Here's another approach to this question:
From How long can a Galaxy class starship last before it needs servicing?, the Enterprise-D can carry 3,000 m^3 of anti-deuterium, which is enough to keep the ship running for three years of normal operation (source: Rick Sternbach and Michael Okuda's Star Trek TNG Technical Manual).
Based on data from the Brookhaven National Laboratory, I'll estimate the maximum density of deuterium (in liquid or solid form) at ~ 0.2 g/cm^3 = 200 kg/m^3 (since it would be a liquid or solid, even vastly increasing the pressure wouldn't change the density much). Since anti-deuterium should have the same density:
200 kg/m^3 x 3,000 m^3 = 600,000 kg anti-deuterium
Now let's assume the Enterprise-D's engines could convert that anti-deuterium to energy with 90% efficiency (the manual specifies a minimum efficiency of 88% up to warp 7.0), by combining it with an equal amount of normal matter. Using E = m c2, that would give us a total of:
1,200,000 kg x (3.0 x 10^8 m/sec)^2 x .9 = 1.0 x 10^23 kg m^2/s^2 = 1.0 x 10^23 J of energy
[J = joules]
And that total energy output, sustained over a 3-year period, would give us an average power output (for propulsion, which should be the main power consumer; total will be more, since they'll also be running the stereo and A/C) of:
1.0 x 10^23 J/(94,608,000 s) = 1.0 x 10^15 J/s = 1.0 x 10^15 watts = 1000 terawatts
[There are 94,608,000 seconds in 3 years.]
[By comparison, total current worldwide energy generation (all sources -- coal, gas, oil, nuclear, hydroelectric, wind, solar, geothermal, etc.) is about 15 terawatts = 2 kilowatts/person.]
We can compare this figure to one that can be estimated from the power usage chart for the engines (Fig. 5.1.1. p 55), and accompanying explanatory text, in the same Star Trek TNG Technical Manual. On p 57, Sternbach and Okuda say the Enterprise is able to cruise for an unlimited amount of time (until its fuel is depleted) at warp 6. So let's assume that's our average cruising speed. Now according to Fig. 5.1.1, the power usage (for propulsion) at warp 6 is 3 x 10^6 MJ/cochrane. Of course, those are the wrong units; since it's power, it should be MW/cochrane (MW = megawatts). So let's make that correction.
They further say that a warp 6 field bubble has a field strength of 392 cochranes. Thus the power required for propulsion at warp 6 is:
3 x 10^6 MW/cochrane x 10^6 W/MW x 392 cochranes = 1.2 x 10^15 watts = 1200 terawatts
This is nearly the same as the first value we calculated! [This is probably serendipitous :).] Of course, as mentioned above, there are other power consumers besides propulsion, but I'm assuming that's the big one, at least for sustained operation.
We can also use the graph and figures in the technical manual to estimate a maximum power output. At its maximum theoretical speed of warp 9.8, we have:
8 x 10^9 MW/cochrane x 10^6 W/MW x 2 x 10^3 cochranes = 1.6 x 10^19 watts = 16 million terawatts = 16 exawatts
This is very close to the 12.75 million terawatt (= 13 exawatts) figure quoted by Data (though I don't know how fast the ship was travelling at the time). At the same time, the 13 and 16 exawatt figures seem a little silly to me, even for 24th–century technology, since they're over 100 times the power the earth receives from the sun (174 petawatts)!
Furthermore, at a 90% conversion efficiency, the engines would need to dissipate 1 exawatt of heat, i.e., 10 times the power the earth receives from the sun! [Additionally, according to the tech manual, conversion efficiency tends to decrease at high warp speeds.] Though I suppose they could deal with this by saying the heat is dissipated into subspace...
Interestingly, the Wikipedia article referenced above says the Enterprise-D can maintain emergency warp, 9.6, for 12 hours. Using the same sort of estimates given above, that would require 11 exawatts of power. However, at that power output, the ship would use up its total fuel capacity in 3 hours. So clearly there's not perfect consistency among these different specifications.
Finally:
A 1PW laser accelerator is a thing. Those on the enterprise must —by virtue of needing to go much further and charge much faster— need more power.
It's important not to conflate peak power output capabilities (of things like lasers) with sustained power output. Today we are capable of building a pair of lasers with a combined peak power output of 20 petawatts = 20,000 terawatts. But this device will put out that power for only 150 femtoseconds = 1.5 x 10 ^-13 s, thus delivering a total energy of 3000 joules. It can do one shot/minute so, as impressive as its peak power output is, its sustained power output is only:
3000 J/min x 1 min/(60 s) = 50 J/s = 50 W
And remember that when we are talking about the power output of the Enterprise-D's warp engines, we're referring to sustained power output.
[Notably, the peak power output of these lasers is >1000x the current 15 terawatt sustained power output of human civilization!]