Viewers will forgive a soft webcam. They will not sit through bad audio. If you have one budget line for your first stream, spend it on a microphone and put it close to your mouth: a $129 condenser six inches away beats a $400 dynamic mic on the far side of the desk.
Lighting is second, and it is cheaper than people think. One panel at 45 degrees above eye level does more than any camera upgrade. Two panels — key and fill — is where most streams stop needing to improve.
The capture card only matters if you stream from a console. PC streamers can skip it entirely and use software encoding until the CPU tells them otherwise.
Camera last. By the time you have good audio and light, a mid-range webcam looks better than a mirrorless camera in a dark room. Upgrade it when your audience asks, not before.
