Dr. Nicely concluded, after eliminating other potential sources of the irregularity, that the Pentium chip itself must be flawed. After contacting Intel's technical support service to no avail, he posted an electronic message to a Compuserve bulletin board to ask other Pentium owners if they were experiencing similar problems.
The message set off a flood of Internet bulletin board postings that confirmed the irregularity. The news media (print and television) soon released the story to the rest of the public and the concern over the flawed chip escalated. In November, Andrew Grove, CEO of Intel, posted a message of his own. He acknowledged the problem, emphasized its minimal significance and made an offer to replace the Pentium chip for users of the Pentium processor who are engaged in work involving heavy duty, concentrated floating point calculations.
Meanwhile, Intel worked with Dr. Nicely and other scientists to develop a 'workaround' that it could build into its programs to avoid the flawed calculations. Intel's offer and workaround only accelerated the concern and reaction by the Pentium public. In the face of its self-inflicted PR nightmare and increasingly critical media coverage, Intel decided in December to offer "no questions asked" chip replacement program to all of its Pentium customers. The company would take a one time charge of $475 million against its earnings for the fourth quarter of 1994 to pay for this program.
Perceived Product Risk
Intel's initial decision to tolerate the flaw was based on a quantitative analysis of its customers' risks. After Intel concluded that these risks were extremely low, it decided that nothing needed to be done. A few months later when its customers were made aware of the flaw, many of them felt they had been intentionally deceived and that they had paid a premium price for a damaged product. Intel's customers ultimately concluded that something had to be done.
With all the debate exchanged over the months following the public exposure of the Pentium flaw, the involved parties did manage to agree on one thing. The heart of the issue concerned risk. The issue in this debate can be broken down into two questions: did the Pentium flaw expose Intel's customers and Pentium owners (end-users) to undue risks and were those risks significant enough that Intel should replace all flawed chips
Intel's "at-risk" customers, specifically IBM, were concerned that the flaw might diminish its reputation as a best product provider and, consequently, its sales would suffer. There was concern the public might view IBM as the source of the problem rather than Intel, to which the problem rightly belonged. The "at-risk" Pentium owners had already bought their computers. They were confronted with an involuntary risk. It was clear that the customers' and the Pentium owners' motivations to react were based on their assessments of risk.
Intel's approach to resolving product defects reflected an implied assumption that the assembler would be responsible for dealing with the end-users (Pentium owners). Intel also assumed that information about the flaw did not need to be passed along to its customers due to the insignificance of the flaw. When millions of computers with flawed Pentium chips passed into the hands of end-users, both assumptions would prove to be poor.