I think your insight is spot on. We're already starting to see the shift towards "hard/deep tech". My observation is that investors are scrambling away from pure software plays and LLM wrappers, and more towards foundational tech or infrastructure.
IMO that's still "tech moat" thinking. The real moat nowadays is GTM traction, and "creativity/taste". All software can be copied, but the creation of new concepts/solutions and trends, is still uniquely something that requires a deep understanding of the "WHYs" and "HOWs" as I've described in my writing.
I think it's amazing how deep you can go in something as simple as rendering diffs, and I'm really grateful you shared that writeup (in fact I've been following you loosely since the Kiip days as a then aspiring designer+developer type).
I've always been curious how products like ag-grid are able to allow you to lazy-load 10,000,000+ rows in a table without the blanking you described, and I imagine this type of method describes how it can be done.
My question would be where you think this space is moving next? I have my own hypotheses but before sharing I'd love to hear your thoughts!