That came at a cost and the market size of people that really really wanted / needed that field toughness was considerably smaller than the general office usage market.
Whereas if not for that, you could do both. Design a solid chip and then put dozens of them in a big iron cabinet for big money but also offer desktops with just one of them for prices that compete with Dell. Except that Dell's customers expect to open their existing Office documents and run their Windows API proprietary software and then won't buy from you.