Rocks, Racks & Rights: Canada’s Blueprint for the AI Energy Wars
Meta’s planning a 5-gigawatt data center called Hyperion. That’s not a server farm anymore. That’s a small province worth of electricity, humming 24/7 to train the next GPT killer. While Zuckerberg’s engineers calculate cooling requirements, I’m sitting in Vancouver watching our city’s AI community explode from studio meetups to a thriving network of companies and professionals, wondering if we’re about to repeat every colonial mistake in the playbook… or finally write a different ending.
Here’s the thing nobody in Silicon Valley wants to admit: AI isn’t software anymore. It’s heavy industry. It needs copper, lithium, cobalt, rare earths. It needs massive amounts of electricity, water for cooling, and political stability to keep the lights on. It needs everything Canada has spent 150 years pulling out of the ground. Except this time, instead of shipping raw logs and crude oil, we could be shipping intelligence itself. Or better yet, keeping it here and selling access on our terms.
The old story goes like this: Canada digs stuff up, America makes it smart, China makes it cheap, and we buy it back at markup. That story made a few people rich and left most communities holding empty mines and poisoned watersheds. The new story could go differently, but only if we stop treating AI like it’s just another app and start treating it like the civilizational infrastructure it’s becoming.
The Great Convergence
Stuart Muir gets it. That’s why I’m heading to his Power Struggle podcast this Sunday, to connect dots that shouldn’t need connecting but somehow do. Energy and AI aren’t separate conversations anymore. They’re the same conversation. The IEA says global data center electricity use could double by 2030, reaching Japan’s entire current consumption. BC Hydro’s Site C dam adds 1,100 megawatts to our grid. A massive project that took decades and billions. A single AI training cluster could swallow that whole.
But here’s where it gets interesting: BC already showed its hand. The province extended its moratorium on new crypto mining connections to “preserve clean electricity for people and businesses.” That’s precedent. That’s the government saying: not all compute is created equal. We get to choose what kind of digital future we power.
So what kind should we choose?
The Energy Reality Check
Let me be clear about something before we go further. I’m not Team Fossil or Team Fantasy. I’m Team Outcome. Canada will continue to extract resources, build infrastructure, and yes, probably export LNG. But let’s stop pretending the climate math is simple.
The IEA itself says gas only beats coal if methane emissions stay truly low… and we’re nowhere near that today. Only about 5% of global oil and gas output meets near-zero methane standards. Satellite data keeps catching massive plumes nobody reported. There’s peer-reviewed research suggesting some LNG supply chains might be worse than coal on a 20-year basis when you count the methane leakage.
So here’s my position: if AI wants gigawatts, it owes gigaton outcomes. Real, measured, verified outcomes. Not marketing decks. Not promises. Satellite-verified methane controls. Published water use. Waste heat recovery that actually heats buildings. Community benefit agreements with teeth.
We can’t keep outsourcing our integrity to PowerPoint presentations.
Rocks → Racks → Rights
Start with the rocks. Canada has what the MIT researchers call “the periodic table of battery materials”: lithium, nickel, cobalt, graphite, all the rare earths that make electrons dance in useful patterns. The federal Critical Minerals Strategy talks a good game about “mines to mobility,” but that’s thinking too small. Every Tesla battery could be powering AI inference. Every copper vein could be carrying the thoughts of machines. The question isn’t whether we mine. It’s whether we capture the value chain.
Traditional extraction looks like this: dig it up, ship it out, buy it back. The Canadian special. But imagine this instead: critical minerals stay in-country for processing. Processing feeds local battery and chip manufacturing. Manufacturing supplies Canadian-owned data centers. Data centers run on Canadian hydro, nuclear, and geothermal. The compute gets packaged as a utility… metered, regulated, and accessible to Canadian startups, researchers, and yes, Indigenous nations who’ve been locked out of every previous industrial revolution.
That’s rocks into racks. But the rights part? That’s where Canada could actually lead instead of follow.
The Sovereignty Stack
When a Wet’suwet’en nation member’s data gets processed, who owns the inference? When an AI trained on Canadian medical records makes a breakthrough, who profits? When our clean electricity powers the training run that creates AGI, do we get a seat at the table or just the electric bill?
These aren’t hypothetical questions anymore. They’re the questions that will determine whether AI becomes another vector for extraction or a tool for regeneration. And Canada, with its particular mix of resources, energy, and half-decent civil society, might be the only country positioned to answer them right.
