I don't believe the physical antenna itself is difficult or expensive. The expensive part is all the digital phase control circuits that are needed to actively steer the antenna beam. A electronically scanned array antenna is actually a whole lot of little antennas fed by an equal number of phase control electronics. The phase control electronics need to be small and inexpensive (since there are a lot of them) and also able to accurately adjust the phase of a very high frequency signal (24++GHz). That is difficult.
https://en.wikipedia.org/wiki/Passive_electronically_scanned...
Example pulled from a google search: https://www.fairviewmicrowave.com/sma-8-bit-programmable-pha...
$12,000 (less in quantity ;-)
Aha! Thank you.
Not sure how their antennas are constructed - if they are fabbed on a PCB, often at microwave frequencies you need exotic substrate material (e.g., Rogers) to cut down on losses. Those substrates are expensive compared to FR4 (fiberglass common in PCBs).
Metal additive manufacturing has never solved a cost problem AFAIK. The problem it solves is the manufacture of parts that have physical characteristics that are difficult to machine, or in the case of Inconel, materials that are simply difficult to machine period. Neither of those characteristics are going to be a problem for the Starlink antennas.
A small series (when producing other tooling would be too expensive; a single, bespoke part would be the extreme thereof) or ones that would be impossible to machine otherwise are candidates for additive manufacturing.
Mass production is usually less suited for that.
EDIT: Sorry for asking a question I guess, what's so wrong I'm getting so many downvotes?