Computer scientist: For several decades, the number of transistors on

Every 18 months:
number of transistors on new computer microchips has doubled --> leading to microchips’ speed doubling

From mid-1990 into the next decade, each such doubling led to double cost of producing.

(A) The only effective way to double the computing speed of computer microchips is to increase the number of transistors per microchip.
Not given. Increasing number of transistors increases speed. But is this the only way of increasing speed? Not known.

(B) From the mid-1990s into the next decade, there was little if any increase in the retail cost of computers as a result of the increased number of transistors on microchips.
Retail cost of computers is unknown. We only know the cost of producing microchips increased.

(C) For the last several decades, computer engineers have focused on increasing the computing speed of computer microchips without making any attempt to control the cost of producing them.
Did the engineers make an attempt to control costs? We don't know. May be they did but couldn't control much 1990s onwards.

(D) From the mid-1990s into the next decade, a doubling in the cost of fabricating new computer microchips accompanied each doubling in the number of transistors on those microchips.
Yes. We know that mid 1990s onwards, each such doubling - "double number of transistors leading to double speed" was accompanied with "double cost". Hence this is implied in the argument.

(E) It is unlikely that engineers will ever be able to increase the computing speed of microchips without also increasing the cost of producing them.
What is likely in the future is not known.

Answer (D)