Except that you can't. 5k is above 'discretionary spending' for a drone, but not worth it to convene meetings with upper management, fly in sales people, do an extensive demo/feasibility study, etc. Spolsky wrote a post about it years ago, and I generally found it's true. There is a gap between 'cheap' and 'expensive' for which it's hard to justify the sales expenses. While doing 1M deal is excruciatingly painful, it's not 200 times more painful than doing 5k deals, so in the end I prefer the one 1M.
The exception is software which is so widely regarded as being good and necessary that it's trivial to expense it. For example: Visual Studio, SQL Server, Oracle, AutoCAD. Software that has acquired a strong reputation can short-circuit the typical management heavy buying decision process. Instead of consulting with the software maker the decision process becomes focused on the question of need, alternatives, and budget.
To date it hasn't been very common for software to live in that niche, but it certainly can.
The management-heavy buying process still applies, however - as support and service contracts become extremely important for large organisations, and are essential to their operations.
There are pros and cons to both. Someone paying you $1M may start to act like they own you. Worse yet, if you don't have much other business, it might be true. Also, the process of selling will likely be long, drawn out, and complicated, wasting a lot of time you could use for development.
Also, it might be a lot harder to grow your business from 1 to N $1M buyers, whereas it might be comparatively easier to grow from 100K people paying $10 to 10M paying $10 or, say, 10M paying $10, 10K paying $1k, etc.
Indeed, a sizable chunk of those 100K individual customers are likely going to find a way to nag, review-extort, and support-whittle your $1M down to almost nothing.
I'd rather that $1M revenue was split down into something more like 200 sales of $5K.