When Chris Shelton, chief technology officer for electricity giant AES, takes the stage at a tech conference in San Francisco this week, he’ll be the rare name in energy. The rest of the lineup looks more like traditional Silicon Valley tech: Google, Nvidia, Facebook and others.
Shelton is headed to San Francisco for the AI Summit to talk about what he described to GTM as “one of the most exciting things out there” -- which is how the energy sector will be using artificial intelligence. Though, Shelton contends, it’s still early days when it comes to how the sector will embrace AI in practice.
According to Shelton, AES is exploring how AI can improve awareness, efficiency and maintenance of the company’s grid systems and assets like solar farms and gas plants. It can also help AES make better predictions of how those systems and assets will operate.
AES manages 36 gigawatts of energy capacity across 17 countries and employs 19,000 people. The company, which generated $14 billion in revenue last year, is also building another nearly 5 gigawatts of energy capacity and has been growing its energy storage division focused on selling battery systems.
That such a large player is moving swiftly to embrace AI shows just how big of an opportunity the technology could be for the energy industry. Tech giants like GE and IBM are building prediction and maintenance systems for electricity, while many startups have emerged to tackle more niche energy issues like lowering the cost of selling solar panels or making office buildings more comfortable and efficient.
While Shelton is quiet on many of AES’ AI details -- due to the early nature of the initiatives, he said -- he describes some of the company’s interest in AI systems around advanced neural network design, natural language processing and machine intelligence. Artificial intelligence is not a single technology, but a suite of data science tools.
“The technology is fascinating because we can overlay it on our existing assets,” said Shelton. Assets, like a solar or battery farm, produce a lot of data, he explained, and AI tools can take all that data and use it to run the farm more efficiently at a lower cost, or to produce more power.
AES is both partnering with -- and looking to partner with -- AI vendors like GE, which it has worked with in the past. At the same time AES is also building its own AI technology, similar to its investment in systems for lithium-ion batteries, said Shelton, who helped spearhead AES’ energy storage business. “We’re ready to be patient with the technology, like we [were] with lithium-ion,” he said.
With the increase of digital technology used at all points on the electricity grid, artificial intelligence is expected to be increasingly used to manage, optimize and maintain the grid and power plants. Analysts at Navigant predict that power companies will turn to AI for increased reliability, safety, cyber security, efficiency and better customer experiences.
Companies like Alphabet's Nest have been using AI to offer better energy experiences for some time. Nest’s thermostat lowers the temperature of a home and reduces energy consumption after using AI to learn its occupants' habits. Startup Comfy delivers a similar service for office buildings, but by using a combination of AI and an app for office workers.
Developing AI tools for energy companies could be a major opportunity for startups and traditional tech vendors alike. Expect both new entrants to emerge with brand-new, focused applications, and for AI players like Google, GE and IBM to use data tools from the web and consumer internet for energy services.
Shelton said AI really started to catch AES’ attention about a year and a half ago. But he expects the technology to be able to be deployed pretty swiftly, particularly compared to how long it takes power companies to install large physical assets on the power grid.
Today, though, AI for energy is just getting started. “It’s similar to the dawn of the internet” in some ways, he said, noting that it could fundamentally change the way energy systems are built and operated.
