The argument is that there is also a burden/liability cost of bringing on an additional 25,000 citizens. They will have students that go to NYC schools, they will have disputes that need police officers, they will have emergencies that need EMTs/Firefighters, and they will have have cases that may burden the court system. They will also use the subway and they will litter and they will wear down the roads. These add burden costs to the city.
Yes they will easily generate more than $6,100 in revenue per year for the city from income tax alone (6.65% is the NYC current tax rate in the $79,000 - $215,000 income bracket. This would bring in $7,248.50 a year based on the average of $109K income from income tax alone. Yes these employees will buy lunch everyday, they will buy TVs and iPads and bed sheets and alcohol and stuff. They will pay 8.85% sales tax on all these goods (only 4.5% of which goes to the city). If we assume they spend 50% of their income on sales taxable goods then they would pay another $9,646.50 a year in sales tax.
So the question is how much on average does one citizen cost the city (the burden cost). As long as it is less than $16,895 (the sum of both previous figures) than things should be ok.
The city also monetizes its' citizens through other means such as traffic tickets, property tax, state lottery sales, cigarettes, alcohol purchases, etc. So this $16K number is a low-ball figure.
I am inclined to believe that the burden cost of a citizen is well below $10k a year. For easy math if we assume that a citizen does cost the city $10k per year, and the tax subsidies do cost $6,100 per year then that means the overall burden cost is $16,100 which would be the break-even point for an amazon employee. Money the city can earn on that employee above that figure would generate a net-benefit for the city. Anything below that figure would be a net-determinate to the city.