Chris Waddell: Data centres have become an increasingly interesting topic in Canada. What’s going to happen to them?
In the U.S. there is so much concern about these giant installations that 18 states have either banned them altogether or put moratoriums on new ones. Another 16 states are discussing various issues about the centres.
Here in Canada, the city council in Hamilton, Ontario talked about putting a moratorium on them, but last month the vote to do that was defeated 10-6. Mississauga is talking about it. On August 2 there was a demonstration in north Edmonton, where Meta plans to put a large data centre.
When I proposed this topic for our Roundtable I asked a few questions. I found that people are concerned about water and electricity use by data centres, and what that might mean for the cost and availability of electricity. This is so particularly in smaller communities, because some data centres are being located outside major cities, since these installations need a lot of space.
Some people may even be concerned about the fact that most data centres seem to be built with borrowed money, so if we get into any sort of debt crisis, will we have partially-finished or empty data centres around in various locations, as a blight on smaller communities?
So I want to ask this: what policy should we have, if any, on data centres, and what is the impact if we decide we don’t build them in Canada?
Peter Nicholson sent a very interesting note with a bunch of thoughts about the topic, and he concluded that the backlash is far more about artificial intelligence than it actually is about data centres. I’d like him to talk about that. His conclusion may well be correct, but at the same time I’ll ask: does that answer the question of what we should do about these places, as a public policy issue? Peter, go ahead.
Peter Nicholson: No, it obviously doesn’t answer that question. Look, data centres are not new. We’ve got thousands of them around the world. Your internet – in fact, this internet Roundtable conversation – would not be taking place were it not for a data centre; in fact, all kinds of data centres. We have them all over the country and all over the world, and they have not generated anything like the kind of backlash that we’re seeing now. Obviously, if you’re living right next door to one, the noise could be a problem. There’s no question that they consume land. Perhaps there was a time when the cooling water wasn’t recycled the way it routinely is now. There may have been a time when people worried about the electricity consumption, but I don’t think that’s going to be an issue today because most of the new data centres are bringing their own electricity with them.
Where I would be concerned about electricity is not with respect to grid rates, but that they do generate, if they’re gas-powered, a pretty humongous amount of carbon dioxide (CO₂). To put a number on it, a one-gigawatt gas-powered data centre, such as the Meta one under construction in Alberta, generates about four megatons of CO₂ per year. To put that in perspective, the oil sands in their entirety are generating about 80 or 90 megatons. So even one large data centre is a not-insignificant generator of greenhouse gases.
But oddly enough, that hasn’t been primarily the debate. The reason I say it’s about AI is because the arguments that are being made are almost manifestly countered in an objective way. Certainly with respect to water, these things are really clean. Land is a problem, but Canada’s got lots of it. So if you leave aside the greenhouse gases from gas-powered centres – and I’m not suggesting we should leave that aside – let’s just realize that otherwise they actually are a pretty benign industrial facility. So the opposition has got to come from some deeper source, which people are either not quite aware of or, for some reason, don’t like to articulate.
There’s one other relevant point, that the scale and speed of the buildout, and the size of the investment in these data facilities, are really humongous. In that sense, they aren’t entirely analogous to the point I started with, which is that we’ve had data centres for a long, long time. Somebody once said that quantity eventually becomes quality if you’ve got enough of it, and I think that’s, to some extent, the issue, not yet in Canada, but in the United States.
Chris Waddell: In Toronto, there was – I think it’s still there, but I’m not sure – a Royal Bank data centre on Front Street, almost beside what was the SkyDome, and is now the Rogers Centre. Of course, nobody put any identification on the building at that point because they didn’t want it being a potential attractor for someone who might want to disrupt the bank or something like that. So data centres have been around, but they tended to be fairly small, I think.
Peter Nicholson: When you drive into Montreal from the airport, you pass this building in Dorval, a Bell data centre, but dating from the telephone era. In other words, it’s a telecom switching centre. For years it didn’t have any insignia on it at all.
Chris Ragan: This discussion emerging now, about the costs and benefits, puzzles me because we seem to be talking about data centres as if they are somehow very different from a steel mill, a pig farm, an Amazon fulfillment centre, or any other big structure.
