The cold-start problem is the early phase where a network has too few participants to be valuable to any of them. Andrew Chen's 'Cold Start Problem' enumerates standard solutions: pick a small atomic network (a city, a campus, a vertical), seed one side first (often supply), use single-player utility to attract one side without the other, and use 'tipping' moments to break out of the niche.
Getting through cold start typically requires unscalable founder work, manual recruiting, hand-curated supply, or even running the supply side as the company itself in the early days.