Indigenous data sovereignty isn’t some nice-to-have progressive add-on. It’s the prototype for how all communities should relate to AI infrastructure. The First Nations Information Governance Centre has been developing OCAP® principles around ownership, control, access, and possession of data for decades.
But here’s the difference between what I’m proposing and the usual corporate playbook: equity stakes and governance from day one. Not consultation. Not engagement. Not late-stage PR fixes after the permits are signed. Real ownership. Real decision-making power. Real revenue sharing. Where projects touch Indigenous lands, Nations should have board seats, not just benefit agreements.
Vancouver’s Lizard Brain Problem
But we have to be honest about our weaknesses too. Vancouver, my beloved city of glass towers and impossible housing prices, has a problem. We know how to do resource extraction. Hell, our stock exchange is basically a global casino for junior mining companies. That pump-and-dump muscle memory, that extractive reflex… it’s deep in our lizard brain. We see a resource, we figure out how to flip it, not how to build with it.
You can see it in our tech scene. We grow companies to Series A, then watch them get acquired by Seattle or San Francisco. We train brilliant engineers at UBC and SFU, then wave goodbye as they head south. We’ve internalized our own colonial position so thoroughly that we do it to ourselves before anyone else gets the chance.
Breaking that pattern means rejecting the false choice between resource economy and innovation economy. The real innovation is figuring out how to be both. How to power the future without repeating the past.
The Energy-Intelligence Covenant
Here’s my proposal, and I’ll be dropping this on Stewart’s podcast like a challenge coin: Every gigawatt of power we allocate to AI must come with a gigaton covenant. You want our clean electricity to train your models? Show us the carbon you’re cutting. You want our minerals for your chips? Show us the community benefits. You want our engineers for your moonshots? Keep the IP here and share the profits.
This isn’t anti-business. It’s pro-business-that-makes-sense. Microsoft’s already signing deals to restart nuclear plants for AI. Google’s investing in geothermal. They know the energy math doesn’t work without massive clean power expansion. The question is whether that expansion happens TO communities or WITH them.
BC could lead by example. Imagine a provincial Clean Compute Compact that:
- Mandates a public Compute Registry: Every data center over 10MW publishes power sources, water use, methane controls (if using gas peaking), waste heat recovery rates, local employment, and Nation partnerships
- Prioritizes verifiable outcomes: Interconnection priority goes to projects that can prove carbon reduction, grid stability benefits, and community value
- Requires demand-response agreements: When the grid’s stressed, compute flexes. No exceptions
- Creates public compute utilities: Shared GPU clusters accessible to startups, researchers, and First Nations at cost
That crypto moratorium? That was practice. This is the real game.
The Transparency Test
If Resource Works and the industry associations want to be “the adults in the room,” as they like to say, then let’s act like adults. Adults show their work. Adults publish their data. Adults admit when the methane math doesn’t add up.
I’ll ask Stewart this on air: Will you support mandatory, satellite-verified methane standards for any gas that claims climate benefits? Will you back a public registry for AI and data center loads? Will you support equity stakes for Indigenous nations from day one, not as a late-stage checkbox?
If the answer is anything but yes, then we’re not having an adult conversation. We’re having a marketing meeting.
Jobs, Dignity, and the Trades Renaissance
The San Francisco tech bros picture AI jobs as PhDs optimizing algorithms in glass offices. But real AI infrastructure needs electricians, pipe fitters, HVAC specialists, network engineers, security guards, and maintenance crews. It needs the same people who built Canada’s industrial economy, except this time they’re building server racks instead of oil rigs.
A single data center can employ hundreds directly and thousands indirectly. But only if we build them here, own them here, and operate them here. The alternative is becoming a battery… storing energy for someone else’s intelligence explosion.
This is where the vision gets practical. Every megawatt of AI compute should generate local employment metrics. Every data center should be required to partner with local technical colleges. Every AI company using Canadian power should contribute to a sovereign compute fund that provides free GPU hours to Canadian researchers and entrepreneurs.
The Both/And Economy
The binary thinking has to stop. It’s not:
- Resources OR innovation
- Environment OR economy
- Indigenous rights OR development
- Regulation OR progress
It’s both/and, always has been, always will be. Canada’s superpower has never been choosing sides. It’s been synthesizing opposites. We’re the country that made federalism work (mostly). We figured out how to be North American without being American. We can figure out how to be an AI superpower without selling our soul.
