Mar 8, 2024 · 38m · catalyst
The early days of AI on the grid
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of Catalyst, host Shayle Kann and energy strategist David Groarke examine the emerging role of artificial intelligence across electric utilities, assessing high-value operational use cases, structural grid constraints, and the competitive vendor landscape.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Shayle holds 25.9% of the talking time here. How this is scored →
Shayle as informed peer 5.3 Guest teaching 4.0 Guest disagreement 0.8 Shayle pushing back 1.5
speaking balance: gold is Shayle, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 31:02 Conceding startup funding limitations
David pushes back slightly on the startup outlook before conceding to Shayle that venture capital figures do not prove market adoption against incumbent grid giants.
Hardest push from Shayle ▶ 30:54 Challenging VC funding as a proxy for utility successShayle interrupts David to demand clarification on whether $1.5 billion represents customer revenue or venture capital, pointing out that venture funding does not guarantee utility market viability.
Biggest teaching moment ▶ 9:48 Explaining grid physics and algorithmic constraintsDavid systematically outlines why AI algorithms struggle with power systems, citing Ohm's law, Kirchhoff's law, and the absence of millisecond-level data granularity.
Shayle holds their own ▶ 6:54 Calling out the failure of the first smart grid waveShayle demonstrates deep industry memory by directly stating that if the goal of the initial digital wave was to lower transmission and distribution costs, the sector failed.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Shayle as informed peer | Guest teaching | Guest disagreement | Shayle pushing back | Why |
|---|---|---|---|---|---|---|
| Framing the Role of AI in Utility Operations | 6 | 2 | 1 | 1 | Shayle draws on his 17 years in the energy sector and experience during the first smart grid wave to contextualize AI's current role. David agrees and builds on the premise by highlighting rising OPEX and improved algorithm development costs. | |
| Structural and Physical Barriers to Grid AI | 5 | 5 | 1 | 1 | Shayle prompts David on why the power sector is slower to adopt AI than sectors like legal tech. David educates on hard physical barriers including modeling Kirchhoff's and Ohm's laws, data latency gaps, and aging workforce dynamics. | |
| Categorizing AI Capabilities Across the Power Sector | 4 | 5 | 0 | 0 | Shayle asks for a clear taxonomy of AI subcategories relevant to utilities. David delivers an educational breakdown covering computer vision, predictive analytics, digital twins, distributed edge AI, and explainable AI. | |
| Midroll Sponsor Break from Bloom Energy and Engie | 5 | 4 | 0 | 1 | Following the midroll ad read, David explains the use of AI in wildfire mitigation and vegetation management. Shayle astutely notes why this is an ideal early AI use case: it is operationally isolated from core real-time grid reliability. | |
| Use Case 2: Customer Propensity and EV Detection | 5 | 4 | 1 | 1 | David explains non-intrusive load monitoring and EV detection from AMI smart meter data. Shayle sharpens the distinction by noting that the data is not new, but modern analytical methods and EV market penetration enable practical adoption. | |
| Use Case 3: Asset Health and Dynamic Line Rating | 5 | 5 | 0 | 0 | David details core operational AI applications including transformer health predictive maintenance and dynamic line ratings (DLR). Shayle listens and summarizes the high volume of practical, non-generative AI use cases across the grid. | |
| Utility AI Supplier Landscape: Startups vs. Conglomerates | 6 | 3 | 2 | 6 | When David presents a statistic showing $1.5 billion invested in AI startups, Shayle immediately intervenes to clarify that venture capital investment does not equate to utility customer traction or proven commercial victory. David concedes the distinction. | |
| The Road to Autonomous Grids: Incremental vs. Radical Change | 6 | 4 | 1 | 2 | Shayle sets up a structured three-part scenario framework for the future of grid automation (pipe dream vs incremental vs radical change). David evaluates the technical sensor requirements and lands firmly on incremental change. |
Statements from this episode (14)
Groarke: US smart meter penetration is over 70%
“All of these smart meters were at over 70% smart meters in the US.”
Groarke: Power delivery infrastructure cost nearly equals generation cost
“Just the cost of infrastructure needed to deliver power is nearly equal to the cost of generating power itself, right?”
Groarke: Power sector OPEX is increasing about 14% annually
“OPEX costs are up about, you know, 14% a year.”
Groarke: Modeling grid physics is the primary barrier to full automation
“The primary reason why, you know, we're not on a path Towards, you know, full automation, and there's, you know, been a spade of use cases launched is just modeling grid physics accurately is incredibly difficult, right, for an algorithm. So, you know, solutio…”
Groarke: Half of the utility workforce will retire in 10 years
“50% of the workforce is going to retire in the next 10 years. You can see that happening right now. So you have this kind of field technicians are leaving, right?”
Groarke: Utility LLM use focuses on enterprise tasks, not grid operations
“We're seeing some application of LLMs across the industry, but usually at the enterprise level around regulatory documents around, you know, generating text or guidance. Less on the actual power sector side of the operation itself itself.”
Groarke: PG&E had a $1.3B wildfire rate case in 2022
“PG&E in, in, in 2002 22 had a 1.3 billion dollar rate case, right, for the wildfires that happened that year over a three-year period.”
Groarke: Utilities spend billions annually on vegetation management
“Anything that helps vegetation management, which utilities spend billions a year on collectively is interesting.”
Groarke: Duke and SCE use live AI for customer EV detection
“So that non-intrusive load monitoring, using external data, identifying consumption, and selling new products and services is happening now. And that's live at utilities, right? Just like the wildfire example there's live at Duke and Southern California Edison…”
Groarke: AI predictive maintenance cuts utility equipment downtime by up to 50%
“You can reduce downtime Of some of these critical components by like, 30, 50%.”
Groarke: Utility AI startups raised $1.5B across 80 rounds since 2021
“Startups received 1.5 billion, right, from 80 rounds of funding in related to those deployments since twenty-twenty-one.”
Groarke: Grid startups focus on edge applications while conglomerates dominate core operations
“I think most of the startups we found were playing in the Dior integration space or EV charging management place. We saw some startups in the Enterprise type of use case space even looking at regulatory documents with LLMs, for example as you go down the list …”
Groarke: AI is not creating a massive new utility market
“I can't say that there is an enormous new market emerging because of AI. As a result, I think AI is kind of riding the wave within the sector.”
Groarke: Grid data lacks the coverage required for real-time automated decisions
“The data right now is, is not available, right, for, let's say, real-time decision-making across the grid, right? Like, what we have is AMI, right, which is minutes to hours. We have SCADA, which is, you know, seconds to minutes, and then we have PMUs, which a…”