Separately Berkeley, MIT and at least four other universities are creating the Center for Energy Efficient Science. It will conduct research in moving toward semiconductor circuits that can operate on a millivolt of power.
"We could be using a million times less energy to process information," said Eli Yablonovitch, a Berkeley professor who will work with the center.
In a panel discussion Yablonovitch and others called for a new generation of power engineers who can apply the techniques of the Internet to craft a smart electric grid.
"We need new thinkers to have an impact on this area," said Randy Katz, a Berkeley professor who helped launch a low-power research effort at the event in 2009.
"This is an opportunity to think about what is the right background—it's not the old Handbook of Power Engineering," Katz said. "It's an opportunity to train a new generation of people who understand both IT systems and how power moves around," he added.
David Culler, a Berkeley professor working on an initiative for energy-efficient buildings, said engineers need to understand a variety of mechanical, civil and electrical disciplines in this sector. "I really worry we are not training people for the wide range of issues coming up," Culler said.
He called for a smart grid that uses Internet-like techniques such as distributed services and separately-defined implementation layers that can evolve independently. "Just like we have virtual networks as overlays on the Net, there's no reason we can't have virtual private grids--that's how you evolve the infrastructure," Culler said.
Katz agreed, adding that new regulations including a carbon tax are needed to motivate utilities and power users. "In order to have the innovation take place the true cost of energy has to be reflected, it's the only way to get people to invest," he said.
Finally, Berkeley professor Michael Franklin formally announced the AMP Lab, a new research center seeking to drive cloud computing to the next level. The center aims to address what Franklin called the scalability problem involving algorithms, machine learning and people.
Machine learning algorithms and data analytics don't scale to increasingly large and complex data sets. Meanwhile cloud services lack crowd-sourcing tools to harness large groups of people over the Internet to tackle shared problems.
The lab is a spin-out of a Berkeley center developing software that will help individuals use cloud computing to launch new Web services. The new lab wants to enable many people to collaborate to collect, generate, clean and make sense of large data sets, he said.