Demand Response AI Patents: Who Leads, Where the Gaps Are 2026
- Filing has flattened, not grown. activity rose from 53 families in 2017 to a peak of 98 in 2025, but the 2022 midpoint of 54 shows the middle years added little — this is a field consolidating claim positions, not one still accelerating.
- Grid control and business-process claims dominate. H02J (power supply & grid systems, 456 records) and G06Q (business/commerce data processing, 377 records) together outweigh the AI-specific G06N class (179 records), meaning most filings wrap machine learning inside grid or market-operations claims rather than claiming the model itself.
- The US receives more than double the next office. 358 records filed at the USPTO against 143 in India and 94 at the EPO — a filer building freedom to operate outside the US should not assume US claim scope maps cleanly onto those other jurisdictions.
Filing growth compares 2021 (30 records) with 2024 (47) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 834 records in scope (CR5), not by the ranked leaders only.
What this patent set covers
This landscape covers patent families that combine demand-side management or load flexibility concepts with load forecasting, reinforcement learning, or AI-based flexibility optimization. The search spans records published between 2015 and mid-2026, drawing on 834 patent families tracked in the assignee ranking. Because publication lags filing by roughly eighteen months, the 2026 count of 38 understates actual filing activity for that year and should not be read as a real decline on its own.
The technology composition points to a field where AI methods are typically embedded inside grid-control or market-participation claims rather than filed as standalone algorithms. That has practical consequences for claim drafting and for anticipating where an examiner will search prior art — a load-forecasting claim is as likely to be classified under power systems (H02J) or commerce data processing (G06Q) as under the AI computing class (G06N) itself.
Let an AI agent run this analysis on your own technology
Pick a task. Every answer cites the patents behind it.
Filing trend and technology composition
Two views of the same 834-family dataset: how filing volume has moved year over year, and which technical subclasses carry the claim density.
A plateau after 2022, not a straight climb
Filings moved from 53 in 2017 to a peak of 98 in 2025, but the 2022 midpoint of 54 sits almost where the series started — the growth is concentrated in the last few years rather than spread evenly across the period, and the partial 2026 figure of 38 should be read as an undercount given publication lag.
Grid and business-process classes outweigh pure AI claims
H02J (456 records) and G06Q (377 records) lead the composition, ahead of G06N (179), G06F (160), G05B (192), H04L (88), G01R (76) and G05F (54) — evidence that most inventive activity here is filed as grid operation or market-participation methods that happen to use AI, rather than as AI model claims in their own right.
Shares are the percentage of the 834 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Demand Response System AI and Machine Learning with Eureka
This page is one run against one query. Ask Eureka your own question about demand response system ai and machine learning and every answer comes back with the patent numbers behind it.
Try EurekaCited foundations and a representative claim
System and method for phase balancing in a power distribution system
A phase balancing system includes a load forecasting module, a phase unbalance identification module and a demand response module. The load forecasting module determines a load forecast for the distribution system for the period of interest and the phase unbalance identification module determines voltage unbalance on the distribution system for the period of interest. The demand response module estimates an available demand response on the distribution system for the period of interest and allocates an optimized demand response from the available demand response to minimize the voltage unbalance on the distribution system for the period of interest.Filed by General Electric Company, published as US20120074779A1 — a 2012 filing that ties load forecasting directly to a demand response allocation step, making it a natural prior-art anchor for later AI-driven flexibility claims.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20170006135A1 | Systems, methods, and devices for an enterprise internet-of-things application development platform | 1,900 |
| 2 | US20100332373A1 | System and method for participation in energy-related markets | 1,021 |
| 3 | US20080177678A1 | Method of communicating between a utility and its customer locations | 897 |
| 4 | US20150094968A1 | Comfort-driven optimization of electric grid utilization | 493 |
| 5 | US20110106328A1 | Energy optimization system | 363 |
| 6 | US20180191867A1 | Systems, methods, and devices for an enterprise ai and internet-of-things platform | 350 |
| 7 | US20160305678A1 | Predictive building control system and method for optimizing energy use and thermal comfort for a building or… | 333 |
| 8 | US8359124B2 | Energy optimization system | 313 |
| 9 | US20150316907A1 | Building management system for forecasting time series values of building variables | 306 |
| 10 | US20090187284A1 | System and Method for Providing Power Distribution System Information | 277 |
Citation counts favour older, longer-indexed records — treat this table as a map of influence on later filings, not as a ranking of current commercial relevance. The top-cited record, an enterprise IoT application platform, draws citations from far outside demand response specifically.
Each row carries its publication number; clicking a row searches Eureka by that number.
Put your own technology through the same analysis
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →What the filing pattern tells a strategist
Three patterns worth acting on: where claim density sits, how citation influence is distributed, and what the co-filing data says about how invention teams are structured.
Grid-system claims still carry the most weight
H02J is the largest single subclass by a wide margin, meaning any new load-forecasting or flexibility-optimization filing is likely to be examined against a dense body of grid-control prior art before it is compared against AI-specific literature.
Influence is concentrated in a small set of old filings
The most-cited record in this set, an enterprise IoT platform filing, draws far more citations than any demand-response-specific record, a reminder that citation counts here reflect broad indexing reach rather than relevance to demand response strategy specifically.
