One of the cases specifically being looked at in this investigation is Washington D.C. Google tells me that around 60% of rentals in D.C. are in large apartment buildings, but that's just a quick search so take that with a grain of salt. ProPublica says that 90% of these apartments are using the software. If we accept these numbers are close, and knowing that 90% of buildings doesn't necessarily mean 90% of all apartments, we can guess that means somewhere around 50% of all rental properties in D.C. are priced using this software. The number might be higher, since I don't know if any non-apartment buildings are using this software, or it might be lower if the data is inaccurate.
So yes, the remaining 50% of the market could undercut the 50% using the software. However, it's still possible for 50% to have an effect on the housing market. This is because the housing market is generally considered to be inelastic and because it is difficult to increase supply quickly.
Compare it to, for example, soda, which is an elastic market with a lot of production. If Coke and Pepsi lost their minds and started charging $100 for an 8 oz can, they're creating a lot of potential for other companies to undercut them at, say $2 a can. As long as $2/can is profitable, they can win Coke and Pepsi's market from them overnight.
If 50% of rental properties started renting for $1 million/week, obviously, very few people can afford that, and I doubt anyone would pay it. But what could the other 50% of the market do? They could keep their rates lower, but all the rentals will fill up. Again, numbers are hard to find, but vacancies in D.C. are probably around 6%, so prospective renters will play musical chairs for the 50% of rental properties that have more reasonable rents. There's a potential here for companies to build new properties and undercut the 50%, but building is lengthy, expensive, requires permitting, etc. The market can't respond very quickly, and because new construction tends to be expensive, rents from new buildings tend to be higher.