I've been through the R&D cycle many times on many products. Engineering a derivative of something that has been done before is easier than doing something cutting edge. It's also better to have all your R&D resources focused on one product instead of two (as they had with the Model X and 3 in development at the same time). Having a factory built for 500,000 cars a year that is currently making about 80,000 is also a plus.
However, most engineers are terrible at making schedules and Elon Musk is worse than most. I learned the hard way to make an accurate schedule, you need to break things down into small tasks, on the order of a day or two, assign an estimated time for each task, add up all those times, and assume each person only works a 40 hour work week, then pad the total by 20%. If you do all that, you will be done almost exactly when your schedule said you'd be done. Everyone will end up working more than 40 hours a week for at least part of the project too. Working 50-60 hours or more is your contingency dealing with the inevitable things that crop up that take longer than you thought.
Elon Musk is good at breaking things down into small tasks, assigning estimates and rolling it all back up. The problem is he tends to assign the most optimistic times to each task, doesn't build in any contingencies, and works on the assumption everyone is working 60 hours a week instead of 40. When the inevitable problems crop up, the schedule slides.
It takes a lot of discipline from top to bottom to make schedules like I did above. It's painful to go to your manager and tell them that's what it's going to take and just about all managers will push back. The one time I was allowed to go through that whole process on a large project, we finished within a week of when I said we would. All the other parts of the project allowed their managers to push them into shorter schedules with fewer resources than they needed and everyone but my part had problems and delays. In the end my part of the project was the only milestone that hit even a year within the target.
This was when I was at Boeing when they were developing the 777. In the end the plane ended up 2-3 years late and $1 billion over budget. One guy in my group was responsible for $100 million of that cost overrun.
If Tesla is careful and focused, they have a shot at getting the Model 3 into production by the end of 2017, But even established car companies with much larger R&D budgets than Tesla, better established supply networks, and lots of factories take at least as long as 1 1/2 years to get a car from the prototype stage into production. Look at the new Lincoln Continental due out this fall. It was a year ago Lincoln showed off the concept/prototype at a car show. Press reports I read at the time pointed out that the concept they showed off was much closer to production ready than most concept cars, but it won't be available until this fall.
Getting from where they are now to a production car is no trivial task, even if all the tech is well established.