Visitors to Alchemy Semiconductor Inc.'s Austin, Texas, headquarters are often nonplussed to find the place inhabited by gremlins and lizards. If they dip into the candy dish at the receptionist's desk, they'll really be taken aback: It may be filled with cockroaches.
The beasties are mere props, of course, as are the large monsters, rendered in the style of late puppeteer Jim Henson, that lurk in the office. Indeed, although Alchemy seeks to be known for designing low-power system-on-chip microprocessor cores based on the MIPS architecture, it has found that word of mouth has also spread about its funky decor. Some of its partners which include Microsoft, Epson and Wind River not only are interested in Alchemy's Au processor line but have been so inspired by Alchemy's menagerie that they've started collections of their own.
"It's interesting when we visit our partners; many of them now have these animals," said Ray Stephany, co-founder and engineering director. "We'll be walking down the hall there and someone will pull us aside and ask where we got a squid or other creature. It's quite something."
Although Alchemy is barely two years old, the history of the wild creatures goes back to the heyday of Digital Equipment Corp. While the monsters are mainly around for the engineers' enjoyment, they also serve as a quick way to determine whether a newcomer will mesh with Alchemy's avant-garde staff.
"Co-founder Jim Montanaro is known industry-wide as a purveyor of art," said Rich Witek, co-founder and chief technology officer. "Going back to Digital Equipment Corp., he used to give out glow-in-the-dark cockroaches and other bugs. That grew into other parts of the animal kingdom, and here it's grown into a unique form of creatures that I hope don't really exist. The sculptures are good filters: If someone comes in and is bothered by them, they're probably not going to be a good fit."
Today, the founders are working to convince customers that their first two products the half-watt, 400-MHz Au1000 and the PCI-compatible Au1500 are a good fit for future systems. The chips are aimed at what the company calls Internet edge products systems that bring new capabilities to the Internet, not the standard Web-connected devices.
Though they're designing devices that will enable Internet access, the Web is not a central part of their design efforts, largely because the 86 staff members are all in the same facility. Because they developed tools while at Digital and later worked at Cadence Design Systems Inc., they aren't very enamored of the once-touted scheme to access EDA tools via the Internet. But that doesn't mean that the Web isn't helping them get to market quicker.
"We have a good tool base in-house, and we have not seen anything we need on the Web," Witek said. "But we use it a lot. If you were trying to collect information four years ago, you'd send letters or call a sales rep, then wait for information. Now, you can download a file and get parts specs or IEEE standards."
Getting to market quickly is one of the big challenges for the startup. Now that the initial processor core is complete, successive family members are arriving quickly. The Au1500 was unveiled last month, just a few months after the February unveiling of the flagship Au1000. But even though development cycles must be short, product lifetimes in the embedded market can be long.
"When you go back and look at StrongARM, it's still alive today," Stephany said. "Cores have longer lifetimes than many other types of components."
The technology is drawing attention, but finding a niche in the crowded embedded microprocessor market is not quite as simple as designing a nifty chip, one analyst observed. "From a technical standpoint, they do have clear differentiation, with low power and high integration," said Linley Gwennap, president of the Linley Group (Mountain View, Calif.).
"The trick for them is to find customers where that combination of features is the right mix.
"The chip has very low power for a MIPS core, but it's not low enough for a cell phone. They have to find applications where power is not quite as critical, say in something that plugs into an outlet but needs low power so a fan isn't necessary."
History shows that the core team of five designers who moved to Austin seven years ago have a pretty fair chance at creating technology that will find a market niche. The StrongARM chip is the most recent in a string of seven processors that the team members helped design. The five have stuck together, moving from Digital Equipment to Cadence, which provided seed money when the team decided it wanted to spin out of Cadence and focus on developing processors, not the tools to build them.
That long history helped the partners gather $15 million in financing and forge agreements with such companies as foundry giant Taiwan Semiconductor Manufacturing Co. and ASE Group, a leading maker of IC packaging.
"We've got enough credibility that major players have partnered with us," said Witek. "The only way a startup can succeed is to leverage partnerships. When we deal with people, they know us and we know what they've done, so neither side has to worry about what can be delivered. It's much cleaner when you're not trying to figure out who people are."
While the team's success at Digital Equipment helped them get started, they've had plenty of reminders that in business, it's only today that matters. That's true even inside the company. Employees like John Treat, who joined Alchemy after graduating from the University of Texas at Austin in 1999, have little appreciation of Digital's glory days.
"I had heard of Digital Equipment Corp., but I didn't have much of an impression of it," said Treat, a VLSI design engineer. But he's learned a lot about the company since then, and he's also identified a benefit to keeping design teams intact. "They have been together a long time. That gives them a sense of loyalty. They work together well, and that makes it easy for me to learn."
Wowing a new grad is one thing. But impressing grizzled venture capitalists, even in the boom years of 1999-2000, proved more difficult than the partners expected. That's partly because the venture capital community at the time was burning to invest in Internet companies.
"Venture capital was plentiful but not always accessible; it was going to the dot-coms," Stephany said. "We guys with real products did not get the respect we wanted. Now, we get that respect, and the dot-coms have gotten 'mad com disease.' "
The founders are focusing during the downturn on designing products that will be ready when the uptick eventually comes. They have found one benefit from what's being called the tech wreck: "The price on used hardware is down quite a bit," Witek said.
The challenges of getting a company going and entering a crowded field during a down market are another reason that monsters haunt the Alchemy offices. The staff often burns the midnight oil, so the founders wanted keep the atmosphere light.
"At a startup, you spend far too much time working, so we wanted it to be fun," said Witek. "That doesn't mean we won't yell at each other, but you do need a spirit of camaraderie."
Since the founders know each others' strengths and weaknesses, Alchemy's management isn't as structured as in some startups. When they do yell at each other, they understand it's usually because there's disagreement on technology, not on personal issues. That also benefits the newcomers on the team, the founders contend.
"We've got a fairly open environment," Witek said. "We do a lot of design reviews. I've seen a lot of other companies, when they do design reviews, it's petrifying for the engineers. Everyone in the room is looking for errors. We don't do that. We're more interested in fixing errors. I'm not saying we don't have strong egos, but we don't have to worry about proving ourselves to each other."