What do you see as the downside of tubes?

"If there was any way that transistors could be made to make the music that tubes can, I would not be making tube amps"

Atmasphere, is there a scientific explanation for why this is? What is it exactly about tubes that inherently make them sound different than solid state?

Well. actually, a couple of decades ago Bob Carver demonstrated that characteristics of an audio amp can be taylored to match those of any other amp. Also, Behringer, (sorry about that Rodman99999) makes a device that can be added to a system to simulate tube sound. There is a selector switch to choose what tube type you want to match.

But if tubes turn you on, it's probably easier to just use tubes.

Mapman-I sure it has to do with the flow of electrons in solid state junctions of pnp or npn materials vs the flow in a vacuum from cathode to plate thru the grid(s) in your output tube; somehow harmonically there is a difference that can be heard.
I sure Ralph will expand on your question as well if he sees it;thats the best I can do.

Actually, Carver showed that an amp could be *modeled*, but even in his tests the modeling did not match the actual behavior of a tube amp.

Mapman, here are some primary differences.
1) Tubes are often much more linear than transistors. Triodes are the most linear form of amplification known. This results in far less need or no need at all for feedback to reduce distortion.

2) tubes produce mainly lower-ordered harmonics, transistors tend to produce higher ordered harmonics, especially odd-ordered. This can be easily seen in the overload characteristics of the amplifiers where these behaviors are exaggerated. Anyone who has run an amplifier into clipping and observed the output with an oscilloscope knows what I am talking about

3) tubes tend to high impedance while transistors tend to low impedance. In addition, there are capacitive effects in transistors that are magnified at higher currents. The effect of this is that both transistors and tubes will sound better (lower distortion, smoother sound, greater transparency) when driving higher impedances.

Atmasphere...As I recall (and it was a long time ago) after "modeling" Carver inserted networks to match what was modeled. Listening tests by a gang of golden ears were unable to distinguish the modeled amp from the real thing.

May I submit that a D/A conversion (digital amp) can be much more linear than any analog component (eg: tube).

Clipping characteristics are of academic interest. I don't operate my amps into clipping.