The maintenance of the ocean general circulation requires energy input from the wind. Previous studies estimate that the mean rate of wind work (or wind power) acting on the surface currents over the global ocean amounts to 1.1 TW (1 TW = 1012 Watts), though values remain highly uncertain. By analyzing the output from a range of ocean-only models and data assimilations, we show that the tropical Pacific Ocean contributes around 0.2 to 0.4 TW, which is roughly half of the total tropical contribution. Not only does this wind power represent a significant fraction of the total global energy input into the ocean circulation, it is also critical in maintaining the east-west tilt of the ocean thermocline along the equator. The differences in the wind power estimates are due to discrepancies in the wind stress used to force the models and discrepancies in the surface currents the models simulate, particularly the North Equatorial Counter Current and the South Equatorial Current. Decadal variations in the wind power, more prominent in some models, show a distinct decrease in the wind power in the late 1970s, consistent with the climate regime shift of that time and a flattening of the equatorial thermocline. We find that most of the wind power generated in the tropics is dissipated by friction in the mixed layer and in zonal currents with strong vertical and horizontal shears. Roughly 10 to 20% of the wind power (depending on the model) is transferred down the water column through vertical buoyancy fluxes to maintain the thermocline slope along the equator. Ultimately, this fraction of the wind power is dissipated by a combination of vertical and horizontal diffusion, energy advection out of the tropics, and damping by surface heat fluxes. Values of wind power generated in the tropical Pacific by coupled general circulation models are typically larger than those generated by ocean-only models, and range from 0.3 to 0.6 TW. Even though many models simulate a 'realistic' climate in the tropical ocean, their energy budgets can still vary greatly from one model to the next. We argue that a correct energy balance is an essential measure of how well the models represent the actual ocean physics.