The historical filing volume in this corpus is dominated by large semiconductor and telecom device makers rather than specialist optics firms, reflecting the fact that most claim density sits in H01L (semiconductor device structure) rather than optical systems. Several of these same large holders show zero filings in the most recent year, which suggests they are defending existing portfolios rather than actively expanding them. A full ranking by family count, rather than raw document count, gives the fairer picture since it neutralises continuation filings and multi-jurisdiction duplicates.

Filings rose to a peak of 329 in 2019, held near 217 at the 2022 midpoint, and have fallen sharply since, reaching a partial count of 11 by 2026. Part of that recent drop is a publication-lag artefact, since patent applications typically take around eighteen months to publish, so the last one to two years in any dataset are undercounted. But the multi-year decline from 2019 through 2022 predates that lag window and points to a genuine slowdown in new filing activity, consistent with core architecture claims already being staked out.

Both terms appear together in this search scope because they describe closely related device families: avalanche photodiodes amplify photocurrent through impact ionisation, while single-photon avalanche diodes (SPADs) run in Geiger mode to detect individual photons, and both are claimed heavily in H01L filings. Patent claims in this space tend to differentiate on dark count rate mitigation, quantum efficiency and temperature dependence rather than on the base avalanche mechanism itself, since that mechanism is old, well-established prior art. A 2026 filing from STMicroelectronics illustrates this pattern, claiming a dark-count calibration method built on top of a conventional avalanche photodiode array rather than a new base device structure.

The IPC composition shows a heavy concentration in H01L relative to thinner coverage in branches like G01S (radar and positioning integration) and H10P, which points to comparatively under-claimed ground around sensor fusion, temperature-compensated readout circuits, and wavelength-selective filtering at the pixel level. Co-assignee data also shows only ten collaboration pairs across the whole corpus, meaning most filers are working solo rather than jointly developing integration-layer technology. That combination — thin IPC coverage plus low collaboration density — is a reasonable signal of open claim space, though it should be checked against a fuller freedom-to-operate search before drafting.

The United States leads by a wide margin with 2,595 filings in this corpus, more than three times the next largest office, Europe's EPO, at 801. WIPO's PCT route follows at 496, close to Japan's 494, while China and Canada trail at 218 and 163 respectively. That distribution suggests most applicants treat the US as the primary market to protect first, using PCT filings to preserve international options before committing to additional national phases.