into higher volume markets effectively, they are OK. And as a given market matures, Ember is able to take advantage of the increased volumes with its chip sales. The piece that also typically comes with this is that as the networks become pervasive and the data becomes so readily available, then a lot of these companies take the maintenance in-house. So that is a big risk for the integrators. George: This is maybe a slightly different counter- example, but we spoke with some people in the Oil and Gas industry and they are just absolutely reluctant to add people to their organization. So if you can provide them the data they need and, ideally, let them reduce their headcount, and that is a solution that is very attractive in that industry. Bob: Obviously, from our standpoint, either way, either one will detract from the attractiveness of the market to us because we don’t quite care whether we are installing the network for the service provider or for the OEM. So, more power to them. I just think it gets a little bit risky as that market matures because sometimes these guys are pretty fickle and they go from one stage of wanting to outsource to, then a little bit later, deciding to save a little bit of money because they already have facilities people. But either way, it’s absolutely fine for Ember. George: You have obviously been around. You understand the dynamics of that whole outsourcing game. Bob: So as the ecosystem matures, we want to make sure that our partners are strategic partners and that they fit with where we want to go. That may mean that we have to go convince partners that we don’t work with today that Ember is a great partner for them, rather than just get excited about whoever knocks on our door, or whoever we trip on in the lobby of a customer. So you may see some minor changes from us there, but fortunately a lot of our partners have matured at the same rate and in the same markets as we have. So some of our activities will just be re-confirming the partners we already have. But an ecosystem, in the complexity of the markets that we deal with, I think will always be required. George: Is STMicro a key strategic relationship for Ember? Bob: We think that the relationship is absolutely key for Ember because it confirms Ember’s strength in the market and it strengthens our position and also gives us access to technology and even an access to the market. And then, of course, it helps to confirm that ZigBee is real. It was a big message for ZigBee when TI bought Chipcon and also Freescale doing much bigger types of things inside ZigBee now, and then you have got ST putting a lot on the line to work with Ember. So from that side, it is a significant relationship for Ember. But do I see that ST will fundamentally change Ember and influence where we go? No. I think that they will help us accelerate in the market and help to confirm what we wanted to do, but its all in the same vein as we expected to do before the relationship. On the back-end, what they give us is they give us access to technology. They give us more power in working with some of the big suppliers on the manufacturing side, whether it’s the tools people like Cadence and Mentor, or it’s the foundries like TSMC. Certainly having a relationship with ST is a huge help on that side. But that is invisible to our customers. On the front-facing side, which our customers see, there are some key applications that are looking at ZigBee where ST is going to be a very big help because of their involvement in those markets. So certainly things like the mobile phones, things like set top boxes, these are areas where ST is a leader with great relationships and with huge customers. So they can certainly help make sure that ZigBee gets a very fair hearing and evaluation inside these big customers, which would have been much harder for Ember to do by itself. Also, as I talked about earlier, ST has a broad distribution network and a very broad product portfolio so they get access to a lot of different customers and can really help us to be able to serve customers that we would normally not be able to serve. George: We are seeing a trend towards selling products with higher levels of integration (wireless node plus sensor/sensor interface possibly combined with management/control software) Do you see this same trend and what does it mean for the industry? Bob: The trend we are seeing is that customers want better tools, they want more support. Tools are a funny thing for me. Having come from large semiconductor houses, I was just accustomed to the idea that tools were something that partners do. My thought had always been that tools is something we have to get out of, its just not our core competence. As I talked to customers, they all highlighted it as one of Ember’s differentiation factors. They told me how much they love our tools and how much that made all the difference for us. And so based on this, we will stay in the tool business. But what we are seeing is that customers want to still be able to run the application, so they don’t want us to become an ODM, but they do want reference designs, they want tools that can be used very simply. They want to make their job easier because when you are working with R&D guys, they have plenty of time to play around with the difficulties of making something work. But the good news is that we are now past that and we are closer to the core product engineers who are trying to make something that will work in the field. Things are coming up now like determining how you test RF in a production environment. It’s not a trivial problem. So do they want help? Absolutely. Do they want us to take more ownership of their end products? That we are not seeing. George: Are you seeing a trend toward customers who are more in the product engineering role having a fear of the whole complexity of RF and wanting help in managing that complexity? Bob: Yes, we see that more in the building automation companies than in the home automation because the home automation companies tend to be more technology savvy. So they tend to be a little bit more courageous whereas the industrial, building automation, and medical companies are really looking for an awful lot of guidance from us. That is a challenge for sure. It is also a bit regional, no surprise there. The guys in Asia really want a lot more help on reference designs and help to make it easy and the guys in Europe and the U.S. are able to do a little bit more on their own. George: One of the other things we have seen talking to people is that many applications seem to have a sweet-spot of maybe a couple hundred nodes at the most. The key is that the application can be replicated quite a few number of times so you actually get some interesting numbers if you sell the application. Where do you see the sweet-spot in terms of the size of the networks that people are trying to deploy? Bob: We do see more and more large-scale deployment. We are seeing challenges in terms of trying to get these two and three hundred networks running