Of course that number isn't correct, as no car on the road produces that much torque. Even a Bugatti Veyron produces "only" 1,000 lb-ft, give or take. Clearly something's amiss here, but the problem the dyno had in reading the Tesla's torque apparently doesn't come down to its electric powertrain. (Nor does it have anything to do with the northerly latitude or the interference of polar winds.) It comes down to the shiny, ten-spoke alloys.
Apparently this type of dyno measures torque by running horsepower and wheel revolutions through an algorithm. It measures horsepower at the wheel (which, at 436 hp, wasn't far off of Tesla's own rating of 416 hp) and uses a stationary optical sensor interfacing with a reflector on the wheel. Every time the reflector passes the sensor, it counts one revolution. But since the Model S has shiny ten-spoke wheels (and we presume because it was taken outdoors under bright sunlight), the sensor thought that each passing spoke was one revolution of the wheel... when it was, in fact, ten times too much.
The machine can set to compensate for up to a quarter rotation, but not a tenth. And since the software can only calculate up to 2,000 lb-ft, that's the number it spat out. Watch the video to see the Tesla on the dyno.
UPDATE: Turns out there is a lot of discussion on why the dyno registered its overzealous figure. The explanation above comes courtesy of a commenter on Jalopnik, later quoted by DragTimes, who claims to be an engineer at Dynojet, other explanations include the effects of torque multiplication at the wheels and the fact that the Tesla has one single gear instead of a multi-geared transmission. We're not dyno experts, and we don't claim to know which explanation is correct. In any case, the video is interesting, and you're welcome to watch it above.