Within this 897-record dataset, one assignee leads with 42 records, with the fifth-ranked company at 29 and the tenth at 14. The top 5 combined account for 19.5% of all 897 records, meaning leadership is meaningful but far from a monopoly. The remaining field spreads across a long tail of the other 95 ranked companies, most holding only a handful of filings each, so freedom-to-operate work should not stop at the top few names.

Filing rose from 12 records in 2017 to a peak of 115 in 2021, then declined to 94 by 2024 — an 18% pullback over that span, using 2024 as the most recent year that can be treated as complete. Counts for 2025 and 2026 look lower still, but that is largely a publication-lag artefact: patent publication typically trails filing by around 18 months, so the most recent years are always understated at the point of measurement. Reading the recent dip as a genuine slowdown would be premature.

Polymer processing and solutions (IPC class C08J) is the single densest area at 28.8% of the 897 records, followed by plastics preparation (B29B) at 20.8% and polymer compositions (C08L) at 20.6%. Shaping of plastics (B29C) and layered products (B32B) also carry meaningful shares. Because records often carry multiple IPC codes, these figures overlap rather than sum to 100%, which is expected and reflects how multi-step recycling processes get classified across several subclasses at once.

Classification density is comparatively lower in containers and packaging (B65D, 7.6% of records) and addition polymers (C08F, 7.7%), versus the 20-29% density seen in the core processing and preparation classes. That gap does not guarantee an easy filing — it means fewer records currently occupy that specific claim space relative to the field's core, which is worth validating against a targeted prior-art search before committing R&D or drafting budget to a specific claim construction.

US12468268B1 claims a machine-learning based method for adjusting additives to a plastic recycling input stream to hit target physical or mechanical characteristics in the output, based on sensor input and a trained model determining an administration scheme. It represents a newer category of claim — process-control and monitoring software layered onto physical recycling operations — distinct from the materials and mechanical-separation claims that dominate the older, most-cited records in this dataset. Companies deploying sensor-driven additive dosing in a recycling line should review its claim scope specifically, rather than assuming older materials patents cover the same ground.