smoothly and seamlessly and so I don’t know that that is going to be the sweet-spot, I suspect that the sweet-spot will be smaller than that. I suspect that we don’t hear much about those because they run smoothly and easily. In the bigger ones, you have to pay attention in how you handle the messaging, how many times you send the messaging, how you hand it back off again, or otherwise you will see performance degradation. So we are having discussions with customers and our customers are very often deploying on stacks that are quite a few versions older so we have to get them up to the latest version that can handle the density of nodes that they want to go to. We are seeing real deployments of large-scale networks, but probably that would be an overstatement to say that’s the sweet-spot. George: One of the things that we have been curious about is the use of wireless devices as a replacement for the classical embedded microcontroller in something like, I’ll pick on photocopiers. Are you seeing those kind of applications come in your sights, or is it still an emerging application? Bob: To be fair, our sights are very much focused right now on the home and building automation side. There are guys that are doing this type of development, and they are the “low touch” guys so probably for the scale of what they are doing now they would not be needing our help beyond just getting a development kit and maybe a little bit of training. My view on what you are going to see in the roll-out here, and even the types of applications you mentioned, is first the simple one-for-one replacement of wired for wireless. And then the second will be adding on to the existing infrastructure to get better performance, more information, better customer enjoyment. And the third is now I have this infrastructure and its tied into the control environment. I can really now do some ‘cool’ things to leverage that. For example, in a building automation application, you start out just replacing the wires and then I add some more temperature sensors, and maybe a little bit of control to save some energy and help my tenants get a little bit more of what they want. Now they have this infrastructure and they may want to do a bit of location tracking of my assets or asset management of how things are being used. That would get into things like the photocopiers. So I think that the photocopiers are emerging because they will become the next step after you have an infrastructure to tie them into. George: So it sounds like you really to have to boot- strap the infrastructure to get to full deployment. It comes back to Metcalfe’s Law: the size of the network goes up by the number of people who are attached to it. Bob: That’s right. So you will see a faster emergence of these new applications once you have the infrastructure that you can tie into rather than having a photocopier network around the building and a laser printer network around the building, if you can tie all of those into some financial analysis that you are doing to manage your energy expenses and security system and time cards and this is one more report you can do to help you more efficiently manage your assets, then it provides great value to the end customer. George: How does the EM260 network coprocessor play in your product line? Bob: One thing that is key to Ember in a variety of areas is the EM260 product. While we are very, very happy to have ST as such a strong partner and while we will leverage that relationship deeper than we will some other ones, we also want to make sure that we are microcontroller-agnostic. We are absolutely able to work with anyone. We have no problem working with TI processors, we have no problem working with Atmel processors, and the 260 is really a clear implementation of that strategy. And I think that really separates us because we complement microcontroller vendors, as opposed to trying to compete with them. What that does for customers is that if you have an application that is wired today, with minimal effort you can be up and running with a wireless solution. So you have time to market, you have ease of resources. We really play into the installed base of microcontrollers. While there are other people that do bits and pieces of that, basically they are doing it with a radio and we have a full network co-processor with the EM260. George: I am not sure that your competition has recognized the value of that. I take it from the standpoint of having been a development engineer and if you are being asked to adopt a new technology which ZigBee really is, and at the same time give up all you old tools, it’s probably four times as hard to sell that kind of a situation. The alternative is much easier: you just keep the standard tools you have, stay on the same microprocessors you are on today, and just learn a little bit of new stuff around ZigBee and a network coprocessor. I think that is not recognized as being a key selling feature. Bob: While software is only software, we had one partner we were talking to and I will try to be as diplomatic as I can here when you hear the comment. He said to us that he looked at one of these software stack guys and their stack works really, really well on our micros as long as you are not running any other application. That’s great. When you have to combine processing of your network messages with processing your application, it certainly can be done but it’s not for the new guys. Its just one more added complexity and we take away all of that stuff. We think that is going to be huge. The market right now is a little bit early to predict, but there are two schools of thought here and one of them is that the EM260 is going to be a more popular product than the EM250. This is something that technically was a minor achievement, but for the vision of the market and Ember’s position in the market this was huge. George: Is there anything else you would like to tell our readers as a final note? Bob: I would like to add that we are very encouraged by what ZigBee is doing. We really see them now behaving the way you would an industry standard to. And it sounds like you are way ahead of understanding what they are trying to accomplish based on reading the summary of your report and our discussion. They know, and Bob Heile specifically knows that this is the time to make ZigBee dominant. We cannot miss our opportunity and this window. People are craving a standard and ZigBee is the best thing out there for a whole range of applications. We have customers that are saying they want to help ZigBee to be the dominant standard, but they need help in the form of something that is released that they can rally behind. There are a bunch of things converging that are really going to help give the lift into ZigBee’s wings. We are very encouraged by that and we look forward to seeing it reach its final state in the next few months. More information about ZigBee… This interview ran in our Sept. 13, 2006 newsletter West Technology Research Solutions, LLC © 2007 All Rights Reserved |