They already have fabs, many fabless companies, and plenty of cheap engineers. This is likely money that will go into R&D improving existing tech, I.P. to steal more market, and possibly EDA if they're smart. They don't need to drop $10 billion if they build on what's already out there and do it smart. My plans for a platform SOC were expected to cost somewhere between $10-50 million for 45nm and 28nm by carefully leveraging MPW's and cheap engineers. I could kill whole markets with $1-10bil. My imagination looses detail at the $100bil point.