well, actually, a 9.0 has 32 times the energy of an 8.0. It doesn't do 32 times as much damage though.

And I don't recall ever seeing displacements of 100m on a fault rupture - 10-12 is pretty bad and I've read of 16-19 (all metres). Don't know where you got your 100m from but the USGS reports "about 20 meters"

link here

Currently available models of the 26 December main-shock fault displacement differ in many interesting details, but are consistent in implying that fault-rupture propagated to the northwest from the epicenter and that substantial fault-rupture occurred hundreds of kilometers northwest of the epicenter. The data upon which the modeling is based do not permit confident resolution of the extent of rupture beyond about 500 km northwest of the main-shock epicenter. The finite fault model shown here implies that the width of the earthquake rupture, measured perpendicular to the Sunda trench, was about 150 kilometers, and that the maximum displacement on the fault plane was about 20 meters. The sea floor overlying the thrust fault would have been uplifted by several meters as a result of the earthquake.

Note that for subduction earthquakes, vertical movements are typically 1/10 of horizontal ones. For strike-slip quakes such as the San Andreas system, average vertical movements are less than 1% of horizontal but local fault asperities (bends and kinks) can show significant uplift and downfall.

As for Silicon valley, you've been watching too many James Bond films! There's a balance of subsidence and sedimentation there over geologic time. I don't think a likely outcome of even a colossal quake would be major flooding of the valley, but there would be local changes to drainage and damage to water and drainage systems that could lead to small local floods.

Now, the fundamental question, would a 9.0 be possible on the San Andreas system? Well, since it would involve more than 32 times as much energy, thatat strain energy would have to accumulate for 32 times as long - i.e. 3200 years rather than the historic ~100 year periodicity. So San Francisco (and Los Angeles) are safe for a long time from such an event. Additionally, that same web page I referred above has a useful hint:-

Since 1900, earthquakes similarly sized or larger than the 26 December earthquake have been the magnitude 9.0 1952 Kamchatka earthquake, the magnitude 9.1 1957 Andreanof Islands, Alaska, earthquake, the magnitude 9.5 1960 Chile earthquake, and the magnitude 9.2 1964 Prince William Sound, Alaska, earthquake. All of these earthquakes, like the 26 December earthquake, were mega-thrust events, occurring where one tectonic plate subducts beneath another.

it seems to be the case that such massive quakes occur ONLY in subduction settings and that strike slip systems are limited to lesser events.