Honeywell used to make an 8051-type uC that was
good to 225C. But (a) it's obsolete and (b) when it wasn't,
they wanted $877 a pop through Digi-Key. They do have
two in stock if you're that kind of desperate.
Nothing else that turns up in a >200C search is in stock
except for a 16/32-bit ARM product from TI (probably one
of the Vorago modified products)
Now I do know, from bouncing off the high temp market
a few times, that most of the serious down-hole instrument
people use automotive / MIL-B grade product which they
have personally characterized for suitability and durability,
because of the pricing I mention (commonly applied to
any product which falls outside mainstream high volume
test, or has a narrow end market, or in this case both).
In general an epi-on-P+-handle or a SOI product stands a
good chance of being useful well outside standard MIL
temp range. I have tested 5V flash ADCs on heavy-handle
JI and ported to silicon-on-sapphire substrates, and both
worked properly out to 300C (at which point, a few months
in, the Delta oven burned itself up and the experiment
was over). But no manufacturer has the infrastructure,
and almost none the experience or the patience (or the
business case) to try and qualify a 300C product with
years of service@temp reilability. Hell, 300C sockets are
hen's teeth, and accelerated testing needs higher....
Anyway it's highly likely that you've looked at some parts
which are indeed capable, but the mfr will not make that
claim because backing it up costs money and holds risk.
Proving lot-level reliability in production for 200C or 250C
is doable. As the only customer you'd be asked to pay
the whole NRE and wait a good long while. You might
end up getting a shorter lifetime guarantee (if you call
such forward looking statements, backed by an experiment
done years or decades ago, as guarantees) than you had
hoped. Science does what science does.
You might find HiTEC (High Temperature Electronics
Conference) publications which let slip, that kind of info.
Last one I went to was back in '93 I think but it's still
going on, annual or biennial, I believe. Even seeing who
is bothering to show up, might give you some leads who
you could hit up on LinkedIn or through author contact
info on the papers.