The target market is important. Example? Consumer or automotive, they have different failure requirements. A TV that fails is not that important as the chip of an airbag. Other things to consider:
- There must be a business case of the number of chips going to be sold. Volume will drive the tech choice (total cost of wafer, mask set, test, …).
- The trade-off power-area-speed is always there. More recent tech has lower consumption even leakage sometimes. And there are quite some libraries you can choose from (if available and if affordable), faster but more consumption and more area or slower and better for power. But more recent tech will be much more expensive than older tech.
- You need to find the resources that have experience in that tech. If the tech is new, you will not find so many. And price per hour will be higher because more demand than supply.
There are a lot of factors to consider. If you have never done an ASIC project and have never gone through the budget calculations, it is important to find somebody who has that experience.