It's likely possible, but it's more a question of whether it's optimal and what the potential costs and drawbacks in terms of efficiency and usage might be for persisting with an arbitrary cap that was arguably only meant to be temporary.  If demand vastly outpaces available throughput, this results in a backlog and poor service.  I assume you mean "possible, whilst still being efficient"?

The biggest question we currently face over developmental direction (in my opinion) is whether off-chain transactions are completely optional, or if they become the only viable way in which to transact because there isn't enough space on the backbone of the network to support all the participants.  There's also the all-too-apparent matter of rival implementations, possibly gaining popularity, which express no interest in maintaining a 1MB cap.  Ultimately, customers vote with their feet.