Valley Wonk: DSL Man
No, John Cioffi, a Stanford University engineering professor whose research involves signal processing, sports two DSL lines at home, along with an analog PBX about the size of two laptop computers. And he tinkers with them in his spare time. Working with , he's got the lines running at four times their expected speed, he says.
Cioffi's explaining his home life to me over coffee, a meeting arranged simply because I'd always wanted to meet the guy (and I had a column due). Cioffi created the DSL industry, as far as chipheads are concerned, and his name is regularly invoked at broadband conferences.
For the most part, Cioffi, who's looking professorial in a V-necked sweater, enjoys the reputation. But, he wryly concedes, "It's balanced by when your neighbor's DSL doesn't work and they ask you to fix it."
There's also the lawsuits. Cioffi is repeatedly being subpoenaed and deposed as an expert witness, with calls from up to five lawyers in one day. "Alexander Graham Bell, after the invention of the telephone – they say he was in court 600 times." So, he's still got a lot to look forward to.
I wanted to know what Cioffi is up to lately, because his latest DSL work is starting to hit the chip market (and I had a column due).
But first, a quick history: While at Stanford, researching ways to get more bandwidth down a copper line, Cioffi founded a chip company, Amati, which went public in the mid-90s and eventually got sold to Texas Instruments Inc. (NYSE: TXN). The Amati legacy remains: Peter Chow, once a student of Cioffi's at Stanford, is CTO of TI's DSL technology center. And driving down Highway 85, you can see an old Amati trailer parked on an embankment.
Now, Cioffi has some new optimization techniques for DSL, in the form of Dynamic Spectrum Management (DSM). It's yet another trick for squeezing speed from copper. After a few years of development, the technology is finally appearing in commercial chips, with firms like ElectriPHY Corp. and Ikanos Communications Inc. quick to note their DSM prowess (although some say Ikanos doesn't count because it implements DSM "Level Zero," a vanilla version that involves less complexity than Levels 1 through 3).
It's in DSM that Cioffi found the impetus for his next startup, Adaptive Spectrum and Signal Alignment Inc. (Assia). He's been on a company-founding fast since Amati, although he's been enticed to join four boards of directors, not including Assia, and three advisory boards. Startups flock to the guy. To invoke Cioffi's name in PowerPoint slides is like saying Alex Rodriguez was on your little league team, or the recently departed RL Burnside once jammed with your band, apparently.
Assia puts DSM into practice. The idea behind DSM is to collect the parameters of each DSL line and tweak each connection accordingly to maximize throughput. What's new is that DSM takes into account the effect each line has on neighboring lines, a factor that's long been irksome for telcos. "We collect that data back in the phone company somewhere, throw it around, and say, 'Could this line go faster?' " Cioffi says.
The appeal of the new startup was twofold. First, it isn't a chip play. "It's more of a service and software that's sold to a phone company. Chips are being commoditized so rapidly, especially in the communications area."
But more important, Assia isn't just churning out another product. Cioffi sees the potential for creating a new type of business, much the way Amati did. While carriers have the technology in trials, it's unclear exactly how the company will make money out of its services. That's the part that seems to pique his interest most.
"Amati was a very special startup, as is Assia, because it was not only creating a business, it was creating an industry. I didn't realize it at the time, but those are special startups. They don't follow the normal rules. It's very difficult to get venture capital, because the return on investment doesn't follow a model."
Even as the DSL drumbeat goes on, Cioffi doesn't turn his nose up at other broadband technologies – one of his board seats is with PON chip vendor Teknovus Inc., for example (see Teknovus Grabs $5M and Teknovus Gives EPON Chip Details). Don't hold your breath for him to buy that cable modem, though.
"Cable's good for TV, but it's not really good for data. There's the sharing problem in the upstream direction. It wouldn't surprise me if the cable opportunity diminished because of competition from satellite and DSL," he says. "And with the phone companies putting a lot of money into distribution [things like FTTx projects], cable could get the squeeze."
— Craig Matsumoto, Senior Editor, Light Reading