To further prove my point. Yesterday, HTC announced a phone running the latest and greatest Tegra 3. Later that afternoon S4 was announced.
There you go.
a) Wait for a quad-core Krait. Given HTC's close relationship with Qualcomm, they likely know it won't be for a while
b) Wait for a Tegra-compatible LTE chip, which hasn't been proven in the market yet. Still late to market, and with parts of questionable integrity.
c) Fall back on the dual-core chip and ship now. Lose a bullet point in the US market, but actually be on the market.
When the Tegra 2 showed up, it was not generally considered suitable for phones; the one found in a few phones is a slightly different model to the original. Also, the Tegra 2 was quite flawed; it lacked NEON, in particular, hurting software compatibility.
It's also possible that supplies of the S4 are restricted; it's a 28nm part, and TSMC's rollout of their 28nm process hasn't gone as smoothly as they might like. The Tegra 3 is 40nm, which is now a mainstream process.
[1] http://www.theverge.com/2012/2/26/2823824/htc-one-x-att/in/2...
But in this case, it's because Nvidia chose to go with the more marketable "quad core" than with a "faster dual core". Also because Nvidia's phone version of chips ships months after the tablet version, on top of the already several months delay of the Tegra 3, which seems to happen all the time with Nvidia lately.
I agree with your sentiment that technology is evolving fast these days, though, and there's little breathing room for technology companies.
Interesting view... How do you define obsolete?