This hurts my brain but it sounds like what you are saying is that there's a "direct" flight UA1851 SJC-IAH-BWI. And when you look at availability:

(1) Availability shown as a solution for a one-way journey SJC-BWI:
UA1851 - SJC-BWI - has inventory F9 T9 L0

(2) Availability shown as a solution for a one-way journey SJC-IAH:
UA1851 - SJC-IAH - has inventory F9 T9 L4

(3) Availability shown as a solution for a one-way journey IAH-BWI:
UA1851 - IAH-BWI - has inventory F9 T9 L0

(4) Availability shown as a solution for a one-way journey SJC-BWI:
UA1851 - SJC-IAH - has inventory at least L1 (let's say it's F9 T9 L4)
UA1851 - IAH-BWI - has inventory at least L1 (let's say it's F9 T9 L4).

The situation described in #4 is totally reasonable (sort of) -- a flight can have "more" inventory when it's part of a connection to a longer journey than when it's searched nonstop. Classic example is UA79 NRT-ICN which in my experience has much better inventory if you are connecting onward like NRT-ICN-SFO vs. as the nonstop.

You know, on most flights, there is an earlier flight SJC-IAH. I **wonder** whether if you do a search SJC-BWI and take a look at inventory on a fare solution that includes a wide-open early-morning SJC-IAH plus UA1851 IAH-BWI, I wonder if you would see L4 (or whatever) on the second sector. I think probably you would.

This is a pretty cool corner case, congrats. I don't think technically there is any reason why this shouldn't happen. After all, the inventory for "the direct flight" and "the individual flights as part of a particular married-segment journey that happens to be the direct flight's city pairs" are different things, under the hood … I think. But the result is not intuitive.