The Lessons of New-Market Disruption

Teradyne was successful. Hewlett-Packard was not. Professor Clark Gilbert writes about how two companies had such different results with disruptive innovation.
by Clark Gilbert

In the early 1990s, both Teradyne and Hewlett-Packard identified new technologies that had enormous potential to cause new-market disruptions. As with most disruptive innovations, these technologies offered to provide smaller and less expensive solutions for customers as the tradeoff for reduced performance. That initially led managers inside both companies to resist devoting scarce resources to developing those technologies because their lead customers were saying they wanted additional performance features, not fewer.

But believing they had a chance to enter new markets, senior managers at Teradyne and HP forced their organizations to focus on these new technologies. The CEOs of both companies got involved, creating separate, financially autonomous divisions for the projects to ensure that the technological breakthroughs would be developed and commercialized.

Teradyne's efforts were rewarded with a new line of business that created revenues in excess of $200 million over the next several years. HP backed out of its efforts after two years with multimillion-dollar losses and a considerable amount of bad press. How did two stories that began so similarly end up so differently?

Teradyne: Testing Opportunity

In 1994, Teradyne, a leading provider of machines that test the quality of microprocessors and other integrated circuits as they come off the assembly line, saw an opportunity to create a more compact and cost-efficient test system by using two new technologies.

The first, CMOS (complementary metal-oxide semiconductor), was aimed at simplifying multicomponent test systems. Typically, silicon wafers moving through production are fed into the test system, which sends electrical impulses into each wafer and compares the actual readout with the expected readout. Each wafer is mapped to indicate good regions and bad regions, and this information is then used in cutting the wafer into individual components.

Development groups need to be starved so they develop the right focus and priorities.
— Alex d'Arbeloff, Teradyne

At that time, each test system comprised multiple machines that took up more than twenty square feet of scarce factory floor space at the semiconductor facility. Teradyne engineers knew that CMOS technology could enable tighter integration of the test system architecture. With this advance, the entire system could be fit into a single box the size of a mini refrigerator and stacked on top of other line equipment. Better yet, this tighter integration would greatly reduce the manufacturing costs.

The second technology was Windows NT. Teradyne's test systems included various software programs that directed the equipment to run tests and report the outcomes. Teradyne found that by using generic components offered with Windows NT software, it could slash the cost of delivering the product to integrated circuit manufacturers.

Despite the significant cost advantages offered by using CMOS and Windows NT, Teradyne's business units did not want to devote any engineering resources to leveraging these technologies. Teradyne's established customers tested high-end components such as microprocessors, for which CMOS architecture was too slow and inaccurate. What's more, a switch to a Windows NT platform would force customers to abandon their investments in customized test software developed for UNIX systems.

Teradyne's top customers were not interested in sacrificing accuracy and scrapping already installed software for the cost benefits of CMOS chips and NT. Instead, they had an extensive list of demands for new high-performance features that engineering was trying to keep up with.

Believing there were other markets in which these technologies could be profitably leveraged, Teradyne's chairman, Alex d'Arbeloff, created a new subsidiary called Aurora, whose sole mission was to commercialize a low-cost CMOS/NT circuit tester. Its general manager was granted autonomy, capital, and freedom to recruit star engineers from other divisions.

D'Arbeloff presided as chairman of Aurora's board of directors, believing that, "with things customers aren't asking for, in markets that are too small to be interesting, without the CEO backing, the thing is never going to happen."

In choosing a target market for Aurora, the team was guided by d'Arbeloff's financial expectations for the new division; Aurora's initial revenue expectations after product launch were $1 million in year one and $11 million in year two. D'Arbeloff's message was clear: Reach profitability and then grow the business.

He thus forced Aurora to identify and focus on a single, small market opportunity. He starved the group of the engineering resources necessary to enter multiple markets.

"Development groups need to be starved so they develop the right focus and priorities," he says.

Without the restriction of large short-term revenue expectations, the Aurora team rejected the idea of becoming a low-cost solution for the high-end market. Instead, it chose to create an entirely new market for test equipment: microcontrollers, which are very inexpensive commodity components that perform basic computations in household appliances, like irons and toasters. At the time, microcontroller manufacturers were just beginning to use automated test equipment. Although the microcontroller market was a small one, it was a perfect fit for Aurora. Commodity component manufacturers had significantly lower test-requirement demands than microprocessor manufacturers and therefore commanded a significantly lower price.

Moreover, the microcontroller market was a perfect beachhead from which to expand into markets with higher performance requirements.

Because Teradyne did not have to satisfy aggressive short-term revenue expectations, the product architecture could be built around the price point demanded by the cost-sensitive customers in its target market.

Hewlett-packard: Small Drive, Big Hopes

In 1991, Hewlett-Packard's Disk Memory Division (DMD) played a small role as a provider of high-end 2.5" disk drives. HP's management explored strategies for transforming DMD into the market leader, but other companies were already too firmly entrenched in the 2.5" drive market. HP engineers believed, however, that a new opportunity was on the horizon, because 2.5" drives would be too large for next-generation portable devices, such as ultrathin laptops and personal digital assistants (PDAs). They speculated that if HP could develop a drive only 1.3" in diameter, the company could establish itself as the market leader and set the next industry standard.

