Making Energy Programmable

We’ve previously written about the storage gap. The proliferation of solar has led to an abundance of energy when the sun is out and a dearth when it’s not. Batteries help address this gap by storing excess energy in times of plenty and discharging stored energy as needed. The push to build out storage capacity is just one example of how the edges of the grid are manipulating energy in increasingly complex ways. 

Originally, the grid was designed to have energy flow unidirectionally: from centralized power plants to end-customers, similar to how centralized mainframe computers were connected to “dumb” terminals . And just as the PC revolution pushed greater computing capabilities to the edge, energy systems are evolving into a more multi-directional network, with energy now moving between the grid, behind-the-meter generation and storage, and end-consumption. Though technological and societal pressures are pushing this transformation through, it’s been occurring on legacy infrastructure ill-equipped for the complex demands of a more decentralized energy future. To effectively enable the grid’s further decentralization, we will have to deploy more capable and flexible – which is to say programmable – energy infrastructure at the edge. 

New software is one key component of new energy systems at the edge. Over the last hundred years, we’ve built out massive physical energy networks, but we’ve been slow to add a software layer on top. Our portfolio companies Leap, David Energy, and Ostrom are leading the charge in building those software networks to coordinate how and when devices send, receive, and use electrons. Just as social networks made internet users readers and writers, these new energy networks empower customers to engage more actively in energy markets as more than just passive consumers. 

However, software is only as good as the hardware it runs on. The machine learning boom of the last decade wouldn’t be possible without innovative computer hardware like Nvidia’s GPUs. Unfortunately, the energy world is still running on the equivalent of vacuum tubes. Transformers, the functional interface between the grid and the edge, are bulky boxes that step voltage down by simply running it through two copper coils. They are analogue systems that inflexibly transform voltage at a fixed ratio. Circuit panels similarly sit at a powerful juncture as the control center for electricity in the home. Current panels bluntly flip loads on and off, but a future smart panel might offer the ability to modulate the amount of power a given load has access to. Companies like Span are innovating by adding digital controls to the circuit panel, but the value of better controls is limited by how capable the underlying hardware they control is.

One way to make these systems programmable is to incorporate power electronics technology. In power electronics, devices process energy with specialized transistors (e.g. SiC) like a computer uses transistors to process information. Instead of using digital controls to intelligently switch between many different pieces of narrow hardware, you could program a more general-purpose circuit to adapt to all kinds of different situations. In a world of dissimilar sources and loads, unpredictable energy flows, and complex end-uses, such device programmability will be invaluable. Software networks will tell devices what to do, and smarter hardware will physically be able to do more, making the software all the more valuable in turn.

Nvidia stock has nearly tripled in value since ChatGPT’s release a year ago, largely as a result of its critical role in the AI revolution we’re living through. We believe there’s now a similar opportunity to design the programmable hardware building blocks of the electrification revolution.