Intel let a bug slip through their design of the new Sandy Bridge P67 chipset used along with their second generation Core processors recently introduced. This actually isn't unusual -- new chipsets may have hundreds of "errors" that need addressing, but most are so minor that they don't really affect the operation of the chip. The problem with this mistake, though, was that eventually it would lead to failure on a major part of the chip.
The H67/P67 chipsets carry two sets of SATA ports. These are the ports typically used to connect mass storage devices like hard drives, CD drives, and DVD drives. The two sets operate at different transfer rates, one using the older 3 Gb/sec standard, and the other the newer 6 Gb/sec rate. The problem is that a transistor in the 3 Gb/sec controller was improperly biased, and that would eventually lead to higher leakage current through the transistor. The higher leakage changes the system's operating parameters leading to failure of that controller. The 6 Gb/sec controller is not affected, nor is anything else in the chipset.
Original estimates of the cost of overcoming this manufacturing defect were placed at $300 million in lost sales and $700 million to handle the recall and replacement of the bad components. News coming from Intel now indicates that the hit may not be that bad. For one, it's highly unlikely that a system would need both sets of controllers. High-end workstations would want to use the faster controller for better throughput. So the chips are still usable for those types of systems. A simple BIOS change to remove access to the 3 Gb/sec controller is all that's needed.
Intel plans to ship a new revision of the chip by the end of February. If they can, that would slow down production of those motherboards in the pipeline, but not remove them. In addition, Intel's quick response to the problem and rush to correct and replace those components in the field also helped mitigate the overall damage and reduced expenses. And, as stated, there are still possible uses for those chips already made. It won't be cheap, but it also won't be a billion dollar blunder.