Back in 2014 some of the early ASICs could (like Zeus products if I recall correctly).
Key is that Litecoins "scrypt" is just nScrypt with a N=10. The Zeus products (and the promised-but-never-delivered GAW Vaultbreaker) could go up to N=12, but with rapidly declining hash rates. Vaultbreaker was suppose to be able to handle N=14. Its why alt-coins like IMACredit used an N=16, but they were before their time. Back then, it was all about the fastest hash, unlike today where things like ZCash's SOL run at similar speeds and are obviously acceptable.
What I didn't know is if the L3 had N hard-coded or not.
Key is that Litecoins "scrypt" is just nScrypt with a N=10. The Zeus products (and the promised-but-never-delivered GAW Vaultbreaker) could go up to N=12, but with rapidly declining hash rates. Vaultbreaker was suppose to be able to handle N=14. Its why alt-coins like IMACredit used an N=16, but they were before their time. Back then, it was all about the fastest hash, unlike today where things like ZCash's SOL run at similar speeds and are obviously acceptable.
What I didn't know is if the L3 had N hard-coded or not.
Well, L3 has N factor hard-coded. (Scrypt-N) N factor is a value which increases the memory requirement to compute the hash for the same algorithm which renders the ASIC with less than required N factor obsolete.
It is similar to DAG implementation for Eth hash based coins like Ethereum where the memory requirement for mining it will increase over time, thus producing ASICs for them quite a challenge.