Inventor collaboration is thin and clustered
Only ten co-assignee pairs appear across the whole dataset, and the strongest link is a single named-inventor pair appearing together 14 times — collaboration in this field looks like small fixed teams filing repeatedly, not broad cross-assignee joint ventures.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to demand response system ai and machine learning, with the prior art for and against each one.
| Assignee | Co-assignee | Shared families |
|---|---|---|
| SUN DAVID | CHEUNG KWOK | 14 |
| SUN DAVID | XIAO YING | 8 |
| SUN DAVID | WANG XING | 8 |
| SUN DAVID | CHIU BUT CHUNG | 8 |
| CHEUNG KWOK | XIAO YING | 8 |
| CHEUNG KWOK | WANG XING | 8 |
| CHEUNG KWOK | CHIU BUT CHUNG | 8 |
| OPUS ONE SOLUTIONS ENERGY CORP | OPUS ONE SOLUTIONS USA CORP | 5 |
With only ten recorded co-assignee pairs across 834 families, most filers in this space are protecting positions independently rather than through joint filings.
Who is filing, and where momentum has gone quiet
Recent-year momentum data shows a field where even the more active recent filers have pulled back sharply, and several long-standing names show no filings at all in the latest tracked year.
The most active recent filer has cut back sharply
Strong Force EE Portfolio 2022 LLC recorded just one filing in the latest year, down 93% year-on-year — even among assignees still filing, the pace has dropped off steeply rather than held steady.
Established grid incumbents show no latest-year activity
General Electric shows zero filings in the latest tracked year in this dataset, consistent with a broader pattern where legacy grid-technology holders are not the ones driving current filing volume.
Some recent entrants have stopped filing entirely
An energy input/output smart-panel assignee shows a full year-on-year drop to zero filings, illustrating how quickly a newer entrant's visible patent activity can go quiet in this space.
| Assignee | Recent year | YoY |
|---|---|---|
| STRONG FORCE EE PORTFOLIO 2022 LLC | 1 | -93% |
| General Electric Company | 0 | — |
| Energy I/O Smart Panel Co., Ltd. | 0 | -100% |
| OPUS ONE SOLUTIONS ENERGY CORP | 0 | — |
| C3 AI, Inc. | 0 | — |
| C3 AI INC | 0 | — |
| SUN DAVID | 0 | — |
| CHEUNG KWOK | 0 | — |
Where to take this analysis
The dataset points to specific next steps depending on whether the goal is freedom-to-operate, competitive tracking or claim drafting.
Map claim scope against the H02J and G06Q clusters
Before drafting a new load-forecasting or flexibility-optimization claim, run it against the two largest subclasses in this set rather than assuming the AI computing class is the primary prior-art pool.
Explore claim scope in EurekaWatch momentum, not just cumulative counts
Several assignees with meaningful historical filings show zero or sharply reduced activity in the latest year — treat that as a signal to re-check before assuming a competitor is still actively building this position.
Track assignee momentum in EurekaProbe the under-claimed branches directly
Reward-design methods for reinforcement-learning dispatch and federated load-forecasting approaches show thinner direct coverage than the core grid-control clusters — worth a targeted search before committing drafting resources elsewhere.
Run a white-space search in EurekaCommon questions about demand response AI patents
This dataset tracks 834 patent families published between 2015 and mid-2026 that combine demand-side management or load flexibility concepts with load forecasting, reinforcement learning, or AI-based flexibility optimization. Filing activity moved from 53 families in 2017 to a peak of 98 in 2025, with a flatter middle period around 2022. Because publication lags actual filing by roughly eighteen months, the most recent year's count is understated and should not be read as a hard decline.
The dataset shows a small number of assignees with sustained filing activity alongside a long tail of single-filing entrants, a pattern typical of an emerging cross-disciplinary field. Recent-year momentum data is more informative than cumulative counts here: several historically active filers, including established grid-technology holders, show zero filings in the latest tracked year, while even the most active recent filer recorded a sharp year-on-year drop. Checking momentum rather than lifetime totals gives a better read on who is currently active.
The bulk of filings sit in H02J (power supply and grid systems, 456 records) and G06Q (business, commerce and administrative data processing, 377 records), followed by G05B (control and regulating systems, 192), G06N (AI computing, 179) and G06F (electric digital data processing, 160). This means load-forecasting and demand-response AI methods are more often classified under grid-control or business-process categories than under the dedicated AI computing class, which matters when scoping a freedom-to-operate search.
Reinforcement-learning reward design for flexibility dispatch, federated learning across distributed load-forecasting nodes, and explainable-AI justification layers for automated demand bids all sit adjacent to the dense H02J and G06Q clusters but show comparatively thin direct claim coverage in this dataset. These branches are worth a targeted prior-art check before assuming the space is occupied, since the core grid-control and market-participation claims are far more crowded. A first claim in these areas should tie the AI mechanism explicitly to a measurable grid or market outcome to avoid overlap with the dense clusters.
Not necessarily. The most-cited record in this dataset is an enterprise IoT application platform filing cited 1,900 times, far ahead of any demand-response-specific record, which reflects broad indexing reach across many unrelated fields rather than relevance to current demand response strategy. Citation counts inside any searched patent corpus favour older, longer-indexed filings, so they are better read as a signal of historical influence than of present-day commercial weight. A newer, less-cited filing can still be the more strategically relevant one to watch.
Research Demand Response System AI and Machine Learning in depth with Eureka
Go past this page: query the whole demand response system ai and machine learning corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company’s registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.