Is proprietary data the most valuable?

There are arguably very few proprietary datasets. Datasets with no competition. Is proprietary data the most valuable asset?

How many truly unique, proprietary datasets exist? Twitter, or X, was one of those datasets. They sold at one point the whole firehose or a small 10% sliver of the feed. When GNIP was still around, you could buy keywords. The price was expensive but arguably justified maybe because there was nothing else like it.

Now that Twitter is no longer selling data, or hasn’t decided if they will, or has increased prices tremendously essentially making it inaccessible to 99% of buyers, who will step in and be the new competition? Or will there be Twitter-like products for different niche verticals? Reddit has been known to sell its data and I have heard it has succeeded. What about StockTwits? StockTwits has one of the longest historical datasets, similar to Twitter, but focused exclusively on the markets. They even invented the cashtag $.StockTwits did sell access to the public firehose in the past.

Is it time for them to reopen the firehose? Is the data even more valuable now in a world of LLMs dominating discussions? Should every new ChatGPT for Finance be licensing StockTwits? Do the quants want the firehose feed? Or a derivative of the feed? Or sentiment feeds built on top of the feed?

In a world of exclusive, proprietary data being so valuable, I wonder what other unique datasets have been in existence for many years and are under-monetized. Maybe the opportunity is to roll up these undervalued data-rich assets and create a modern data & information services company!