Episode Summary
Executive Summary: Sean McAvoy explains how Grid Beyond uses AI software to unlock flexibility in energy assets—batteries, generators, industrial loads, and data centers—and connect them to power markets. The conversation covers the boom-bust cycles of battery markets, the rise of local, millisecond-scale AI decision-making, and how data centers are driving a scramble for fast, reliable power solutions.
Main Topics: Grid Beyond’s origin and platform strategy (Priority: 5/5): Sean describes how he moved from AI software into energy by turning battery-related patents into a software business, ultimately building a platform that sits between energy assets and markets to forecast, optimize, and bid power automatically. AI at the edge for ultra-fast grid response (Priority: 5/5): The discussion highlights a major shift from cloud-centered decision-making to local inferencing at the site level, enabling sub-50 millisecond response times for grid events, blackouts, and brownouts. Battery market cycles across regions (Priority: 5/5): Sean outlines a repeating pattern where battery markets saturate, prices fall, financing becomes harder, and then demand returns as load grows—illustrated by the UK, ERCOT, Japan, and Australia. Data centers as a new flexibility challenge (Priority: 5/5): Data centers are increasingly central to grid planning. Grid Beyond helps orchestrate cooling, workloads, batteries, and generators to manage curtailment, reliability, and interconnection constraints. Resource mix shifts: gas, nuclear, geothermal, and storage (Priority: 4/5): The interview emphasizes that no single resource is a silver bullet for powering new load. Customers are considering nearly every option available, from gas turbines and batteries to geothermal, nuclear reopenings, and future SMRs. Collaboration and the ‘bring your own capacity’ model (Priority: 4/5): Sean argues that utilities, data centers, retailers, and generators must collaborate so new load can come online without shifting costs to ratepayers. Batteries can be deployed where they help decongest the grid and speed interconnection. The energy transition is still happening, but with speed bumps (Priority: 3/5): Despite setbacks and shifting policy priorities, Sean says the transition continues, increasingly driven by hyperscalers and private companies building their own energy management capabilities.
Key Arguments: Energy flexibility is now a software problem: AI can coordinate batteries, loads, generators, and markets much faster than manual or cloud-only systems. Battery markets are cyclical; as capacity floods a market, prices fall and financing gets harder, but demand growth eventually restores pricing power. Local inferencing matters because grid events require millisecond-level response, making edge AI essential. Data centers generally do not want to curtail; the practical path is to use batteries, generators, workload shifting, and smarter siting to reduce disruption. Batteries help data centers and grids, but they are only supplemental; firm baseload power is still the top need. The market is moving toward hybrid solutions and utility collaboration rather than a single clean-energy fix. The largest tech companies are becoming energy operators themselves, filling gaps left by broader policy or market slowdowns.
Data Points: Battery market saturation in the UK: Occurred about 5-6 years before ERCOT - Sean says the UK saw battery saturation before Texas, causing prices to drop as supply increased. Battery project size: 50 MW to 200 MW - Examples of lithium battery projects Sean cites when discussing financing and monetization. Battery project cost: Hundreds of millions of dollars - Used to explain why falling market prices can undermine project economics. Last strong battery pricing year in Texas: 2023 - Sean says ERCOT’s last really good year for battery pricing was around 2023 before prices weakened in 2024-2025. ERCOT pricing drop: Started in 2024 and fell further in 2025 - Describes the post-boom decline in battery market returns in Texas. Japan battery pricing: Around $200,000 per megawatt in Tokyo for FCR - Illustrates how new markets can be highly lucrative at the start of their battery buildout cycle. Data center curtailment threshold in Texas: 70 MW and over - Sean references Texas Senate Bill 6 as an example of curtailment requirements for large facilities. Response target: Sub-50 millisecond response - Industry goal for local AI decision-making to react to grid events quickly. Event timing example: Thursday at 4 o'clock - Illustrative example of how Grid Beyond can predict and prepare for an energy event in advance. Data center battery duration: 4 to 8 hours - Sean notes longer-duration batteries are now possible depending on chemistry. Curtailment window: 30 minutes to 1 hour - Sean says this is a practical curtailment duration for some data center use cases. Gas turbine lead time: 3 years - Used to show why gas is not an immediate solution for data center power demand.
Pivotal Quotes: "There’s nothing like this, like the pace at which it’s changing, the pace of the capabilities that are at your hands these days." — Sean McAvoy: He is comparing the current AI era to previous tech acceleration cycles and emphasizing unprecedented speed. "We uncover flexibility and once we find it, we ask the customer what they want to do with it, whether it’s revenue generation, cost savings or resiliency/slash reliability." — Sean McAvoy: This captures Grid Beyond’s core value proposition in helping customers monetize or use flexibility. "The energy transition is still happening. It has gone through a couple of bumps." — Sean McAvoy: Sean’s summary of the broader transition: slowed by policy and market swings, but still advancing.
Implications: Energy infrastructure is becoming an AI-orchestrated system. Expect more edge computing, hybrid power deals, utility collaboration, and fast-growing demand for flexible assets as data centers, batteries, and industrial loads reshape grid planning.