The GenAI Patent Surge Is Coming for Physical AI: What Autonomous Systems Companies Should Be Watching
More generative AI patent families were published in 2024 and 2025 than during the entire preceding decade, according to a recent analysis by the World Intellectual Property Organization (WIPO) and analytics firm, EconSight.1 The numbers are striking: GenAI patent families surged from roughly 14,000 in 2023 to over 37,800 by 2025, nearly tripling in just two years.2 But what may be most significant for companies building autonomous systems, robots, drones, and other physical AI solutions is not the volume itself. It is who is filing and what those filings are targeting.
As reported by Bloomberg Law, some of the fastest-growing patent activity is coming from companies better known for energy, infrastructure, manufacturing, and telecommunications than for developing the AI models that first captivated investors.3 Among the top GenAI patent holders through 2025 are SoftBank, State Grid Corp. of China, Bosch, and Inspur Group. Notably, many of these companies are not claiming the underlying AI models themselves; they are protecting specific applications within existing businesses. Bosch, for example, is staking out its position as a leader in AI automation tools for physical manufacturing, and State Grid’s filings are tied to grid optimization, predictive maintenance, and infrastructure planning.
This pattern should resonate with anyone tracking the autonomous systems patent landscape. The same dynamic is unfolding in physical AI.
The Physical AI Connection
The term “physical AI” has emerged as the unifying concept for AI systems that perceive, reason, and act in real-world environments, enabling machines to anticipate behavior rather than simply react to it. At CES 2026, NVIDIA CEO Jensen Huang described the moment as “the ChatGPT moment for physical AI, when machines begin to understand, reason and act in the real world.” At GTC 2026, NVIDIA announced major expansions to its physical AI platform spanning autonomous vehicles, industrial robots (through partnerships with FANUC and ABB), and humanoid robotics foundation models.
The underlying technical challenges across physical AI are structurally identical to those mapped by the European Patent Office’s original autonomous vehicle taxonomy.4 Perception and sensing, communication, data processing and AI, and vehicle control each have direct equivalents in robotics, drones, surgical systems, and defense platforms. The taxonomy that began with self-driving vehicles now applies to warehouse robots navigating dynamic environments, surgical systems performing precision manipulation, agricultural drones adapting to weather and terrain, and humanoid robots interacting with humans in unstructured settings.
Three Implications for Autonomous Systems Companies
The GenAI patent surge, combined with the broadening of patent activity beyond traditional AI platform developers, raises several strategic considerations for companies in the autonomous systems space:
- Incumbent GenAI patent holders may assert broadly. Companies entering adjacent physical AI markets (robotics, drones, agtech, defense) should anticipate that incumbent GenAI and AV patent holders may assert their portfolios broadly, particularly where claim language is drafted at a level of abstraction that captures perception, planning, and control innovations regardless of the specific physical platform. Notably, sophisticated companies are treating this as a period to build and to position, in anticipation of enforcement activity that most expect to intensify. Four areas appear most likely to generate future enforcement activity:
- the technical infrastructure enabling AI systems to run efficiently;
- applied sectors including financial services and healthcare;
- agentic and multimodal systems as they mature; and
- non-practicing entities.
- Foreign filings create prior art risks. The volume of GenAI patent filings from Chinese state-owned enterprises and other international filers may carry consequences even if those patents are never enforced in U.S. courts. Once a foreign patent application is published, it becomes prior art, potentially blocking U.S. companies from later patenting the same idea. For autonomous systems companies, this means that patent landscape monitoring is not optional. WIPO’s data confirms that seven out of the top ten inventor locations published more GenAI patent families in the last two years than in the entire preceding decade.2 Companies building physical AI solutions should conduct freedom-to-operate analyses before entering new markets or launching new products.
- The patent-versus-trade-secret calculus is shifting. Much of the most strategically significant AI innovation, including model weights, training data, and proprietary methods difficult to detect in a competitor’s product, is increasingly held as trade secrets and may never appear in patent counts. For autonomous systems companies, the question of what to patent versus what to protect as a trade secret requires a systematic assessment of the technology stack. Innovations that are detectable in the product’s behavior or outputs, or that are likely to be independently discovered by competitors, are generally better served by patent protection; training data sets, proprietary labeling taxonomies, and internal model architectures that are not discernible from the shipped product may be more effectively protected through trade secrecy.
Generalize The Playbook
The expansion of GenAI patent activity beyond traditional AI platform developers mirrors a trend that has been building for several years in the autonomous systems space. The playbook that AV startups developed for patent strategy between 2015 and 2022 must now be generalized. What to patent, where to file, and how to align IP activity with product development and funding milestones are questions that every autonomous systems company must answer deliberately and early.
For autonomous systems companies, patents are not merely legal instruments; they are strategic business assets that, when managed deliberately, support every phase of the company’s lifecycle from formation through exit. The GenAI patent surge underscores that point. Companies that are not traditionally associated with AI are now aggressively staking out IP territory in applied AI domains, and the line between GenAI and physical AI is becoming increasingly difficult to draw.
As WIPO has cautioned, a large portfolio is not necessarily the same as a strategically potent one, and the data do not allow us to readily distinguish between the two.2 That is precisely why the first strategic question for autonomous systems companies is not “should we file patents?” It is “do we understand which layers of the autonomy stack contain our defensible innovation, and are we protecting them deliberately?”
The GenAI patent wave is being seeded now. For physical AI companies, the time to position is before that wave arrives.
DISCLAIMER: The information provided herein may not be applicable in all situations and should not be acted upon without specific legal advice based on particular situations.
__________________________________________________________-
1 World Intellectual Prop. Org. (WIPO), Patent Landscape Report: Generative Artificial Intelligence (2024), https://www.wipo.int/web-publications/patent-landscape-report-generative-artificial-intelligence-genai/en/index.html.
2 WIPO, SPARK: Patent Trends Update in GenAI (July 2026), https://www.wipo.int/web-publications/spark-patent-trends-update-in-genai/en/index.html.
3 Christopher Yasiejko, GenAI Patent Surge Extends Well Beyond Familiar Platform Firms, Bloomberg Law (Aug. 10, 2026), .
4 European Patent Office, Patents and Self-Driving Vehicles: The Inventions Behind Automated Driving (Nov. 2018), https://www.epo.org/en/news-events/in-focus/ict/self-driving-vehicles.
