If you’re a glass-half-empty person, you might say the mesh networking technology pioneered by Boston’s Ember Corporation is a solution in search of a problem. If you’re a glass-half-full person, you’d probably call the company’s eight-year history a case study in flexible thinking. Regardless, after years of market struggles, Ember seems to have found a niche where its technology for self-organizing digital radio networks will shine: smart-energy systems designed to give utilities and consumers more control over how they use energy.
Andy Wheeler and Robert Poor, who built experimental wireless sensor networks for the U.S. Defense Advanced Research Projects Agency while they were students at MIT, founded Ember in 2001 with $3 million in seed funding from Polaris Venture Partners, DFJ New England, Stata Venture Partners, and Ethernet co-inventor Bob Metcalfe. (In subsequent funding rounds over the years, the company has raised an additional $78 million.) The idea was to find applications for the concept of ad-hoc mesh networking, in which pairs of transceivers in a network set up two-way communications as conditions allow, and messages hop from node to node until they reach their destination (roughly the same way they do inside the Internet).
In its early years, Ember made mesh-networking software for other companies’ microchips, usually equipment designed to be embedded in temperature sensors for petroleum refinery pipes and other similar devices. But over time, the company has evolved into a chipmaker in its own right—just one that happens to have a special expertise in the software running on its chips. And it has gone open-source, investing heavily in ZigBee, an open industry standard for low-power, low-bandwidth mesh networking.
Wheeler and Poor have both moved on to other companies. Meanwhile, Ember has moved on to new application areas—with the biggest near-term payoffs likely to emerge from a utility industry initiative called the Advanced Metering Infrastructure. In pilot AMI projects in states like California and Texas, computerized “smart meters,” or energy gateways, are being attached to utility customers’ homes. The devices communicate wirelessly both with utility control centers and with in-home thermostats, displays, and smart appliances, allowing utilities to dial back electrical usage during hours of peak demand and giving customers more information about their energy consumption patterns. The home side of these communications depends on technology from Ember, which is the leading supplier of communications chips for AMI devices.
“The core technology is not that different from what Andy and Rob had in mind,” Ember’s president and CEO, Robert LeFort, says. “But it’s being used in applications they never had in mind when they were developing it.”
LeFort joined Ember in 2006 from semiconductor maker Infineon Technologies, where he had spent four years as president of the company’s North American operations. I interviewed him about Ember’s development and its business opportunities last week at the company’s headquarters in the Fort Point district of South Boston. An abridged version of our talk follows.
Xconomy: How did Ember evolve from a software company into a chip company?
Robert LeFort: It’s just very hard to make a business case for embedded software. You typically earn a royalty on a per-instance basis, and you don’t usually cover even the cost of developing the software. So you need to be doing more than just software. The end devices and modules [using Ember’s software] were interesting to us, and chips seemed to be a better fit. So in 2003 we introduced our first chip, the 2420, which we still sell today. And in 2004 Ember acquired RF [radio frequency] technology and the chip team that developed it from Cambridge, UK-based Cambridge Consultants. [The company’s main products today are newer microprocessor models called the EM250 and EM260, both announced in 2006.] We monetize our software through the chips. I think it’s fair to say that there’s nobody who uses our software who doesn’t also buy our chips. And if you look at our competitors from the 2000 era, the ones who went up the chain and sold just software have not fared so well.
X: Talk a little about the company’s recent turn toward smart energy applications.
RL: There were a variety of things that converged, and some other things that fell off the table. When I first came, we were very excited about commercial buildings and lighting. Wireless mesh networking is perfect for that application, but we learned that it is a very fragmented market that moves very slowly. It’s still a good market for us, but it’s incremental. The home automation market is also good, but somewhat limited until they get the costs down and the maturity up.
Meanwhile, I think smart energy will work really well. It’s something we saw from being so tightly involved with the ZigBee Alliance, just from being down in the trenches and talking to guys about what they were doing. I would say smart energy only made up about 10 percent of our revenue in 2008, but we expect it to be 40 or 50 percent in 2009.
X: You mentioned the ZigBee Alliance. As I remember it, when ZigBee started out years ago it was mainly pitched as low-bandwidth alternative to the Bluetooth standard for home entertainment and automation applications. The one that sticks out in my mind was a remote control for opening and closing your blinds, which seemed kind of frivolous to me at the time. But Ember became instrumental in adapting the standard for more advanced applications, like monitoring devices for their energy usage. How did Ember help to advance that standard, specifically?
RL: We were big proponents of the ZigBee Pro networking standard, and fundamental to that is true ad-hoc routing. We believe that RF links are inherently unreliable. Unlike a Bluetooth device or a cell phone, a temperature sensor that loses its signal can’t move six inches to the right to get a better signal. You need a network that can handle that efficiently. We wanted to build a network where you wouldn’t be constrained by
