At Motorola, Tom Beaver helped make the VMEbus a mainstay of the electronics industry. Now he's hoping he'll be able to steer another technology into the realm of classic. As the new chief at tiny 3DSP Corp., Beaver has turned his attention to system-on-chip digital signal processing, with a view to guiding his company's DSP core technology into design wins with OEM customers and chip makers alike.
"Nobody really owns the category of scalable and configurable DSPs today," said Beaver, chief executive officer and president of 3DSP, a 1997 Irvine, Calif., startup. "What we want to do is to win the DSP IP intellectual property architecture war. I think we'll get to $150 million in about three years, which will make us No. 1 in the DSP IP-licensing core space."
Though that sum is dwarfed by the revenue figures of an industry giant like Motorola, where Beaver spent 30 years, pulling it off would represent something of a quantum leap in the fledgling DSP IP industry.
"Today that market probably totals $50 million to $60 million," said Will Strauss, president of research house Forward Concepts (Tempe, Ariz.). "The DSP Group is probably the poster child for that market, and it did about $30 million last year. The market these guys at 3DSP are addressing will grow at just under 30 percent per year faster than the DSP market in general certainly for the next three to four years. Every mainstream semiconductor company has got to have DSP in its portfolio, and many OEMs do too."
Indeed, Beaver and the other executives at 3DSP are focusing on exactly those key markets: chip companies and OEMs that don't want standard parts. The first deal with an IC supplier came in February, when 3DSP linked up with National Semiconductor Corp., which didn't have scalable DSP cores among its offerings. National will use both the core and 3DSP's Hi Fi development tool set.
Beaver believes that both markets will play a role in the emergence of system-on-chip (SoC) technology. And the National deal notwithstanding, he predicted that the first successes will be at the more traditional OEMs, which are making more of their own chips.
"The reason that a lot of the hype around SoC has not materialized is because it's been looked at only from the perspective of semiconductor companies," Beaver said. In fact, "Most of the SoC work is going on at big companies like Cisco and Tellabs, and at smaller companies too. Another category is companies with no customer-owned tools, which may include a division of a large company or a smaller company."
Beaver's mention of Cisco and Tellabs isn't by accident; voice-over-Internet Protocol, audio functions and digital subscriber lines will be some of the key markets 3DSP will address.
For those without tools or links to fabs, Beaver and company will suggest partners that have both. For those who want to do their own work, 3DSP has developed the Hi Fi tool set, which addresses the specific design quirks of DSP cores yet works with tools from Cadence, Mentor and many others.
If 3DSP's strategic approach sounds familiar, that's because the company considers itself the DSP "alter ego" of processor-core vendor ARM Ltd., Beaver said: offering intellectual property, development tools and the DSP core itself.
That plan has been in place since the company's founding in 1997 by Kan Lu, its chief technical officer, and fellow Texas Instruments veteran June Jiang, who is vice president of DSP development and chief financial officer. Lu cited those three elements of the startup's strategy when he gave the company its name.
The first years of operation were devoted to developing the technology. With that in place, and the market poised for takeoff, the founders decided it was time to bring in Beaver.
"There's a tendency with engineers to feel that if you've got a great product, it will sell itself," Beaver said. "Boy, do we need to change that here. We're too much the best-kept secret around. They have been relying on the spoken word to promote the product. That only lasts so long before you run out of friends."
VME, PowerPC
During much of his three decades at Motorola, Thomas L. Beaver headed the computer systems group that is now the core of one of the few bright sectors for the struggling giant. He was closely involved in semiconductors too. "When I look back on my career, the two most fulfilling things at Motorola were that VME became a standard and the PowerPC was reconfigured to become a success," said Beaver, a Chica-go native who signed on with Motorola after earning BSEE and BSBA degrees from Marquette University, on whose board he now serves. He later went on to obtain an MBA from the University of Minnesota. "Originally, we thought it the PowerPC was a processor to take on Intel."
After leaving Motorola, Beaver served as executive vice president of sales and marketing at Astec Power and then as chief executive at Wyle Electronics. He left Wyle when the company, one of the largest distributors in the United States, was acquired by Arrow. Afterward, "I was talking to a friend in venture capital and asked him what I should do," said Beaver. "He said why not run this company, 3DSP, which was initially an IP license play that will evolve into ASICs. I talked to them four or five times. I was a little skeptical at first."
But the lure of heading a small but energetic company soon sold him on 3DSP. "I've got it in my veins to be the CEO," Beaver said. "I've already run a division of Motorola and I didn't want to run a division of Arrow. It's kind of a been-there, done-that feeling."
The culture of what's essentially a startup is utterly different from that of a large corporation. "Today, we're at about 60 people, though we're growing that substantially," Beaver said. By contrast, "Over most of my 30 years at Motorola, at any one time I probably had 5,000 people reporting to me."
Small can be beautiful, of course. Beaver noted that there won't be much bureaucracy to slow things down at 3DSP, eliminating one of the big headaches that comes with managing a division of Motorola or a large company like Wyle.
Since leaving school, Beaver has seen major shifts in the way ICs come out of development. "I grew up with three, four or five passes at silicon before we got a working part," he said. "Today it's rare when we don't have a working part the first time."