Artificial Intelligence (AI) has emerged as one of the biggest secular megatrends of our time. AI is powering the fourth industrial revolution and is increasingly being viewed as a key strategy for mastering some of the greatest challenges of our time including climate change and pollution. Energy companies are increasingly deploying AI tools to digitize records, analyze vast troves of data and geological maps, and potentially identify problems such as excessive equipment use or pipeline corrosion.
One such company is Dutch energy giant Shell Plc (NYSE:SHEL). In May, Shell announced plans to use AI-based technology from big-data analytics firm SparkCognition in its deep sea exploration and production in a bid to improve operational efficiency and speed as well as boost production.
According to Bruce Porter, chief science officer for Texas-based SparkCognition, Generative AI for seismic imaging has broad and far-reaching implications, adding that the technology can dramatically cut exploration timelines from nine months to less than nine days. Back in 2018, the company partnered with Microsoft to incorporate the Azure C3 Internet of Things platform in its offshore operations. The platform uses AI to drive efficiencies across the company’s offshore infrastructure, from drilling and extraction to employee empowerment and safety.
Shell is, however, not the only Big Oil company employing AI in its operations. Back in 2019, BP Plc (NYSE:BP) invested in Houston-based technology start-up Belmont Technology which helped the company develop a cloud-based geoscience platform nicknamed “Sandy.” Sandy allows BP to interpret geology, geophysics and reservoir project information thus creating unique “knowledge-graphs” including robust images of BP’s subsurface assets. BP is then able to perform simulations and interpret results using the program’s neural networks.
Meanwhile, utilities are employing artificial intelligence to make the electric grid more reliable and efficiently. Typical use cases in the industry include load and weather forecasting, grid management, predictive maintenance, enhancing the output of wind and solar resources, wildfire risk assessment, faster storm recovery and methane leak detection. Three years ago, Google teamed up with IBM to search for a solution to the highly intermittent nature of wind power. Using IBM’s DeepMind AI platform, Google deployed ML algorithms to 700 megawatts of wind power capacity in the central United States–enough to power a medium-sized city.
IBM says that by using a neural network trained on widely available weather forecasts and historical turbine data, DeepMind is now able to predict wind power output 36 hours ahead of actual generation. Consequently, this has boosted the value of Google’s wind energy by roughly 20 percent.
However, like all technologies known to man, AI has a dark side to it: high energy consumption.
According to Sreedhar Sistu, vice president of artificial intelligence for Schneider Electric, excluding China, AI represents 4.3 GW of global power demand today, a figure that could grow almost five-fold by 2028. Running AI tasks typically requires more powerful hardware than traditional computing tasks. According to a study by Alex De Vries, PhD candidate at the VU Amsterdam School of Business and Economics, AI consumes 85-134 terawatt-hours (TWh) of electricity each year, or about as much as the energy consumption of the Netherlands. Indeed, AI is nearly as power-hungry as mining bitcoin: a new research by Digiconomist has warned that ‘‘if not managed properly, AI could be responsible for as much electricity consumption as Bitcoin is today in just a few years’ time.’’
AI-servers are power-hungry devices. According to Digiconomist, a single NVIDIA DGX A100 server consumes as much electricity as several U.S. households combined. This implies that powering hundreds of thousands or even millions of these devices starts to add up quickly and could start to strain the very power grids they are supposed to make more efficient.
Whereas the supply chain of AI-servers is currently facing some bottlenecks that might moderate AI growth and power consumption in the near future, experts have predicted that it won’t be long before these bottlenecks are resolved. Indeed, a KPMG survey has found that 83% of industry executives are confident that semiconductor chip shortages that hit at the height of the Covid pandemic will largely ease by the end of 2023.
The Digiconimist report contains a call to action to be mindful about the use of AI. The paper notes that emerging technologies such as AI and previously blockchain tend to be accompanied by a lot of hype and fear of missing out, leading to the creation of applications that yield little to no benefit to end-users. Digiconomist says this waste can be mitigated by attempting to build solutions that provide the best fit with the needs of the end-users.