A solid-state transformer replaces the single magnetic core of a conventional transformer with a chain of power-electronic conversion stages, and the isolated DC-DC stage — typically built around a dual active bridge and a medium-frequency transformer — is where most of the switching, control and topology innovation actually happens. Patent claims tend to target this stage specifically because it's where efficiency, component count and control complexity are won or lost, separate from the magnetic core itself or the AC-side rectification. That's why a search restricted to DC-DC-specific claim language (dual active bridge, isolated DC-DC stage, high-frequency DC-DC) returns a materially different, more focused set than a general SST search.

The corpus shows a mix of Chinese grid-affiliated research institutes, universities and established industrial power-electronics firms, with citation influence still concentrated in early Chinese filings on cascaded modular topologies. In the most recent tracked year, activity skews toward research institutes and universities rather than large industrial programmes. Some previously active industrial filers show a sharp year-on-year drop to zero filings, so 'leading' by cumulative filings and 'leading' by current momentum point to different organisations.

Of the tracked receiving-office filings, 36 out of roughly 58 documented routes are in China, far ahead of the US, India, the EPO and PCT combined. This reflects both the scale of China's grid modernisation programme, which has driven heavy utility and university R&D into SST for DC distribution networks, and the presence of large state grid research institutes that file domestically as a matter of course. Anyone assessing risk or opportunity in this space needs to treat Chinese prior art as the primary body, not a secondary check.

Filing volume grew from a single record in 2017 to a peak of 14 in 2025, passing through a midpoint of 7 in 2022 — a real increase over the decade but not a steep or accelerating one, and the middle years were essentially flat. The 2026 figure will look lower than it eventually is, because patent publication typically lags the actual filing date by around 18 months. Overall, this reads as steady, moderate interest rather than a technology in a hype phase.

The core DC-DC and power-conversion classification (H02M) is dense, appearing in every record in this dataset, so a claim aimed squarely at cascaded modular topology or standard dual-active-bridge control is competing against a crowded field. Secondary IPC classes — EV propulsion overlap, digital-processing-based fault prediction, AI-based control, and insulation/measurement overlap — each carry only one or two records, which suggests those adjacencies haven't been claimed exhaustively yet. A first claim combining the DC-DC stage with one of these adjacencies, rather than restating the core topology, has more room to stand on its own.