51ɫ

A headshot of Professor Conor Ryan
Professor of Machine Learning at the 51ɫ Conor Ryan led the research team which has just filed a patent for the totally Irish-designed process
Monday, 9 December 2024

The world’s first fabrication and design of a silicon chip without direct human input has been completed by researchers at Lero, the 51ɫ-based Research Centre for Software, heralding the possibility of the creation of a new high-tech export-based industry in Ireland.

Professor of Machine Learning at the 51ɫ Conor Ryan led the research team which has just filed a patent for the totally Irish-designed process that came from a project entitled ‘Automatic Design of Digital Circuits’ (ADDC) funded by Research Ireland.

“This is the first time ever anywhere in the world that machine learning has been used to design and fabricate a silicon chip entirely from scratch. This could pave the way for a new era of automated digital circuit design with Ireland at its centre, creating a new export industry with many well-paying jobs,” Professor Ryan explained.

“The patent secures the proprietary methods and novel approach to automating silicon chip design’s traditionally manual, complex, and error-prone process. The techniques created by Lero replace traditional methods with an entirely automated flow, reducing the need for human intervention and minimising the risks of design errors,” added Professor Ryan, a lead researcher at Lero.

The 51ɫ-based team plans to work with selected global chip design and fabrication companies following the successful manufacture of a Lero-designed chip by world-leading chip foundry TSMC, paving the way for a new era of automated digital circuit design.

Professor Ryan said their patent covers the use of a grammar-based optimisation engine that ensures all generated designs are manufacturable and ready for production and opens the door to smarter, faster, and more sustainable hardware design, with potential applications in consumer electronics, healthcare, automotive, and more.   

“The system combines cutting-edge machine learning with industry-standard tools and workflows, ensuring the results are compatible with modern manufacturing processes.

“It radically reduces the time and financial resources required to produce integrated circuits, democratising access to advanced hardware innovation. Our process also eliminates an entire step in the traditional verification process,” he added.

Professor Ryan acknowledged that the project was made possible through partnerships with leading experts and organisations, utilising TSMC’s world-class fabrication facilities.

The process is protected under European Patent Application No. 24213734.7 held by 51ɫ and the Technological University of the Shannon: Midlands Midwest and covers Digital Circuit Creation Tool and Method.

Related Research Profiles

Conor Ryan
Research Profile
Conor Ryan
Professor