GFAB6 (a bipolar 6" Fab in Scotland) is being shut down, since it is uneconomic to continue 6" wafer production except for some specialized RF applications with niche materials (GaAs, GaN, SiGe, and a few others). The capacity was about 40k wafers/year or about 110/day, which was apparently not valuable enough to keep the TI bean-counters interested.
The designs could be ported to newer processes on 8" and above, but it requires work and investment to do so. Many of the original NatSemi designers who designed these parts have retired, left or been laid off over the last few years, so it may not be easy or viable to port these designs to new processes. TI also has a partly overlapping series of analog parts from its older Burr-Brown and other acquisitions over the years. Hence, the decision to terminate some of the NatSemi analog parts seems to be driven by economic considerations. In the process, some very unique parts will be killed, with no successors - LME49810, LME49811, LME49830, LME49860, LME49990, LME49600 and LME49610 are examples of unique NatSemi parts with no equivalent or successor part, anywhere in the industry. These are like Rubycon/Jel-Max Black Gates.
In addition, even some of the surviving high-performance bipolar parts (LM3886, LM1875, LM6172, etc.) may be killed in the years to come. The only option seems to be to stockpile a few for future use, while looking for alternatives - I can confidently state there is no equivalent for the LM3886 and LM1875 at their price-points. LM6172 has a few parts from Linear Technology and others that are close, but not exact equivalents - LT1364, LT1208, etc. I suppose one could substitute LME49990 by OPA627/OPA827, but they're not exact equivalents (BJT vs. JFET input), and the latter are 3x to 10x more expensive than the LME49990. ADA4627 is also worth looking at as an alternative.
