Ai Hardware-Software Combo Patent Eligibility.
1. Conceptual Background
AI systems rarely function as pure software. In real-world applications—autonomous vehicles, medical imaging, drones, robotics, smart sensors—AI algorithms are tightly integrated with hardware components such as processors, sensors, GPUs, FPGAs, or embedded systems.
Patent law generally treats:
Abstract algorithms / software alone → Not patentable
Hardware alone → Usually patentable
Hardware + software producing a technical effect → Potentially patentable
The core legal question is:
Does the AI software, when combined with specific hardware, improve the functioning of the machine itself or produce a concrete technical effect?
2. Key Patentability Standards Applied to AI Hardware–Software Combos
Across jurisdictions, courts assess:
Technical contribution
Improvement in computer or machine functioning
Non-abstract application
Novel and non-obvious interaction between hardware and software
3. Detailed Case Laws
Case 1: Diamond v. Diehr (1981, US Supreme Court)
Facts
The invention used a computer-implemented algorithm to calculate the optimal time for curing rubber. The algorithm controlled industrial molding hardware, continuously monitoring temperature and adjusting the curing process.
Legal Issue
Is a mathematical algorithm controlling physical machinery patentable?
Court’s Reasoning
The invention was not just an algorithm
The algorithm was embedded in a physical industrial process
The system transformed raw rubber into a cured product
Holding
The invention was patentable.
Relevance to AI Hardware–Software Combos
This case laid the foundation for AI patents:
AI algorithms controlling robots, machines, or sensors are patent-eligible
Emphasized process + hardware integration
Principle Established
Software controlling hardware to produce a physical transformation is patentable.
Case 2: Alice Corp. v. CLS Bank (2014, US Supreme Court)
Facts
Alice claimed a computer-implemented financial transaction system using generic computer hardware.
Legal Issue
Can abstract ideas implemented on generic hardware be patented?
Court’s Reasoning
The invention was merely an abstract idea
Hardware was generic and added no technical innovation
No improvement to computer functioning
Holding
The patent claims were invalid.
Relevance to AI Systems
This case created a barrier for AI patents:
AI software on generic hardware alone is not enough
Must show specific hardware interaction or improvement
Principle Established
Adding a computer to an abstract idea does not make it patentable.
Case 3: Enfish, LLC v. Microsoft Corp. (2016, US Federal Circuit)
Facts
The invention involved a self-referential database architecture, implemented in software but designed to improve computer performance.
Legal Issue
Is software that improves computer functionality patent-eligible?
Court’s Reasoning
The invention was not abstract
It improved memory usage, processing speed, and data retrieval
Focused on how the computer works, not what it does
Holding
The claims were patent-eligible.
Relevance to AI Hardware–Software Combos
AI models that:
Optimize processor usage
Improve GPU efficiency
Reduce memory consumption
…can be patentable when tied to hardware performance.
Principle Established
Software improving computer functionality itself can be patentable.
Case 4: Thales Visionix Inc. v. United States (2017, US Federal Circuit)
Facts
Thales patented an inertial tracking system using sensors and software algorithms to track object motion more accurately.
Legal Issue
Is sensor-based data processing using algorithms patentable?
Court’s Reasoning
Claims were tied to specific physical sensors
The system improved tracking accuracy
Not a mathematical formula in isolation
Holding
The invention was patent-eligible.
Relevance to AI
Applies directly to:
AI sensor fusion systems
Autonomous vehicles
Drones and robotics
Principle Established
Algorithms integrated with physical sensors producing technical improvements are patentable.
Case 5: McRO, Inc. v. Bandai Namco Games (2016, US Federal Circuit)
Facts
The invention automated facial animation using rules rather than human animators.
Legal Issue
Is rule-based automation abstract?
Court’s Reasoning
The invention used specific technical rules
It improved animation technology
It did not merely replicate human activity
Holding
The invention was patent-eligible.
Relevance to AI Hardware–Software Combos
AI systems replacing manual processes through technical rules are patentable when:
Claims specify how automation occurs
Not just the result
Principle Established
Automation through specific technical rules is not abstract.
Case 6: Siemens v. Comptroller General of Patents (UK Supreme Court Approach)
Facts
Siemens developed a system using industrial sensors + software analytics to monitor machine failures.
Legal Issue
Does software controlling industrial machinery qualify as patentable subject matter?
Court’s Reasoning
The invention produced a technical effect outside the computer
Improved industrial process reliability
Holding
The invention was patentable.
Relevance to AI
Supports patentability of:
Predictive maintenance AI
Smart factory AI systems
Principle Established
Technical effect beyond software execution supports patentability.
Case 7: G 1/19 (EPO Enlarged Board of Appeal – Simulation Case)
Facts
A simulation-based invention lacked a direct physical output.
Legal Issue
Can simulations and AI models be patentable?
Court’s Reasoning
Simulations alone may be abstract
But simulations linked to hardware implementation are patentable
Holding
Patentability depends on technical purpose and hardware linkage.
Relevance to AI
AI models become patentable when:
Deployed on hardware
Used to control or optimize physical systems
Principle Established
AI simulations must contribute to a technical solution.
4. Combined Legal Principles for AI Hardware–Software Patent Eligibility
An AI hardware-software combination is patent-eligible if:
The AI algorithm is not claimed in isolation
The hardware is specifically defined, not generic
The system improves machine performance
The invention produces a technical effect
The claims focus on how the system works, not just results
5. Practical Example of a Patent-Eligible AI Combo
✔ AI model optimized for GPU architecture to reduce inference latency
✔ Neural network controlling robotic arm sensors
✔ AI-driven medical imaging device improving signal accuracy
✔ Edge-AI chips with specialized learning algorithms
✘ Generic AI software running on any computer
✘ Pure data analysis without hardware integration
6. Conclusion
Courts worldwide increasingly recognize that AI innovations lie at the intersection of software intelligence and hardware execution. While abstract AI algorithms are excluded, AI systems that redefine how machines operate are strongly protected under patent law.
RELATED Blog
comments