A Panacea for AI Chip Bottlenecks? White House Puts Up $285M for Digital Twins R&D

The Biden White House on Monday announced a new research and development fund with the goal of making chip manufacturing more efficient and less resource-intensive using "digital twins" technology.

The $285 million fund, part of the $50 billion earmarked for the CHIPS and Science Act of 2022, is part of the U.S. government's effort to support homegrown production of semiconductors -- which, amid today's AI boom, are in especially high demand. One of the goals of the CHIPS Act is to find novel ways to overcome the immense resource, talent and technology challenges of semiconductor manufacturing, as well as to create a framework for more U.S.-based semiconductor production.

To do that, the government is betting on digital twins, a technology that lets users create virtual, software-based replicas of physical objects. Digital twins are useful for simulation, planning and training uses, particularly in industrial settings. For instance, Pure AI blogger Noelle Russell recently described the use of digital twins in construction:

Virtual twins can ingest a vast amount of data and use it to replicate processes that predict potential performance outcomes and issues that might not be obvious in the real-world. They can simulate the performance of various sustainable materials and construction techniques. They can make it possible to analyze data about bio-based materials, passive solar designs, and plans for natural light, and then incorporate that data into building plans.

Notably, Russell added, generative AI "has become a critical enabler of this technology," providing the modeling, prediction and visualization capabilities that make digital twins truly useful.

When it comes to semiconductors, digital twins can enable manufacturers to design, optimize and troubleshoot new chips before wasting a single physical resource, significantly reducing overhead and time to market.

"Unlike traditional, physical research models, digital twins can exist in the cloud, which enables collaborative design and process development by engineers and researchers across the country, creating new opportunities for participation, speeding innovation, and reducing costs of research and development," the U.S. Department of Commerce wrote in a press release Monday. "Digital twin-based research can also leverage emerging technology like artificial intelligence to help accelerate the design of new U.S. chip development and manufacturing concepts and significantly reduce costs by improving capacity planning, production optimization, facility upgrades, and real-time process adjustments."

The CHIPS Manufacturing USA institute, which operates the fund, is now soliciting proposals from nonprofits, government agencies, educational institutions and private firms that can help develop solutions for, and provide research around, the use of digital twins to help facilitate U.S.-based semiconductor manufacturing.

According to the Notice of Funding Opportunity (NOFO), applicants should consider these six goals in their proposals:

  1. Convene stakeholders across the semiconductor manufacturing, advanced packaging, assembly, and test industry to address shared challenges relevant to digital twins, in a collaborative environment.
  2. Improve the state of the art in manufacturing-relevant digital twins, for both unit-level digital twins and the combination of multiple digital twins.
  3. Significantly reduce U.S. chip development and manufacturing costs by improving capacity planning, production optimization, facility upgrades, and real-time process adjustments using digital twins.
  4. Improve development cycle times of semiconductor manufacturing, advanced packaging, assembly, and test and accelerate the development and adoption of relevant innovative technologies, including breakthrough tools, materials, and manufacturing processes.
  5. Advance digital twin-enabled curricula, best practices, and hands-on opportunities for training the next generation of the domestic semiconductor workforce.
  6. Create a digital twin marketplace for industry to access digital models and manufacturing process flows and to de-risk digital twin development and implementation.

"Digital twin technology will help transform the semiconductor industry," said Laurie E. Locascio, director of the National Institute of Standards and Technology (NIST), in a prepared statement. "This historic investment in the CHIPS Manufacturing USA institute will help unite the semiconductor industry to unlock the enormous potential of digital twin technology for breakthrough discoveries."

Concept papers are due from applicants on June 20, and full applications on Sept. 9. Informational events will be held on May 8 and May 16 for interested applicants; registration information is available here.

About the Author

Gladys Rama (@GladysRama3) is the editorial director of Converge360.