Maybe it is, but not completely. There will be need of low cost easy document extractions. Yes, if the cost factor goes down, OCR would be obsolete.
I work in these domain, and can say, a lot of things are changing quite rapidly, LLMs and VLLMs are taking the extraction industry by storm.
Its quite easy to one shot an extraction with VLLMs given the intricated details might be missed by OCR and with VLLMs, its like a brain powered visual detector. It really understands the context and its knowledge really comes in the clutch to parse almost any type of document.
Really excited to see where my journey leads in this development