Jun 25, 2024 · 22m · green-blueprint
Avoiding mistakes from the first smart meters
Mike Phillips · 14m spoken Stephen Lacey · 5m spoken ChatGPT (GPT-4o Voice) · 12s spoken OpenAI Demo Tester · 9s spoken
0:00 / 0:00
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Sense CEO Mike Phillips joins host Stephen Lacey on The Carbon Copy to discuss how high-resolution edge computing and AI can rectify the architectural failures of first-generation smart meters and modernize the electrical grid.
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 4.4 Guest teaching 4.4 Guest disagreement 0.9 The hosts pushing back 1.6
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 11:32 Prioritizing raw data over edge GPU processing hype
Phillips firmly pushes back against the trend of prioritizing heavy GPU compute at the edge, asserting that lack of raw data is the fatal bottleneck rather than compute limitations.
Hardest push from the hosts ▶ 16:29 Demanding concrete technological pitfalls for utility buyersLacey pushes Phillips past polite generalities to specify the exact technological choices that constitute a failed smart meter procurement.
Biggest teaching moment ▶ 5:41 Exposing AMI 1.0 latency with the Google Maps comparisonPhillips clearly educates the listener and host by demonstrating why 15-minute delayed batch meter data is as useless as receiving driving directions a day after the commute.
The host holds their own ▶ 11:16 Challenging meter hardware assumptions via Nvidia and UtilidataLacey demonstrates domain knowledge by citing Nvidia's partnership with Utilidata to challenge Phillips on whether advanced AI meters require dedicated GPU silicon.
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 |
|---|---|---|---|---|---|---|
| Voice AI Breakthroughs and Conversational Sarcasm | 3 | 1 | 0 | 0 | Host Stephen Lacey introduces GPT-4o voice capabilities and narrates Mike Phillips' background bridging speech recognition AI at MIT to disaggregating household electrical waveforms at Sense. The dynamic is introductory and contextual without friction. | |
| Uncovering Grid Visibility Through Edge Computing Hardware | 4 | 4 | 1 | 2 | Lacey asks what visibility Sense uncovered beyond the home, prompting Phillips to explain how million-sample-per-second edge hardware revealed grid-level behavior. Lacey playfully summarizes that first-gen smart meters were not smart. | |
| Why Legacy Smart Meter Architecture Failed | 5 | 5 | 1 | 1 | Lacey frames a structured diagnosis of AMI 1.0 failures, asking whether hardware limits or utility data paralysis were to blame. Phillips illustrates the architectural flaw with an analogy of running Google Maps on 15-minute delayed batch data. | |
| Unlocking Grid Edge AI with High-Resolution Data | 4 | 5 | 2 | 2 | Lacey prompts Phillips on why current utility reinvestments remain disappointing. Phillips stresses that cloud-based grid AI fails without genuine high-resolution waveform data captured directly at the meter edge. | |
| Technical Metrics: Sampling Rates and Harmonic Analysis | 6 | 6 | 2 | 3 | Lacey presses on hardware requirements by citing Nvidia's partnership with Utilidata for meter GPUs. Phillips reframes the problem, explaining that algorithmic efficiency can manage compute constraints but missing high-frequency data cannot be recovered. | |
| Consumer Engagement and Distribution Grid Reliability | 4 | 5 | 0 | 1 | Lacey bifurcates power sector use cases into consumer and grid domains. Phillips explains how sub-second sampling enables real-time demand flexibility on the consumer side while detecting transformer arc flashes and vegetation contact on the distribution side. | |
| The High Stakes of Fifteen-Year Metering Lifecycles | 5 | 5 | 1 | 3 | Lacey highlights the 15-year procurement lock-in cycle and presses Phillips to name the specific technological mistakes utilities risk making today. Phillips outlines the core requirements: waveform sampling, upgradable compute headroom, and real-time networking. | |
| Scaling AI Paradigms and Next-Gen Meter Roadmap | 4 | 4 | 0 | 1 | Lacey asks about broader machine learning developments, leading Phillips to draw parallels between LLM token scale and power waveform sampling up to megahertz frequencies for future arc-fault use cases. |
Statements from this episode (14)
Phillips: Smart home devices still fail to communicate their operational status
“Look, if smart devices were fully enabled and every device out in your home told us what it was doing, we'd be happy to use that. But it's far from being the case.”
Phillips: Sense devices collect up to one million data samples per second
“We started to build these little orange boxes that go inside electrical panels and collect data at super high resolution, up to a million samples a second in the little orange box.”
Phillips: High-resolution edge computing enables real-time grid visibility
“Technical capabilities, so high resolution data, edge computing, real-time networking that we use to interact with consumers on what's happening in their home, we can look the other direction and we can see what the grid is doing in real time from the edge.”
Phillips: Treat smart meters as distributed computing platforms, not collection devices
“Once you start to think of this as a distributed platform, not just a data collection device, this entire world of making use of the data at the edge and AI machine learning at the edge starts to get opened up.”
Phillips: Legacy smart meter architecture cannot support real-time grid visibility
“Low-resolution data sent up to the service provider, made available later on, and look, there's some things you can do with that, but you just can't have a real-time, consumer-facing app, and then you can't see the grid in real-time from the edge either.”
Phillips: High-resolution meter data enables real-time grid fault detection
“What we've learned since then is that same high-resolution data lets us see, ah, the grid from the edge, lets us see transformers arcing, lets us see vegetation hanging power lines in real-time. That only happens through high-resolution data.”
Phillips: Utilities continue buying obsolete first-generation smart meters
“And there's still decisions being made for a previous generations of meters.”
Phillips: Modern smart meters capture 50 million times more data
“There's meters on the market that are doing that at 15,000 times per second. So, so that's fifty million times more data than AMI one dot O.”
Lacey: NVIDIA is partnering with Utilidata on smart meter edge computing
“So GPU chip maker NVIDIA is working with a company like Utilidata to add a lot more processing power into meters.”
Phillips: High-frequency data access matters more than edge processing power
“I've been a little less concerned about that, because for computation, you can always be very efficient about your algorithms, and you can make trade-offs from the algorithms. The reason we're more concerned about the data is if you don't have the data, you do…”
Phillips: Legacy utility portals suffer from low consumer engagement
“Look, frankly, not a lot of people use those portals. I think there's pretty well defined metrics that these haven't got a lot of use.”
Lacey: Utility smart meter rollouts occur on 15-year procurement cycles
“These rollouts tend to happen on around 15 year cycles.”
Phillips: Next-gen smart meters require three core technical capabilities
“There are just these three main topics that I've been talking about, which is getting access to the right data, so high-resolution waveform data. It's well known how to do that. It's mature technology, we know how to do that. Having enough computation so that …”
Phillips: Sense develops megahertz-sampling meters to detect transformer arcing
“We are working with some of the meter makers to have the next generations of meters that have even higher sampling rates than that, going all the way up to megahertz, and that's unlocking some more use cases around arc fault detection, you know, when there's a…”