Even if you are running it constantly, the per token power consumption is likely going to be in a similar range, not to mention you'd need 10+ macs for the throughput.