If you go back to 1993, SPARC, PA-RISC, MIPS, and Alpha all seemed like serious contenders for the next big CPU architecture. To my knowledge, none of them supported unaligned access (though I'm sure someone will point out a counterexample). 20 years later, all of them are dead and x86_64 and ARM are the new "modern architectures." x86_64 supports unaligned access and ARM has been gradually gaining support for it.
So yes, chips that can only do aligned memory access do tend to be older.
If you go back to 1992 you will find PowerPC which is still around. Same as MIPS.
Listen, I know that there are people who worked into retirement without ever seeing anything other than x86 but a) this shows how old x86 is and b) does not give them expertise over different ISAs.
I guess MIPS and PowerPC are still around somewhere, but it's definitely a lot less common to see them these days. For example, it used to be that every Mac came with a PowerPC, and so people were aware of the issues that chip had. Nowadays it's rare to find a programmer experienced with anything but x86. That was really Eric's point when he made that comment.
It's less common because they're hidden inside other equipment. E.g. my RAID cards have PPC cpu's; my set-top box used to be MIPS. They're both licensing hundreds of millions of cores a year.
MIPS is still in the top 4 or so CPU architectures by volume, with x86, PPC, and ARM, with ARM as by far the largest (forecast about 3 billion cores licensed last year).
At least a few hundred million MIPS cores are shipped every year. Depending on what you think embedded x86 volume is, it may be ahead of x86. Of course by revenue x86 still beats everyone.
Intel actually put in an feature to trap on unaligned access starting with the 486 (and it still exists today) - look up the AC "alignment check" flag. Of course almost no one used that feature since x86 supported unaligned access from the very beginning, and I think this flexibility is one of the reasons why it's stayed competitive: lots of features initially, not all of them optimal (e.g. division instruction on the 8088 was extremely slow), then improve on them with each new microarchitecture, the result being that existing software benefits from these improvements. In contrast, an architecture that didn't support unaligned access would not do much for existing software if faster instructions for unaligned access were introduced since they wouldn't be used.
So yes, chips that can only do aligned memory access do tend to be older.