In Go, we achieve concurrency by working with Goroutines. They are like processes that run applications in a computer concurrently; in fact, the main loop of Go could be considered a Goroutine, too. Goroutines are used in places where we would use actors. They execute some logic and die (or keep looping if necessary).
But Goroutines are not threads. We can launch thousands of concurrent Goroutines, even millions. They are incredibly cheap, with a small growth stack. We will use Goroutines to execute code that we want to work concurrently. For example, three calls to three services to compose a response can be designed concurrently with three Goroutines to do the service calls potentially in parallel and a fourth Goroutine to receive them and compose the response. What's the point here? That if we have a computer with four cores, we could potentially run this service call in parallel, but if we use a one-core computer, the design will still be correct and the calls will be executed...