The original announcement wasn't about a new theorem, just a new proof, and it was entirely possible he had one. Real nice of him to withdraw and tweet it.

New proofs of old theorems, including significantly simplified ones aren't unheard of. Selberg-Erdos' elementary proof of the prime number theorem is a famous one. A lesser known example is a one paragraph proof by the then-high-schooler Noah Snyder of Mason's theorem, which is the functional variant of the ABC conjecture!