One campaign with all cities is fine to start. But set location targeting to 'presence' (people IN those places), check the 'locations' report monthly, and add location bid adjustments when one city converts meaningfully better or worse.
Split into per-city campaigns only when the data demands it: different budgets per market, different economics (a city where jobs are bigger deserves higher bids), or a city big enough to earn its own ad copy ('serving Franklin since…'). Premature splitting fragments data.
The waste hides in two places: the interest-vs-presence default, and claiming a service area bigger than you profitably drive. Advertise where you'd actually take the job tomorrow morning.