The race for physical AI will be won in the patent office

Developments in artificial intelligence, machine learning and automation have given robots capabilities that would have seemed extraordinary ten years ago. As competition intensifies, the physical AI race is less likely to be won by those who build the best robot, and more and more likely to be won by those who have the technology that drives its behavior.

This is a particularly complex intellectual property (IP) landscape because such innovations can span hardware, software, AI models and calibration methods. The most valuable invention may not be the robot itself, but the technology that allows it to behave in a certain way.

I have worked for 20 years as a patent attorney and have filed more than 300 patents in the field of industrial automation. The pattern is consistent: the startups that lose are the ones that lag.

Robotics founders must decide what to patent, what to keep secret, and how to protect their most important technologies before their competitors do. Otherwise, they may become locked out of their own core mechanisms, locked out of the market, or purchased for parts rather than value.

In some cases, the robot is not the innovation; his behavior is

Boston Dynamics’ 2022 lawsuit against Ghost Robotics for violating the “basic technology” of how the robot recovered if it fell illustrates this idea particularly well.

Similar challenges are faced by startups working in the field of swarm robotics, where technical innovation is often an interaction between dozens of machines that produces behaviors that no single machine could achieve on its own. This raises a question that the law has not yet fully answered: can the behavior of the swarm be patented, or only the machines that produce it?

Patent offices, including the European Patent Office, have indicated that, depending on the interaction between different entities, patent claims may need to define individual entities and/or the overall system, including how the entities interact. In some circumstances, it may be necessary to limit your request to the combination of these entities.

Although Boston Dynamics and Ghost Robotics managed to resolve their dispute in 2025, the reality is that in ambiguous IP scenarios, startups rarely have the upper hand due to the large budget needed to defend patents. Therefore, in an industry that moves so quickly, you need to implement a strong IP strategy from the start – it can’t be something that comes after Serie A.

A strong IP strategy does not necessarily mean more patents

A strong patent portfolio should be the foundation of any robotics IP strategy. As RoboSense’s head of intellectual property warned last year, patent disputes in robotics are likely to multiply as they did in the smartphone industry in the 2010s, when global legal battles showed how patents could influence licensing, acquisitions and market control.

This doesn’t mean patenting every component. A robotics startup that tries to do this will run out of design time before it runs out of components to patent. And while the fortieth application is being filed, a competitor will submit the one that really mattered first.

Know that some of your best inventions should not be published

A patent is a publication. For some innovations, this is the last thing to do.

For example, for a manufacturing process or calibration technique that is difficult to reverse engineer from the finished robot, keeping it confidential may be a more cost-effective way to maintain a competitive advantage than filing for a patent.

For strategically sensitive innovations, such as in defense technology, a published patent could tell competitors – and adversaries – exactly what problem you have solved and where your advantage lies.

Conversely, a patent is likely to make more sense when an invention is incorporated into the finished product and could be reverse-engineered by a competitor. For example, a new robotic mechanism that allows a robot to move faster, more precisely, or more efficiently would be difficult to keep secret once the robot is on the market. In this case, patent protection can grant the startup a period of exclusivity on the mechanism.

Focus on one or two inventions without which the robot cannot function

When a startup gets to expanding its patent portfolio, the goal shouldn’t be to try to patent an entire machine. Startups should aim to identify one or two new mechanisms, architectures or processes that are central enough to the robot’s function without which it would be really difficult for competitors to copy the robot.

This was an approach we helped London-based food assembly robotics startup KAIKAKU use in building its strategic patent portfolio. The problem KAIKAKU solved was to never move a bowl from one place to another. He was moving it quickly without making a mess. This behavior was achieved through a central motion mechanism that was novel and central enough to the robot’s function to become one of its “crown jewel” patents.

Make it more difficult to design based on the main patent

Once crown jewel inventions are secured, startups can build a broader defensive ecosystem to protect the surrounding architectures that make the core innovation commercially useful – and that a competitor might otherwise use to design around the core patent.

The result is layered. Some patents protect the benefit itself. A wider range of patents makes designing expensive enough that most competitors won’t even try. For startups, this can create not only stronger protection against competitors, but also greater influence in licensing, investments, partnerships and potential acquisition discussions.

Leverage artificial intelligence to make quality patent protection accessible

For a startup with limited resources, even the leanest patent portfolio can represent a huge financial burden. A single European application, properly drafted and prosecuted, costs €13,000 to €18,000 ($14,636.7 to $20,266.2 US) or more once legal fees are included. Multiply that by three or four crown jewels and a defensive ring, and you get a line item that most early-stage founders can’t justify.

The challenge is to find a commercially viable intellectual property strategy without compromising patent quality and without diverting scarce resources from product development.

In this case, advances in artificial intelligence offer a clear opportunity to make quality intellectual property management and patent filing accessible to startups. Most of the work done in a patent is not judgmental. It is documentation, analysis and drafting. When AI does this groundwork and a patent attorney examines, refines, and takes responsibility for what is filed, the economics change.

Filing becomes up to 70% faster and lawyer time goes where it should: on court calls and at a fraction of the traditional cost. In a market like physical AI, that’s the difference between protecting inventions that matter and leaving them to chance.

Don’t count your physical AI patents. Weigh them

The race for physical AI will not be won by the company with the most patents. Not even patents alone will bring a robot to market. But when a company’s advantage rests on a handful of hard-to-replicate mechanisms, leaving them unprotected is an impractical choice.

The right approach is therefore strategic rather than quantitative. Identify the leading technologies that make the robot valuable, decide which should be patented and which should remain secret, and build a defensive ecosystem around them, using an AI-native patent platform to simplify cost and complexity.

For startups, the value lies not simply in the ability to file more patents, but also in the ability to spend more of their limited legal and engineering resources to identify, protect and strengthen the innovations that matter most.

About the author

Dominic Davies is co-founder and CEO of Lightbringer, one of the first AI-native patent services. Dominic is a former software engineer turned patent lawyer in the UK and Europe, who previously founded patent consultancy firm Invent Horizon IP after practicing with London’s leading IP firms. He is also a co-founder and investment manager at Immetric, an investor in intellectual property-rich startups.