Stardent and the Elusive Graphics Supercomputer 333

and product costs. However, the real flaw was introducing a first product that failed to demonstrate clear performance superiority, as discussed in Chapter 8. Although a second set of products overcame this shortcoming, the inadequacy of Ardent’s first product inflicted very painful damage on the firm. According to Allen Michels, Ardent's founder and CEO: "You get one shot."

Rarely does a start-up get a second chance if it misses with its first product. The first product must be designed, sold, and delivered correctly. The only reason the company survived to merge with Stellar in October 1989 was that Ardent's Japanese partners had a basic faith in the firm and the market. Poduska became president and Michels chairman, and in 1990, Michels and Sanders left the organization.

The following are some lessons that can be learned from Ardent's experience with its graphics supercomputer:

• Although a start-up can count on a long-established company to be lethargic and noncompetitive, a high-growth firm less than ten years old is unlikely to let a start-up enter its space.

• Stardent underestimated the competitiveness of Silicon Graphics by classifying it with established companies such as Apollo, DEC, HP, and IBM.

• When attacking a walled city, a start-up shouldn't telegraph its intentions to the inhabitants by distributing promotional coffee cups, towels, and T-shirts. And it should avoid attacking two walled cities at once.

• When developing ~a high-risk, state-of-the-art product that has never before been created, a start-up's founders would be well advised to anticipate project surprises (by a factor of 2) in schedule, resource requirements, product cost, and specifications.

• When a new venture is entering a crowded market where the niches are small and perhaps hard to find, it helps to have a strategic partner with resources.

• Ordinary venture financing would probably have given up on Ardent after three years. Kubota is determined to be successful in technical computing and is prepared to invest for a long-term gain.

• A start-up must not skip the seed stage, especially if it is building a product that the world has never seen before.

• Detailed planning is essential. With a proper seed stage, Ardent would have either cut the product to fit the funding or made a realistic product plan.