expectedwrong hindsight

The $100 Billion Trap

On using funding rounds as chess moves, and two tools that actually work now.

2 min read 328 words #openai #knowledge-graphs #text-to-sql #ai-tooling #money
hindsight — still happening

The $100B trap theory is still playing out. Sam is still scrambling. The non-profit conversion is still the most expensive chess game in technology. Nobody is writing about it this way because superintelligence is funnier.

The $100 billion going into OpenAI isn't really about OpenAI. The theory — and it's a good one — is that the point is to corner Sam. You drop $100 billion into the company, you raise the implied valuation, and now Sam has to find even more capital to buy out the non-profit on his way to full for-profit conversion. The funding round as a chess move. Checkmate expressed in commas on a wire transfer.

It's a beautiful trap, if true. The non-profit has to be bought at fair market value, fair market value just got much larger, and whoever dropped the nine figures gets to sit back and watch Sam scramble to find a number that rhymes with $300 billion.

Nobody writes about it this way because it's funnier to write about superintelligence.

Neo4j shipped a knowledge graph builder and it's open source and it actually works — documents in, graph out, pipeline you can stand up without hiring someone who did their dissertation on RDF. Knowledge graphs have been the "this is finally the year" technology for roughly a decade, always almost ready, always requiring a consultant and two weeks of ontology design before you could graph your first entity. This one is different enough that it's worth saying out loud.

Whether that means knowledge graphs are finally happening or Neo4j has simply built the nicest waiting room in enterprise software, I genuinely don't know.

WrenAI has come a long way. Six months ago it was a promising GitHub repo with a demo that worked if you squinted. Now it's a semantic layer you could actually show someone — English in, SQL out, warehouse underneath, answers that aren't immediately embarrassing. Text-to-SQL has been a graveyard of almost-good tools for years, but this one kept shipping and it shows.

Three observations. One involves billions of dollars being weaponized. Two involve open-source tools getting quietly good. The ratio feels about right for a Tuesday in February.