With the caveat that I found the general whiny tone annoying:
Direct costs may go down, since they have less URLs to worry about. But value goes down as well, since they have less data to mine. There may be some critical mass of data required to have anything of value.
They're complaining about development costs and bandwidth costs.
I don't buy it.
A url shortener is relatively trivial to operate (at least, compared to some of the stuff I'm doing right now). It works, or it doesn't, once you've got it there isn't much you can do to make it better or add features.
So development costs are almost a one-time affair and it seems they had already done that.
What's left is the bandwidth cost.
A Mbit is about $5 /month in bulk. The average 'GET' reguest from a device is maybe 500 bytes, the answer, the redirect is maybe another 500 bytes. So say 1 Kbyte in total (generous!). 1Mbit / sec = 128 Kbyte, so that is roughly 100 requests / second for $5 /month. Serverload is trivial, a single table with a unique key, you can do 1000's of queries like that per second on a single box. It is a trivially parallel problem, simply distribute all urls across all machines. A single box could push maybe 100K requests / second using varnishd or some other front end cache (tr.im uses nginx), so that's maybe 800Kbit/sec out of one box.
You'll need a bunch of IPs otherwise you'll run out of sockets and you'll need some memory for the cache.
Calculate in some margin and you can cut that down to maybe 500 Mbit/sec, so that box would cost you $2500 / month and it would handle a staggering 62500*86400 = 5 billion requests / day.
No way tr.im is at that level, more likely a few tens of millions of requests per day, reduce bandwidth costs accordingly, say $500 to $1000.
It wouldn't take more than a bunch of google ads displayed to the makers of the urls or a donation to keep a service like that going, that's not a whole lot of $ to make on a site. One of my sites which converts absolutely horrible does that kind of money on 27000 visitors / day (mostly from Finland, don't ask...).
This will never be a business, and personally I really don't understand why the likes of twitter do not give a short url out on their own service. But it also does not cost a fortune to run, that's bs.
And if you want to create 'value' you should do something a little more sticky than a url shortener, it literally is a throwaway relationship with your audience.
And if you lock them out for a couple of days because you can't make up your mind if you're going to sink or swim you have really tossed your investment into the water.
Direct costs may go down, since they have less URLs to worry about. But value goes down as well, since they have less data to mine. There may be some critical mass of data required to have anything of value.