Tech startup of the month.

ENGINEERED INTELLIGENCE

FORT COLLINS

ENGINEEREDINTELLIGENCE.COM 

FOUNDED: APRIL 2002

INITIAL LIGHTBULB: Originally from Stuttgart, Germany, Matt Oberdorfer was a tech writer, penning five books about programming, before joining Hewlett-Packard as a manager in his native country. He jumped at an opportunity to transfer to Fort Collins in 1999; an earlier snowboarding trip had whetted his appetite for the Rockies.

Oberdorfer soon pinpointed cluster computing as an emerging industry. A cluster computer, also known as a parallel computer, consists of up to 2,000 traditional computers ("nodes") connected by a high-speed network. Recent innovations in hardware have made clusters capable of performance levels normally associated with multi-million dollar supercomputers. Like supercomputers, clusters are good for attacking big, highly complex problems, such as atmospheric modeling, aerospace design, or simulating car crashes.

Because there was no corresponding market for cluster software, Oberdorfer went out on his own and launched Engineered Intelligence Corp. with an octet of fellow ex-HP employees and other Fort Collins businesspeople. The mission: to build the Microsoft of parallel computing. Now Engineered Intelligence's president and CEO, Oberdorfer, 37, heads a staff of five full-time employees.

FINANCING: Its launch financed by the founders and angel investors, El closed on a Series A funding round with Menlo Park, Calif.-based U.S. Venture Partners in early March. The approximately $2 million infusion will allow El to hire sales staff, open a Bay Area office, and develop other products.

IN A NUTSHELL: "The new processors are getting as fast as the old (supercomputers) were, and if you hook a bunch of them together, you can get supercomputer performance," said Jim Gutowski, Engineered Intelligence's executive VP of marketing and sales. The upside: a cluster computer costs a fraction of the price of a supercomputer.

The downside: programming a cluster. "Today, it's very complex to do," said Gutowski. "Only the super-geeks can do it." To make cluster programming simpler, El developed its primary product, CXC (pronounced "C by C"), which serves as both an operating system and a development environment for cluster computers.

CXC "itself is a very simple language to program parallel computers," said Oberdorfer, who likened it to C and Java. "You can write a simple scientific algorithm without being a parallel computer expert ... and you can create them on your laptop; normally, you have to be 24/7 in the computer room."

Dr. Richard Casey, founder of El client RMC Software, a Fort Collins firm that uses high-speed computing for cancer-drug research and molecular modeling, called CXC "a great benefit" to cluster users and a big timesaver. "If you parallelize (code) once, it should run on a variety of hardware," he noted.

Initial El marketing efforts have tried to create a buzz in the cluster-computing user community. The company has staged three "Grid Wars" events in which competitors design CXC-based "warriors" that do battle in a virtual arena. Another program targets life-science researchers who want to harness the power of parallel computing.

El's marketing strategy is focused on bundling CXC and other El software with the cluster computers sold by such companies as IBM, HP and Dell. The company currently has a contract to bundle CXC on HP clusters and is in talks with other major providers.

QUOTE OF NOTE: "We are penetrating the market as a commodity on (HP) systems. Now that we have financing, the company has to make that jump. We have proven we can be bright and innovative. Now we have to prove we can be a successful mid-sized software company."

[ILLUSTRATION OMITTED]

--Engineered Intelligence President and CEO Matt Oberdorfer

THE MARKET: International Data Corp. pegged the market for supercomputer hardware, software, and services at $7.5 billion, with 13 percent annual growth. Clusters account for about $2 billion of this market now, but the category is growing at a 35 percent annual clip and might eventually render traditional supercomputers, or "big iron," obsolete.

About 250,000 scientists and engineers use parallel computers now. El estimated that 3 million more could use them if equipped with its CXC product. "It's niche market to start with, but (cluster computers) could be in everything in the future," said Oberdorfer. "The real challenge for us is to get to that point (when it is pervasive)."

He has set his sights high: "Microsoft is not parallel, but we are. I want to be in every device."