Not a power issue but a feature issue. No ray tracing stops Indiana Jones and Doom Dark Ages (though you can do it in software on Linux): https://youtu.be/aU2qwlCLWm8 . Doom Dark Ages also added a check for Vulkan Variable Rate Shading, requiring a workaround to spoof it. Mesh shader requirement prevents Final Fantasy 7 Rebirth from running.
Unless you're on the absolute newest stuff with DisplayPort 2.1, HDMI 2.1 has more bandwidth than DP1.4. That'll be Nvidias 2000 through 4000 series. No DisplayPort 2.1 until the RTX 5000s.
And then monitors released during this time generally do the same too.
Also if you want to use it through a capture card, HDMI ones are way more common and cheaper
It's a bit more complicated than that (on Windows) because Steam doesn't make a virtual gamepad to the OS. The way Steam handles the input is by hooking into the games individually. So to use Steam for other games, you need to add them to Steam as non-steam games.
Even open source controller remapping tools (not just Steam Controller) and similar used ViGEmBus which is no longer maintained. You can have it do mouse/keyboard though, those don't require custom drivers.
Seeing as the original Steam Controllers kernel drivers were community reverse engineered rather than Valve contributed, I don't know if I believe in them to make one for the new one either: https://www.phoronix.com/news/Steam-Controller-RE-Kernel
Community servers don't want server-side anti-cheat either. Hell they invented client-side anti-cheats back in the day. Even current day community servers like Face-IT have additional anti-cheats, not less. Same with modded GTAV FiveM (even before the main game added anti-cheats)
I'm using the smaller vision models (Qwen3.5-4B currently) with Frigate, a FOSS self-hosted "AI" NVR. It's good enough at analyzing images to figure out mostly what's happening, and doesn't require the big knowledge base that bigger models have.
Also use a bigger model for summarizing or translating text, which I don't consume in realtime, so doesn't need to be fast. Would be a thing I could use OpenAI's batch APIs for if I did need something higher quality.