In a nutshell: it's war from day #1, the only advantages you have is that you are more nimble and have less burn so changes in conditions will work to the detriment of your opponent. But they likely have more capital and a solid revenue stream as well as more mindshare with their customers.
If you ever plan on going head to head with an incumbent I'd take a leaf out of the playbook of the flea, find a single customer somewhere that you are going to make totally happy whilst conserving those two advantages, keep your burn down and focus on the speed of your turnaround in the interaction with the customer. Then, when your customer runs out of ideas for you to add to the package generalize as much of it as possible and pull in customer #2. Never retire a feature that you built for someone else but do focus on the emerging 'core platform' of features, the overlap between your customers and move peripheral stuff into its own environment (in software: a separate repo with customer specific stuff).
With a more physical product it's going to be harder (but then, everything is harder with a physical product) so you'll need to either go slower or have deeper pockets.
