The dataset shows annual filings rising to a peak of 71 in 2018, then falling through subsequent years to around 30 by 2022 and lower still afterward. This tracks the broader commercial story: crystalline silicon module costs fell sharply over the same period, squeezing the cost advantage that originally justified heavy CIGS/CdTe R&D investment. It does not mean the technology stopped working, only that fewer new claims are being filed into an already dense space each year.

Every assignee tracked in this dataset, including equipment suppliers like Applied Materials and electronics manufacturers like Samsung SDI, shows zero filings in the most recent tracked year. Because publication lags actual filing by roughly eighteen months, a zero in the latest year is not conclusive evidence that a company has exited the field. The more reliable read is the multi-year trend: filing has slowed broadly across the assignee base rather than concentrating activity in a shrinking set of leaders.

This search string does not separate CIGS and CdTe claims into distinct counts, so the 1,511 families reflect both absorber chemistries together wherever they intersect with deposition, buffer-layer, uniformity, stability or material-availability language. In practice, CdTe-specific claims tend to cluster around absorber stability and cadmium-related material handling, while CIGS claims more often touch multi-source co-evaporation or sputtering steps captured in the deposition and buffer-layer terms. A targeted search restricted to one chemistry would be needed to split the corpus cleanly.

The clearest gap is process-level rather than device-level: H01L (semiconductor devices) covers 1,500 of 1,511 records, while deposition-specific coating claims under C23C sit at only 136 and electroplating-based routes under C25D at 35. Large-area uniformity control, Cd-free buffer layer substitution, and flexible-substrate lamination are all named in the underlying search terms but show up far less often in the highest-cited records, suggesting these are still open for narrowly drafted process claims.

Co-assignee pairing is limited — only ten pairs appear across the full dataset, and the strongest pairings link a single corporate assignee to named individual inventors rather than to another company. Combined with the wide spread across receiving offices (the US, China and South Korea each carrying a substantial share), this points to a field where ownership is fragmented across many single-assignee filers rather than concentrated in a small set of cross-licensed players.