My timeline is flooded with GPT-6 Astra generations. We saw something similar earlier this year when Google shipped Project Genie on Jan 30 and the engine names got taken apart, with $U down more than 20% in a single day and $RBLX and $TTWO going with it. The fear was that if a model can generate a playable world, nobody needs an engine. The difference now is that the models are working with the engines. OpenAI published a Playco case study on Sept 3 where an agentic IDE plugs directly into Unity and Godot, and they got three playable prototypes off one grey box with 50% fewer manual fixes than the previous model. I don't think AI replaces game engines, at least not yet, because generating a world and running one are separate problems > state that persists across sessions > physics that behave the same way twice > bugs you can actually reproduce > players staying in sync across devices So while these names keep getting lumped in with the software stocks getting commoditized, I think there could be an opportunity here.