OK, FUD patrol checking in.

"Control four chip manufacturers then you could control 90% of [ASIC chip] production". 90% of ADDITIONAL production, maybe. You still would need to produce > 50% of all hashing capacity that already exists. And you need to hope no other chip designs/manufacturers emerge (since SHA2 hashing isn't a terribly difficult task, that requires a leap of faith. Good luck blowing your $400 million on that gamble.

"By controlling two mining pools you could get more than 50% of the network" Mining pools with 50% can censor transaction but they cannot double spend unless they first take that mining capacity offline (to mine a private fork of the blockchain). Do a Google search for

"Downplaying statistically possible double-spending risks" (http://www.ofnumbers.com/2014/08/18/downplaying-statistically-possible-double-spending-risks/) for more on why pools with > 50% are not the same as having your own equipment that exceeds 50% .

"Control one dozen or two dozen mining centers you could get 50%" Again ... that's a $200 million (or more) investment. If the intention is to have just censorship power then maybe that's feasible. Otherwise what is needed for this hashing capacity to be used to double spend is to take it offline and begin mining a private fork. Just like with pools, taking this action would be something noticeable before the damage is done (six confirmations on a much slower fork have passed) seriously lessening the chances of a double spending attack using this method causing serious harm to the big players (who presumable are monitoring the rate of block solving or have hotwallets or other withdrawal limits that woudl protect from such an attack).

Cool story though.