1️⃣ Buy the biggest keyword package from DataForSEO. Probably their US keyword index.
2️⃣ Download the Bing keyword dump that was leaked a few years ago.
3️⃣ Do GSC-to-BigQuery for a couple of large domains.
4️⃣ Use an LLM to create similar queries. Training or fine-tuning might reduce cost and increase quality.
Billions of real user queries would be tough.
Alternatively, ask a large Google Ads spender with broad inventory (Amazon, eBay, Booking, idealo, Kelkoo, RedBrain, etc) to share the list of keywords they are (not) bidding on. I know this is in the billions for some of them. Or at least it used to be.