Heat pumps can actually be over 100% efficient if you’re measuring it based on heat produced. Because heat pumps aren’t designed to produce heat; They’re designed to move it around via refrigerant. And if you can use 1w of energy to soak up and bring in 3w of heat, you’re now 300% efficient.

So by that metric, a server would be a “bad” heater. It would still contribute to your heating, but not as much as a heat pump would. It doesn’t mean the device is below 100% efficient, it just means the bar for “good” heaters is much higher than 100%.

Every electronic device is 100% efficient after the electricity has already been generated and delivered, sure, but a bunch of efficiency losses occurred before that. If you're comparing methane burned on-site in a furnace to methane burned at a power plant, transmitted to the site as electricity, and then used for electric resistance heating, burning on-site is gonna be better even if the furnace loses more heat up the chimney than the power plant does.

Also, a heat pump is "300%-500% efficient" in the sense that it moves 3x-5x as much heat as it uses. See also: https://en.wikipedia.org/wiki/Coefficient_of_performance