That's how it went for Kepler, Maxwell and Pascal. They introduced the Titan before they introduced the equivalent x80Ti. For Maxwell and Pascal the x80 launched a very long time before the ti, like 6-9 months. Kepler was sort of a transition period but the 600 series is basically like "year one Kepler" and 700 was "year 2 Kepler", whereas the ones after Kepler simply switched to an official two year cadence instead of doing a whole new range every year, but technologically it went similarly with little Kepler one year, then the Titan, then the cheaper Titan derivative the next year.
Now you might say "why wouldn't they follow Turing's lead if they changed things? Why expect it to go back to the old trend?" The reason we expect it to go back is because the reason they didn't do it for Turing and went for a simultaneous launch was because Big Turing was necessary to outperform the 1080ti from the previous gen. For Kepler, Maxwell and Pascal the 104 class chip (680, 980, 1080) outperformed the preceding generation's top chip meaningfully. For Turing, the 2080 was about equal to the 1080ti. This meant that if they didn't launch 2080ti, there was no upgrade path for 1080ti users. Thus, they had no real choice but to release it.
Since we expect Ampere to have closer to a normal generational uplift over Turing thanks to the big node shrink, we also expect them to release 3080 but not 3080ti day one.