But it requires rejecting the learned helplessness that says we’re too small, too nice, too peripheral to matter. When Meta needs 5 gigawatts of clean power, we matter. When the AI supply chain needs critical minerals, we matter. When the world needs a model for ethical AI deployment, we matter.
The Resistance
There’s already resistance building, and that’s good. Seventy-two percent of British Columbians say they’re more concerned than excited about AI in daily life. Fifty-eight percent see unregulated AI as the bigger risk. This isn’t Luddism. It’s wisdom. People know when they’re being sold something that benefits someone else.
BC residents want results with guardrails. Give them clean power, real co-benefits, and consent-based development, and they’ll back projects that add value. But try to ram through another extractive play with nice words about innovation? They’ll see right through it.
The resistance needs direction, not suppression. Channel it into:
- Community review boards for AI deployments with real veto power
- Public compute registries with mandatory disclosure
- Algorithmic audits for any AI touching public services
- Local ownership requirements with teeth
Make the social license explicit and enforceable. You want to build here? Prove how it helps here. With numbers. With contracts. With equity.
The Commonwealth of Intelligence
Here’s the vision I’m carrying into Sunday’s podcast, the one I want bouncing around in mining executives’ heads and teenagers’ TikTok feeds alike:
Canada becomes the first country to treat AI compute as a public utility. Not nationalized… that’s old thinking. But regulated, metered, and accessible like we do with electricity, water, and telecommunications. We build the Commonwealth of Intelligence: shared infrastructure, open governance, transparent operations, and equitable access.
Indigenous nations get dedicated compute allocations. Not charity, but recognition that digital sovereignty follows from territorial sovereignty. Small businesses get affordable GPU hours without selling equity to Silicon Valley. Researchers get the tools to compete globally while contributing locally. Artists and creators get the compute to imagine new worlds without corporate overlords.
The financing comes from where it always should: the value created. A small percentage of all commercial AI operations on Canadian infrastructure goes into the sovereign compute fund. Companies that want exclusive access pay premium rates that subsidize public access. Carbon credits from AI-optimized industrial processes flow back to expand clean energy.
The Moment (and the Math)
This moment won’t last. The compute gold rush is accelerating. Countries are waking up to what computational sovereignty means. The question is whether Canada will be a player or a power source, a partner or a peripheral, a leader or a logger.
We have everything we need: the energy, the minerals, the talent, the space, the stability, and… critically… the social infrastructure to do this right. What we’ve lacked is the imagination to see ourselves as anything other than hewers of wood and drawers of water. But code is the new wood, and compute is the new water, and maybe, just maybe, we’ve learned enough from history to hew and draw on our own terms this time.
Sunday’s podcast with Stewart is just one conversation, but these conversations matter. They’re how we rehearse the future before we build it. They’re how we resist the narrative that says Canada’s role is to provide raw materials for other people’s dreams.
I’ll be recording it myself and posting the full version, because transparency isn’t just for data centers. It’s for all of us trying to navigate between prosperity and principle.
The protesters who block pipelines, the entrepreneurs who build startups, the Indigenous leaders who assert sovereignty, the workers who demand dignity… they’re all part of the same resistance. The resistance that says: not this time. This time we build it right. This time we keep the value. This time we write the rules.
Rocks into racks into rights. That’s not a slogan. That’s a blueprint. And Canada… skeptical, cautious, pragmatic Canada… might be the only country wise enough to build it.
The intelligence explosion is coming whether we’re ready or not. The only question is whether we’ll power it on our terms or theirs. I know which future I’m choosing. Sunday, we find out if Stewart and his audience are ready to choose it too.
Because if we don’t wire intelligence to integrity now, we’ll spend the next century wondering why we gave away the keys to the kingdom for the promise of a few smart thermostats and slightly better search results. Canada can be where humanity figures out how to have its intelligence explosion without an energy implosion, its technological revolution without a social devolution.
We just have to decide we’re worth it. And then build like we believe it.
But let’s build with our eyes open. With methane monitors. With public registries. With real consent. With actual outcomes, not promised ones.
I’m not Team Fossil or Team Fantasy. I’m Team Outcome.
If AI wants gigawatts, it owes gigaton outcomes. Let’s measure that, not vibes.
The future needs power. Power needs purpose. Canada has both… if we’re brave enough to claim them and honest enough to verify them.