MIT, Amazon, TSMC, ASML and others are working on sustainable AI • The Register


Big names in tech are collaborating with academics to develop energy-optimized machine learning and quantum computing systems under the MIT AI Hardware program, an initiative announced Tuesday.

Chipmakers like TSMC and Analog Devices, hardware development lab NTT Research, EUV machine supplier ASML and tech giant Amazon have signed up so far.

The goal is to develop a roadmap outlining the production of next-generation low-power hardware for AI and quantum computing over the next decade. To this end, research will focus on the development of innovative architectures and software at the heart of a range of technologies, from analog neural networks and neuromorphic computing to hybrid cloud computing and HPC. The designs will be tested using proof-of-concept at MIT.nano, the American university’s small-scale manufacturing facility.

The MIT AI Hardware program will be co-led by Jesús del Alamo, Professor of Electrical Engineering, and Aude Oliva, Director of Strategic Industry Engagement and the MIT-IBM Watson AI Lab. It will be chaired by Anantha Chandrakasan, dean of the School of Engineering and professor of electrical engineering and computer science at the university.

“Over the past few years, we’ve seen seemingly superhuman capabilities of AI systems,” Alamo said. The register.

“Used correctly, they are poised to transform many human activities such as transportation, healthcare, education, defense, etc. As advancements in algorithms and datasets continue apace , hardware must follow or the promise of AI will not be realized. This is why it is extremely important that research takes place on AI hardware.”

It’s one thing to build ever more powerful chips and systems to run ever-growing neural networks. It’s another to do it in a way that’s energy efficient and sustainable for our planet, which is what this effort is focused on.

“More optimized hardware has been proposed, but significant new research is needed to realize these concepts,” Alamo said. “Power efficiency is the greatest need. As data sets grow, hardware must expand accordingly and power consumption explodes. It does not scale. We need new hardware .”

The MIT AI Hardware program is industry-funded, we’re told. Alamo declined to say how much its inaugural members have contributed so far.

Semiconductor supply chains are under strain during the ongoing COVID-19 pandemic. Shortages of key materials and high demand have led to a backlog of component orders that are struggling to be filled by overseas manufacturers, prompting governments around the world to invest in efforts to increase local chip production.

Intel CEO Pat Gelsinger has just urged the U.S. Congress to pass the $52 billion Creating Advantageous Semiconductor Production Incentives (CHIPS) for America Act to subsidize the expansion of factories. Meanwhile, the European Commission has proposed €11 billion (~$12.2 billion) in funding to bolster chip R&D under the EU Chip Law.

Alamo said the MIT AI hardware program focuses on next-generation hardware for emerging technologies and is less concerned with global chip crunch.

“Supply chain issues are transitory; we think long term,” he said. “University research is most effective in five to ten years and beyond.” ®

Source link


Comments are closed.