If you look closely at the published torque curve and hp curve for the intake mod, they do get more hp at the high rpm range but it is at the expense of torque (and hp) at the low end of the rpm curve as the stock air intake had been tuned for low end torque (and low noise).
I've been following aftermarket engine mods offered by various aftermarket companies for motorcycles, cars and trucks for over 30 years. With well-tuned modern engines there is often very little extra power to be had without making big compromises in other areas (most often low end torque, smooth throttle response and fuel economy or range). The one thing that has not changed over the years is the Dyno charts published by aftermarket suppliers are notoriously optimistic. Seriously, how many units would an aftermarket manufacturer or supplier actually sell if the charts showed minimal gain with big flat spots in the torque/hp curves? Their goal is to increase internal profits without messing up the driveability of your vehicle so much that customers start badmouthing or returning their products (and if it makes the engine sound a little louder many will be fooled into thinking it's actually increasing performance).
Many consumers don't even know much about how to read a dyno chart. Let's look at the chart published by CorkSport for their CX-5 intake mod and, for the sake of discussion, let's assume it is representative of adding a straight intake to a CX-5:
http://www.corksport.com/corksport-2010-cx-5-skyactiv-power-series-short-ram-intake.html
The main problem with this chart is not what it shows but, what it hides. As was properly pointed out by GAXIBM, drivers spend 95% of the time in the lower rpm range of their engine. My CX-5 spends a considerable portion of it's time cruising between 1200 and 1800 rpm's in 6th gear. Driveability in this range is very important to achieve the fuel economy that makes the CX-5 stand out from it's competitors. In otherwords, the CX-5 doesn't make much torque or hp here but it is enough to allow the vehicle to cruise in this range and maintain speed on relatively flat roads in the higher gears as long as a burst of power is not needed to make a substantial change in speed or climb a bigger hill. It has just enough torque to pull over small rises or slowly increase speed if needed (without downshifting).
But here's the rub. The dyno chart published by CorkSport truncates everything below 32 hp (keep in mind the first VW beetle had a maximum output of only 25 hp). They don't even want to show you what their product does to the engines performance in this critical region. However, you can get a clue of the lurking disaster by looking at the torque curves (the little bit they show you) between 1800 and 1900 rpm's). Ignore the fact that CorkSport was not interested enough in accuracy to provide a torque axis with correct numbers on the right margin (regardless, you can still see the lurking disaster on the left side of the torque curve). Specifically, the chart shows the stock CX-5 engine as making almost twice as much torque at 1900 rpm's compared to the engine with the modified intake.That means the modified engine cannot be practically run in this region unless going downhill, perfectly level with a light tailwind, gradually slowing down or in a lower gear. What happens below 1800 rpm's, they don't show us and yet there is no good technical reason why the Dynojet can't do this.
It is for this reason I do not believe their claims about increased fuel economy simply by bolting on an air intake that is more crudely designed than the stock Mazda intake (because downshifting would be required much more often to deal with small hills, headwinds or light acceleration).
As to their claims of about 3-5 more ponies between 4500 and 6400 rpm's, that is certainly possible but (as already pointed out) this is of dubious value on such a vehicle and I have no way to verify without spending a lot of time/money. How often do you rev past this? If my pre-purchase test drive of the vehicle did not convince me beyond a doubt that it already had enough power, I would have bought a more powerful vehicle, not slapped on a crudely designed air intake that provides less smooth throttle response, more cabin noise and less low end torque. Air intakes on modern cars are designed using sophisticated sensors and computers to smooth out airflow through the rev range by suppressing resonance that can occur at certain rpm's/throttle openings. The CorkSport Dynochart does not represent any scenario except wide open throttle so it really can't depict all the disadvantages such a crude modification will have in real life.