If you really only need to send fuel, just do that. There's multiple fuel pods and a single robotic vehicle that goes around to each satellite to install the fuel pods. Sure, you're still launching a a big robotic vehicle but that vehicle will service multiple customers. It's cheaper for all of those customers to share the cost of one expensive satellite that to each pay for their own new one. The rest of the equipment on board works great, they have just gone past their expected service life (typically 15 years for GEO).

GEO communication satellites have become less attractive as worldwide internet access gets better (either fiber, mobile, or LEO constellations like Starlink) so it's hard to justify spending the money on a new 15 year mission when it's entirely possible you won't ever sell enough bandwidth to pay for it. It's way cheaper just to extend the life of existing assets so you can better guage if you actually should eventually replace it or exit the market.

Instead of propellant, if a satellite sends down a request for cakes and ale, is NASA legally required to comply with that request?