sylvain gignac

Might say that natural gas fits the same logic even more tightly.

How an early cold snap pushed Hydro-Québec's power grid to its limit | CBC News A recent cold snap pushed Hydro-Québec's network near its limit, forcing the utility to issue two rare alerts as it scrambled to meet the surge in demand for power.

An unseasonably cold snap in December pushed HQ’s system near its limit, forcing it to issue two Level 2 alerts. Bad timing: peak demand hit right as turbines were down for maintenance. Often-overlooked factor that kept the lights on: power imports. Might become recurrent.
www.cbc.ca/news/canada/...

Venezuela...with become an american bulge. To be sardonic.

With software, pivots are just a thought spell = tweak the code, search for new investors, boom, magic. In power grids' reality, we’re still bound by the ancient laws of mass, inertia + thermal limits. Can’t ask a transformer to "pivot"; it doesn’t respond like or well to over increasing AI query.

so we’re not dealing with a temporary bulge but a new baseline?

Funny how Big Tech assumes planners + op carry magic dust to handle their ever-changing wishes. I once thought IBRs were hard. Now the real challenge is BT’s built-in belief that the grid should behave like softwares = unconstrained by physics, timelines, or reality.

Not surprising, Crocs originated in Quebec City, where we are accustomed to sub-freezing walks.

Destabilizing loads with expectations outpacing the physical constraints of the grid + steep rampers. Personally their behavior just messes with all system modeling LT or ST.

As of Monday, the lab has been renamed "National Laboratory of the Rockies". The DOE's shift toward a “applied energy” rather than "sole" focus on renewable energy. www.nrel.gov/news/detail/...

Virginia creates a new rate class for loads ≥25 MW, mostly DC starting 01-01-27. Must shoulder more of the grid costs: at least 85% of contracted T&D demand and 60% of generation demand = avoiding that everyone else pays for unused capacity built for DC.
www.scc.virginia.gov/about-the-sc...

Forest caribou grazing...very few left
Moose overlooking my car
So many white-tailed deers frolicking...loosing count
Osprey catching a fish
Bobcats with little ones... lasted few seconds.

Monitoring stays. But it will lag. Tech + interdependencies btw systems might outpace any oversight we can build.

Thanks, just learned that the tech was deployed at home..."carbon-free anode inside an existing smelter in Alma, Quebec."

HQ’s rising demand won’t touch existing long-term PPAs. As for NECEC, it's locked in for 20 years at 9.45 TWh/yr up to 2045. Few smaller contracts will end earlier. But with low reservoirs + fast local load growth in Qc, extending or adding new export deals after these terms might be harder.

Good question heard today = At what point does an external AI, not its physical assets, become a functional part of grid planning or operations? If system operators increasingly rely on AI tools for forecasting, siting, topology optimization, situational awareness, and tutti quanti.

Hydro-Québec anticipe le boom des centres de données Hydro-Québec a presque doublé sa prévision de consommation d'électricité des centres de données, avec la nouvelle vision économique du gouvernement Legault.

"Currently, Quebec has 81 data centers. Consumption in this industry in the province is projected to increase sevenfold, giving an idea of ​​the increase in the number and size of these server buildings over the next 10 years."

ici.radio-canada.ca/nouvelle/220...

NERC tells FERC the rapid growth of large loads = a core reliability challenge. DC/AI loads are scaling too fast to forecast + already causing disturbances. NERC stresses that AI/DC should fall under NERC’s reliability scope.

www.nerc.com/newsroom/ner...

So much confusion / hypes around AI even inside utilities or BPS. As for BPS, seen ML being used for forecasting, some predictive maintenance or fault location. As for LLMs, mostly used to summarize outage logs, searching standards (use this often), or helping with data queries. Different tools.

Do you have a PDF version of the report.

Great study on EGS as a game-changer for western and northern 🇨🇦. It already appears cost-competitive with other baseload options. But the real value is its flexibility, providing a backstop to variable gen and a credible CO2 free-answer to increasing energy deficits.

"...the technology can also be harmful if misapplied. In one high-profile example, around 40,000 Michigan residents were wrongly accused of unemployment insurance fraud based on a state AI system with a faulty algorithm and inadequate human oversight." 🤷‍♂️ // power grid
www.nerc.com/globalassets...

Data Centers and Natural Gas Are Bending the Climate Transition Curve, IEA Says The group’s latest World Energy Outlook reflects the sharp swerve in U.S. policy over the past year.

" ... No other country has made such a substantial recent policy U-turn in the past year, turning against renewables deployment at the same time as it is seeing electricity demand leap up thanks to data centers."
heatmap.news/energy/iea-w...

Another challenge is balancing since data centers can switch computational work btw different regions / balancing authorities for reasons often unknown to operators like price sensitivity, emission... Difficulty to forecast = increased reserves = increased costs for all market participants.

Grid Strategies analysis of summer peak load forecasts

Grid Strategies has analyzed utilities' summer peak load forecasts: an astounding 166 GW, with data centers representing 90 GW. But it cautions that the utility industry has a long history of exaggerated forecasts to justify massive investments in power plants & grid infrastructure.

2025–2026 Winter Reliability Assessment = "Potential for insufficient operating reserves in normal peak conditions" If demand leads to a shortfall of generation and all operating reserves are exhausted, manual load shed may be needed to stabilize the system.
www.nerc.com/globalassets...

Just read The Light That Failed that stresses “What ensured that these revolutions would remain ‘velvet’ was their background hostility to utopias and political experiments.” An interesting read for a North American.

Large computational loads are forcing a permanent shift in how we plan, model, and operate the grid. The era of predictable, passive demand is over. Recommended read for planners, ops, and AI/DC developers. I’m honestly overwhelmed by the scale of the work ahead.

www.nerc.com/globalassets...

From a planning perspective, the real crisis might be more than silicon. It will be heat extraction at scale, substation buildout, grid integration, upstream transmission adequacy, local hosting capacity, transformer shortages + inertia and stability with massive IBR-fed loads.

"Rack densities have gone from an average of 6 kilowatts per rack eight years ago to the point where racks are now shipping with 270 kW,” says David Holmes, the global industries CTO at Dell Technologies. “Next year, 480 kW will be ready, and megawatt racks will be with us within two years.”

Solid analysis on why gas is getting expensive in the Northeast. Interested in getting further studies on how to replace it to tackle winter peaks.