A sharp "Crack!" echoed across Musgrave, South Africa, last February, instantly shorting out eight years of work for Costas Criticos. Lightning fried a broadcast tower that had provided wireless Internet access to one of the city's working class neighborhoods.
Mr. Criticos had relied on that tower as the hub of a nonprofit network that catered to students and low-income families. High-speed Internet in South Africa is several times more expensive than in the United States, so Criticos had shared a connection with neighbors at a discounted rate.
With that tower destroyed, the community's Web connection had closed.
But not for long. A friend of Criticos's visiting from the Massachusetts Institute of Technology brought with him several new gadgets from Meraki, a Silicon Valley start-up that has powered thousands of simple, inexpensive wireless networks around the world.
Like most networks, the neighborhood's busted system worked off a hub-and-spoke model, where information beams out of one antenna to each computer. If that central link breaks, every connection goes down with it.
But Meraki plays by different rules. Its wireless networks string together several small radios that bounce information from one to another. Instead of just radiating in a straight line, data can zigzag from box to box. That way, if one link breaks, the rest automatically reconfigure and find a different path for that e-mail or Web page to travel.
"In really just a few hours, we had the whole community up and running," Criticos says. "Not to mention, we set up the whole thing without the help of technicians, and our costs have never been cheaper."
The nonprofit network now connects several hundred people to the Internet, he says. And thanks to a hundred Meraki nodes placed inside a two-mile area, subscribers get broadband speeds for only $7 a month. (Compare that with South Africa's much slower dial-up access, which can cost $30 a month.)
Meraki's 25-year-old CEO, Sanjit Biswas, can recount a dozen similar stories of low-income neighborhoods that his wireless boxes have helped. He talks about each one with a pride more reminiscent of a parent talking about his kids than a businessman rattling off accomplishments.
"Well, those are all the stories I can think of right now," he says with a hint of humor and modesty. In fact, over the past two years Meraki has powered several thousand wireless networks across 70 countries and opened up the Internet to people who otherwise could never afford it.
While Meraki is a for-profit company, much of its products wind up in affordable housing projects, poor neighborhoods, and developing countries.
"The mission of the company is to bring affordable Internet access to the next billion people," Mr. Biswas says. "We've always felt a social obligation in this work."
Meraki, a Greek word that means putting love and creativity into your work, doesn't offer Internet service. It provides the hardware and software to manage a network. At the heart of the business is the $50 wireless Mini – a wireless router that is neither the fastest nor most powerful on the market. But many have called it the most simple and inexpensive.
Even as major players in municipal Wi-Fi abandon their large wireless projects, those two attributes have carried Meraki networks into Amazonian towns, African cities, and Alaskan outposts. And there's no sign of business slowing down.
A company's wireless roots
Meraki and the Mini trace back to Biswas's studies at MIT in Cambridge, Mass. In 2002, he and company cofounder John Bicket worked on Roofnet, an experimental network that developed the "mesh" model for today's Mini. The project offered graduate students free wireless Internet access if they agreed to put up with the team's system tests and service tweaks.
The two PhD students programmed software and crawled on rooftops throughout the city installing their wireless kits. Unlike the Minis, which are about the size of two decks of playing cards, the Roofnet kits had to be carried with two hands and contained $650 worth of computer parts and antennae.
"We were paying for all this with research grants, so we couldn't afford a van or anything like that," Biswas says. "So we'd take them on the subway. After setting up a hundred of them, I think I now know every street in Cambridge by heart."
And Cambridge got to know them pretty well, too. In 2006, when the team's wireless signal had covered one-third of the city, students would stop them in the street and holler out "Hey, it's the Roofnet guys!" he recalls. Biswas got several unexpected invitations to parties and barbecues that year, including one in Mountain View, Calif., where he helped Google with its early foray into municipal Wi-Fi networks.
As e-mails from CEOs and NGOs piled up, Biswas and Mr. Bicket realized that their work had quickly grown beyond the bounds of academia. Suddenly, they had a flourishing business.
With their advisers' blessing, the two dropped out of MIT, moved to Mountain View, and refined Roofnet into what is now Meraki.
Free access in the US
Almost immediately after launching the company in 2006, they sold a couple thousand kits to Google and hundreds more to groups along the West Coast. One early adopter was NetEquality, a nonprofit that rolled out free Internet access to affordable housing complexes in Portland, Ore.
"The mesh network design helps us keep costs down in so many ways," says David Cannard, director of NetEquality.
First, the wireless technology saved them from ripping open walls to install connection wires in older buildings. Even in new construction projects, he says, simply laying cable from apartment to apartment can be expensive.
Second, since all the Meraki radios talk to one another, the system automatically warns NetEquality if one of the boxes malfunctions, saving Mr. Cannard (or more likely a professional technician) from having to check each box to see where the problem lies.
The information-sharing scheme has allowed one 400-apartment building in Portland to split five DSL lines across 100 Meraki nodes, Cannard says, dropping the cost of Web access to about $1 a month for each apartment.
Of course, there were a few kinks and quirks along the way. Cannard remembers "freaking out" last winter when several of his networks suddenly collapsed.
"Nodes were going down all over," he says. "It turned out that since we just plugged the boxes into normal outlets in each apartment, the residents decided they'd rather plug in their Christmas lights. But so many unplugged them that they were knocking out access to most of their neighbors as well." NetEquality now screws the boxes directly into outlets so that they can't be removed.
Meraki's grapevine growth
Now in its second year, Meraki has yet to advertise its services. Simple word-of-mouth has carried its name through the global geek grapevine.
Criticos heard about it from Biswas's MIT adviser. A school in Ecuador learned through a friend-of-a-friend who worked for Cisco Systems in San Jose, Calif.
Slovakian businessman Marcel Hecko recalls reading about Roofnet on the Internet a few years ago. Soon after Biswas formed his company, Mr. Hecko persuaded his colleagues that "we simply need to get as close to Meraki guys as possible."
Scrounging together enough money, he flew from Slovakia to San Francisco, boarded a train to Mountain View, and then with a GPS navigator in hand, walked 40 minutes to the Meraki office.
When he returned to Slovakia, Hecko rolled out a two-pronged plan for Meraki routers: a nonprofit effort in his home city of Bratislava and a growing business that sells Meraki networks around the world.
Biswas has encouraged others to build their own businesses based on Meraki's network. "Nonprofit networks are crucial," he says. "But in some cases, turning it into a business plan is the best way to ensure the network is sustainable."
For example, Jim Bletas, head of WNI Global in San Jose, Calif., is currently negotiating with telecoms for a massive installation of Meraki nodes in poor areas of São Paulo and Rio de Janeiro, Brazil.
"We want to have a million units" with Internet access by 2009, Mr. Bletas says. "And if it works in poor, urban areas, we'll be looking at rural networks next."