40mbps is still about 8 times faster than the internet, so what happens, assuming that the server you're testing with happens to be very close to AT&T's local entry point to the backbone, is that you'll see a 40mbps download speed from that server, but normally AT&T will get a packet for you every .000002 seconds (which equates to 5mbps if I didn't slip a decimal point) and will send it to you at 40mbps, then wait a relatively long time for the next one. Your speed to AT&T is still 40mbps, but your thruput - the speed you're downloading from some server - is 5mbps. So unless your download drops to under 5mbps, or you're downloading and/or streaming a lot of files at once on some super-fast phone, it's not going to make any difference in actual use between 5mbps and 119mbps.
About the only time a connection like 119mbps makes any practical difference is with wifi, when you have 5 or 10 people streaming movies at once, or people downloading GB-size files at the same time others are streaming. Each one will be using 5mbps of that speed, so if there are enough things coming down the line, you can take advantage of that bandwidth. But phones aren't that fast, and the internet itself isn't that fast, so if you're streaming a video at 5mbps, all of a sudden streaming it at 119mbps isn't going to make the movie run faster - the server is still streaming it out at about 2mbps (or maybe 5mbps if you're watching a 4k stream).
Remember, download speed isn't an absolute, it's from somewhere. when the "where" changes, the speed changes. So you can measure speed using the Ookla app (or use a browser to go to speedtest.net) and get a really nice-looking speed, because they have servers all over the place, but the sites you're usually downloading from don't, so you're limited to the speed you can get from, say, Singapore or Illinois. (Use the Ookla app, change servers to someplace in Siberia or the middle of Africa, and watch your ping go up and your speed go down.)
But what you're seeing is why people are told to study statistics before running speed tests. For the one you posted you ignore the 119mbps (they're called outliers) and average the rest - and your download speed from the server you're testing from is 40.15mbps. (Not that an Ookla speed test really tells you anything - your real-world download speeds do.) The few burst of higher speed don't mean anything.