Whoever set or influenced the 40 nm requirement, it could simply be a compromise between three conflicting goals:

• Minimum cost in terms of development time and money. This goal would be achieved with the 100 nm node size, because they have been there already. However, the node size would not yield the speed and power savings of smaller node sizes.
• Maximum deployability. The larger the node size, the larger the deployability in existing fabs.This would also be an argument for 100 nm, however:
• Minimum node size. The smaller the node size, the larger speed and power savings. They could have strived for 14 nm, however, the time scale would have been much too long and the necessary money way, way over everybody's head.

For whatever reasons, they chose 40 nm as the best comprise between these goals – in spite of, say, 55 or 32 nm.

Scaling further down will happen in the years to come …