AI is evolving from a mere consumer of infrastructure into a force for efficiency enhancement, with the "Infratech" model—which utilizes AI technology to optimize the performance of power and industrial systems—gaining momentum in Silicon Valley. This trend holds promise for alleviating the environmental pressures associated with the AI boom and addressing bottlenecks in grid efficiency.
On February 3, according to a report by tech media The Information, venture capital firm Blue Bear Capital showcased the latest developments in this field at a CEO summit held in Jackson Hole, Wyoming. Its founding partner, Ernst Sack, pointed out that the massive current investments in hardware like power lines, concrete, and fiber optics will ultimately unlock decades of asset-light technological innovation, thereby creating value from these infrastructures. He emphasized that AI technology can "heal" some of the environmental harms unleashed by the resource-intensive AI boom, with the core principle being to make existing systems operate more efficiently.
This strategy of using digital means to enhance the returns on infrastructure investment is demonstrating investment resilience that transcends political cycles. Jane Woodward, a fund manager at WovenEarth Ventures, stated at the summit that digital startups capable of significantly boosting infrastructure returns have the opportunity to achieve a leap in value, a form of value enhancement not constrained by whoever is the sitting president.
Simultaneously, although the U.S. power grid currently faces forced utilization limits—often as low as 50%—due to risks like line sagging from wind or heat causing wildfires, regulatory bodies have begun encouraging the adoption of new technologies that can enhance grid efficiency. Kai-Philipp Kairies, co-founder of battery intelligence startup ACCURE, stated,
"AI truly is our best opportunity to close the electricity gap."
Analysis suggests this shift indicates that the energy industry's discourse is moving from "energy transition" to "energy addition"—that is, achieving a non-linear leap in production capacity by using existing facilities more intelligently. The convergence of capital efficiency and technological pathways is enabling asset-light models to leverage heavy assets.
Since its founding in 2016, Blue Bear Capital has invested approximately $350 million in around 40 startups focused on enhancing grid efficiency, improving the performance of renewable and industrial assets, and optimizing energy data analytics. With rising electricity prices and surging power demand, this investment strategy is reaching a critical juncture.
Reportedly, unlike the "rollercoaster" experience of renewable energy builders last year amid policy changes under the Trump administration, technology providers focused on boosting returns from power assets have sidestepped such policy risks.
Blue Bear's portfolio companies often exhibit a unique combination of traits: they are asset-light software manufacturers, yet their founders are experts in heavy-asset sectors like power grids and nuclear reactors. They utilize AI to parse data and images related to power flows, weather, and physical materials (like battery cells), thereby enabling scarce resources such as water, labor, and energy to be used more effectively.
While corporate venture arms like those of NextEra Energy and energy giant Chevron compete with Blue Bear, they also participate as co-investors. These partners share the consensus that all forms of energy are currently indispensable, and the industry's focus is on how to achieve a net increase in energy supply through technological means. AI is reshaping grid management, enabling everything from millisecond-level decision-making to unlocking latent potential.
The U.S. power grid has long been plagued by inefficiency; to avoid safety incidents, utility companies often set arbitrary cut-off points based on average conditions, limiting power transmission capacity.
Splight, a company backed by Blue Bear, uses AI technology to make automated decisions within milliseconds, safely scaling up power transmission from sources like wind, solar, battery farms, or data centers.
Fernando Llaver, co-founder of Splight and a former Argentine grid operator, stated that before using Splight's technology, power generators were instructed to connect to the grid using only a fraction of their capacity—as little as 60%.
In a recent deployment case, Splight used real-time signals to safely inject 412,448 megawatt-hours of electricity into the grid—power that would otherwise have been unnecessarily curtailed. This volume is equivalent to the annual electricity consumption of a small city. Splight's tools are already widely used in South American and European grids and are expanding into the U.S. market.
Furthermore, large tech companies are also entering this arena. Google X's "moonshot" project, Tapestry, has partnered with Chile's national grid operator, using AI to simulate and plan power flows, and is currently helping to streamline the process for connecting new power sources to one of the busiest grids in the United States. Smart metering giant Itron also acquired Urbint, another Blue Bear investment, last autumn. Beyond grid transmission, utility companies are restarting decommissioned nuclear plants to access clean energy.
For these decades-old nuclear facilities, whose control rooms resemble scenes from old sci-fi movies, startup Nuclearn has developed the first generative AI chatbot specifically designed for the nuclear energy sector.
Nuclearn's large language model complies with nuclear safety protocols and can ingest and understand records created as far back as the 1980s and 1990s to address issues like compliance. Its software has already been adopted by over 80 nuclear reactors in the United States, Canada, and the Middle East, including the installation of the first AI graphics processing unit behind a nuclear environment firewall at the Tennessee Valley Authority.
In the operations and maintenance (O&M) sector, AI-powered infrastructure is seen as an upgrade over the traditional Internet of Things (IoT). Conventional IoT relies on manually installed sensors that are prone to failure, whereas the new model utilizes autonomous robots or drones that can diagnose faults simply by collecting and processing images or detecting abnormal heat signatures.
Nikhil Vadhavkar, co-founder and CEO of solar monitoring company Raptor Maps, another Blue Bear investment, pointed out that in an environment where O&M personnel attrition rates can be as high as 30% annually and O&M companies operate on very thin margins, such early warnings can significantly reduce costs associated with components and labor.
Financial returns in this sector are already beginning to materialize. Omnidian, a solar data analytics platform in which Blue Bear invested in 2017, has seen its annual revenue grow from $300,000 at the time of investment to nearly $50 million last year, with projections reaching $75 million this year, positioning it on a potential path towards a future IPO or sale.
Comments