While a 1.3" disk drive would be the smallest drive in the world (about the size of two postage stamps), it would drastically underperform the 2.5" drives in terms of cost per megabyte and overall capacity, two parameters that have historically characterized disk drive capability. In fact, it had been HP's consistently superior capacity that had enabled the DMD to hold a profitable position.

DMD's existing customers, high-performance laptop and desktop manufacturers, were not interested in the 1.3" model for their current product lines because capacity and cost per megabyte, not size, drove design wins. Consequently, DMD's disk-drive engineers were exclusively focused on finding ways to cost-effectively pack more capacity onto their 2.5" model.

But despite that resistance, HP's senior managers decided to pursue the small disk, so they set up a new group named Kittyhawk. An entrepreneurial manager from R&D was assigned as group leader, and he was given freedom to recruit any star engineering talent he required. These engineers were required to sign a creed that stated: "I am going to build a small, dumb, cheap disk drive." CEO Lew Platt visited the team frequently to provide support and cut through any roadblocks. Platt even carried a Kittyhawk prototype around in his pocket to show off.

These engineers were required to sign a creed that stated: "I am going to build a small, dumb, cheap disk drive.”

The Kittyhawk team had to weigh its market options: Sell the disk drive to DMD's existing customers, who were manufacturing increasingly portable computing devices, or explore emerging opportunities elsewhere.

A leading video game manufacturer, Nintendo, told a senior HP manager that if HP could supply a 1.3" drive for $50, then Nintendo would be interested in purchasing millions of units for their gaming cartridges. According to a Nintendo manager, "the software writers' dream is to have more cheap storage. We're always looking to create more complex games." As HP managers explored this market demand, they realized that there were several other parties interested in a 1.3" disk drive with moderate capacity that could be purchased for $50.

First, though, financial expectations from senior management needed to be considered.

During the planning process, HP senior managers had raised Kittyhawk's annual revenue targets to $100 million within two years of product release. They also set a launch schedule six months shorter than initially planned. Given these new requirements, the proposal to create a $50 drive was out of the question. At that price, the drive could not satisfy the new revenue expectations, and it was unclear whether the development hurdles could be overcome within the launch schedule.

As a result, HP chose to target its small disk at the PDA and ultrathin laptop market rather than to video game manufacturers. At that time, analysts predicted that the emerging superportable market would be huge, and HP engineers thought they could leverage existing intellectual property in drive design from their high-end 2.5" laptop drives. In other words, because it appeared able to satisfy the aggressive revenue targets set by HP's senior management, the superportable market seemed the right choice for the Kittyhawk.

But it wasn't. Kittyhawk did not succeed in either the ultrathin laptop or in the PDA market. When the PDA market failed first to materialize in the early 1990s—the market for handheld devices did not take off until the late 1990s—the Kittyhawk team pushed upmarket and targeted ultrathin laptop manufacturers. But these manufacturers were looking for breakthroughs in drive technology to improve cost per megabyte and capacity, not to sacrifice them. Moreover, at this point, the Nintendo solution was no longer an option as the Kittyhawk team had layered in components and features that inflated the price point well beyond $50, and Kittyhawk's strategic flexibility was constrained by its growing losses.


In the end, HP's Kittyhawk was discontinued and Teradyne's Aurora delivered highly successful commodity component test systems. HP experienced both a financial loss and a morale loss, while Teradyne capitalized on an opportunity and preempted a market migration away from its products. Ultimately, financial expectations drove decisions that handicapped Kittyhawk and helped Aurora.

These stories underscore the point that managers looking to commercialize disruptive technologies must set modest revenue expectations or risk forcing their divisions to overlook emerging markets or target the wrong ones.

If revenue targets are too large, a division may be forced to choose target markets that will not value the disruptive technology. Trying to please customers in established markets, where performance expectations are high, can drive companies to include features that make a product too expensive to satisfy customers in emerging markets.

Because disruptive innovations often see failure before success, flexibility is critical to survival. Companies can retain flexibility by initially focusing on low-cost applications whose requirements satisfy the lowest common denominator. This way, breakthroughs in manufacturing and design can still be leveraged in the event that the product architecture is redesigned for higher-margin products.

Companies have come a long way in proactively addressing the resource allocation challenges of disruptive technology development. As these cases demonstrate, however, even if a company foresees a disruption, and marshals resources within a separate entity to capitalize on it, picking the wrong customer will lead to failure. Aggressive revenue expectations can keep venture managers from choosing to target emerging markets, places where disruptive technologies are likely to find their initial success.

As the HP case shows, sometimes a fledgling division picks the wrong customer and builds the wrong product because corporate expectations leave no other choice.

About the Author

Clark Gilbert is a professor at Harvard Business School.

Daniel Shapero contributed to this article.