Mar 14, 2024 · 26m · green-blueprint
The achilles heel of AI in the power system: data
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 The Carbon Copy, host Stephen Lacey and Snowflake's Titian Palazzi discuss how modern data architecture and artificial intelligence are transforming electric utilities, addressing legacy data bottlenecks, and driving operational efficiency across the clean energy transition.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The hosts as informed peer 5.2 Guest teaching 5.8 Guest disagreement 0.2 The hosts pushing back 0.8
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 23:58 Skepticism of extreme utility load growth forecasts
Palazzi politely pushes back against panic over massive utility load forecasts, pointing out that historical long-term energy projections are often inaccurate.
Hardest push from the hosts ▶ 23:10 Host challenges guest on AI energy drain vs climate gainsLacey directly challenges whether the exponential computing power demands of data centers will cause clean energy progress to run in place.
Biggest teaching moment ▶ 7:57 The 9-month data collection reality checkPalazzi educates the audience and host on the imbalance in AI projects, where data retrieval and cleanup takes 9 months while model training takes weeks.
The host holds their own ▶ 16:56 Detailed breakdown of utility organizational hurdlesLacey articulates specific internal utility dynamics regarding external partnerships, hiring non-energy data scientists, and creating test sandboxes.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| The Evolution of AI and Snowflake's Data Cloud Role | 4 | 5 | 0 | 0 | Lacey sets up the interview with context on Snowflake and time-series forecasting, framing the conversation clearly. Palazzi explains how generative AI and natural language interfaces are replacing older regression models and broadening adoption. | |
| Overcoming Utility Data Bottlenecks and Grid Visibility Constraints | 5 | 6 | 0 | 1 | Lacey asks detailed operational questions about utility data limits and why dirty data holds back AI models. Palazzi explains how data collection consumes months of effort compared to model development, stalling grid interconnection and visibility. | |
| Practical AI Use Cases in Solar, Operations, and Markets | 5 | 7 | 0 | 0 | Lacey prompts for concrete industry examples across the value chain. Palazzi provides rich case studies including rooftop solar predictive maintenance, BP LightSource hail damage mitigation, and utility wholesale trading savings. | |
| Organizational Strategies for AI Adoption and Industry Data Sharing | 6 | 6 | 0 | 1 | Lacey demonstrates domain knowledge about utility staffing challenges and sandbox experimentation. Palazzi outlines the necessity of embedding data scientists with operational teams and advancing cross-utility data sharing. | |
| AI Vendor Integration, Power Demands, and Net Climate Impact | 6 | 5 | 1 | 2 | Lacey presses on the tension between AI computational power consumption and clean energy decarbonization goals. Palazzi provides a measured counterperspective, noting historical forecasting errors and arguing operational grid efficiencies will outweigh data center energy demand. |
Statements from this episode (13)
Lacey: AI time series models boost grid forecasting accuracy by 30-50%
“What they found was that these AI-driven time series models could improve accuracy by 30 to 50%.”
Palazzi: Energy companies are shifting from hard-coded models to production AI
“I think that's a big shift that has taken place, that in production by companies that are serving real customers, AI is used much more commonly. So a shift from hard-coded, pre-set-up, fully visible models into more machine learning and AI.”
Palazzi: LLMs could soon replace traditional time-series forecasting algorithms
“It's quite possible that in the not too distant future, you might actually do forecasting by asking an LLM, can you predict the next few weeks of data? Whereas to date, you would use very different algorithms for that.”
Palazzi: Utilities Spend Far More Time on Data Cleaning Than AI
“What I've seen is that often it can take three months or six months or nine months to get access to all the right data, such as smart meter data for your customer, and then only a couple of weeks or a couple of months to actually build and deploy a predictive …”
Palazzi: Limited Grid Visibility Restricts Renewable Energy Additions
“And the reality is that many of them don't really have visibility into their network in a way that allows them to dynamically manage both new resources being added, and then to ensure that the grid operates in a reliable manner. And then as a result, what they…”
Palazzi: Hail Damage Accounts for the Bulk of Solar Insurance Claims
“Some of the bulk of insurance claims for solar asset owners is actually hail damage to panels, just breaking the glass”
Palazzi: One utility saved $5M in a month improving demand forecasts
“Apparently by improving the predictions, they saved more than five million dollars in a single month by avoiding exposure to a big real-time price spike.”
Palazzi: Relatively few utilities have live, fully productionized AI use cases
“I think the amount of utilities that have fully productionized use cases for AI live is relatively small.”
Palazzi: Embedding data scientists with utility line workers accelerates AI adoption
“So something specific I've seen work very well is to embed teams of data scientists and software engineers with the lines of business. So to make sure that somebody who is a data scientist is actually sitting with a trader for a week, or sitting with the grid …”
Palazzi: California IOUs should share smart meter data with Sunrun and SunPower
“In California, the smart meter data that the three big investor on utilities have actually should be shared with a variety of different players. So for example, the distributed energy companies like Sunrun or Sunpower They can really benefit from having smart …”
Palazzi: 10x compute growth barely increased data center energy use
“What we've seen is that in the last 1520 years, while the amount of compute and data storage has increased a lot, in some cases more than 10 X, depending on the geography and time frame, the energy consumption has actually barely budged. It has barely gone up,…”
Palazzi: Utilities project up to 10x data center power demand growth
“I've seen some utilities in their integrated resource plans mention Growth of two X, five X, even 10 X in the next decade, driven primarily by more data center build out.”
Palazzi: AI emissions reduction benefits will far outweigh data center power demands
“So I generally think that we should proceed with the technology innovations that we're seeing. And in fact, we may not really have the full control over not doing so. And that the advances that we will get in emission reduction and energy consumption reduction…”