They seem to be incredibly good - but aren't there other ways to make a high-rpm rotating shaft (e.g., an automobile axle) rotate at lower friction?

What comes into a question is UHMWPE (ultra-high molecular weight polyethylene). I think it holds potential for direct contact bearings, in the form of two exact-fit cylindrical pieces, with room somewhere along the interior of the outside cylinder for high-pressure grease to be squeezed in. I was thinking that maybe grease can operate better than ball bearings for extreme high RPMs since it is a non-Newtonian fluid.

I don't know what you mean by "two exact-fit cylindrical pieces"? Do you know how long ball-bearings - as "one exact-fit rotating piece between two surfaces" - have been used throughout history?

Two torsional shafts - one exactly circumscribing the other.

you will never be able to build one that exactly circumscribes the other, tolerance issues demand that it will inevitably have some space in between.

but if you want almost no friction at all, consider magnetic bearings. a shaft is literally levitated in place by a series of magnets, with nothing but air or even vacuum surrounding the rotating shaft! not ideal for supporting a whole lot of shear or axial force, but it's the closest we can probably get today to frictionless. the only friction i believe is energy going into making residual current in the metal of the magnets, which is very, very small compared to energy lost in bearings or grease.

There is state of the art bearing technology based on magnetic fields that is claimed by its manufacturers to be virtually frictionless.

I have recently read announcement of clipped nanotubes being a virtually frictionless bearing material. I have coined the word neonaotube to refer to such.