Peter’s probably right that opposition to AI data centres is more deeply about opposition to AI. In terms of their water or electricity usage, I would hope that if we are pricing those things appropriately, then we wouldn’t worry too much about it, right? If they are actually paying the marginal cost of electricity or the marginal cost of water, then I don’t think we should worry about them any more than we would worry about anything else.
If they have negative externalities associated with them, we should worry about that. So when Peter mentions the carbon emissions, if they’re unpriced – and they may not be these days – we should worry about that. But I’m equally puzzled by the people who seem to think that there are huge benefits from AI centres. So here, my ignorance of AI will be revealed, but it just seems to me that if you want to make a pitch that Canada has to be using AI and we can’t afford to get behind the curve on AI, I think that may be right. But I don’t think using AI requires that you have AI data centres in Canada. Maybe it does, but it just seems to me that if you’re using AI, can’t those servers be located in some other place? They could be in Timbuktu, Qatar, South Africa, or wherever. So are there some other benefits from having an AI data centre down the road, or in the middle of a field 100 miles away? Are there some benefits to the supply chain, for example, that we really care about? Maybe there’s a whole bunch of engineering or planning positions, or high-tech IT positions. So maybe it’s the supply chain of an AI data centre that we really like. I don’t know. I’d like somebody to explain to me what the super benefit is of having an AI data centre in our own backyard, because I just don’t get it.
David Dodge: One of the things that one worries about is that our data – your data and my data – is residing on a server somewhere outside the country, and that whoever owns that centre, or the government of the country in which it is located, can then say “A, we’re going to steal your data, and B, we’re going to deny you access to it in the future”. So I think there is a real reason to be concerned about having our own data centres just as much as we are about having our own steel mills or anything else, because it gives you a degree of insulation from the actions of others outside the country. That is not totally unreasonable; it would be useful to have them.
Secondly, consider that there is the Alberta Machine Intelligence Institute, in Edmonton, that is one of the places on the frontier of building new AI models. So if we are to have some of the part of the AI industry where we’re building new AI models, then maybe it is advantageous to have the data centre close at hand. So I don’t think it’s totally unreasonable to think about having data centres nationally. And if you’re in Alberta, if you want to expand out from what you have a lot of, which is natural gas, and help build something that isn’t just shipping natural gas out to other people, but rather burning it yourself, with perhaps an economic spinoff benefit in a totally different domain, that seems to be a pretty reasonable thing to look at.
Chris Ragan: So, if the argument, or part of the argument, is data security, which sounds reasonable, is that a big enough argument that we should say not only should we like the centres domestically, but that maybe we should be subsidizing them in some way, domestically, if we’re getting some sort of national-security or data-security benefit? If that’s something the market’s not going to do all on its own, how do you feel about that?
David Dodge: I don’t think you have to subsidize it. They will build their own. Bell built its own data centres, and for a long time we thought it was a really pretty good idea to have Bell switches under control in the local community, and not elsewhere. I just don’t know how much value, subjectively, to put on the security. I think there is value. I just don’t know how much to put on it.
Chris Waddell: Some of what David’s talking about is an extension of what happened after the U.S. Patriot Act of 2001, I think. We did get concerned about our data being kept on servers in the U.S. and U.S. officials being able to use it, whether it’s data on your health, data on people’s travel, or whatever it might be.
The other issue that the U.S. is the source of most, if not all, of the primary AI software being used in these sorts of ventures. And some people raise as a concern the fear that if you don’t have your own centres, the United States could decide to cut you off, and then you’d be left on your own. We saw a little bit of that with what they tried to do with Anthropic at one point.
I don’t know how realistic that is, but it’s an argument that some people make: that you need your own to be competitive in the world in the future, with so much depending on AI. Again, these are all debatable points, I guess. If you don’t have your own standard, you’re the victim. You’re potentially at the will of whoever actually does, which is largely the United States.
Miville Tremblay: I want to add to the personal-information concerns. I’m not so sure what people would do with my medical file personally, but there are other, more strategic assets. For example Polytechnique Montréal, the engineering university in Montreal, has decided to store all its research data in Canada, not in the U.S. any more, because it feared spying by the Americans. So that’s a concern. But I think the big misunderstood point is the effect of the U.S. CLOUD Act of 2018, which says that wherever in the world a data centre owned or operated by a U.S. corporation is located, the U.S. government can ask Microsoft, Google, Amazon or others to retrieve a file it wants on someone.
That doesn’t mean the U.S. government couldn’t do that by itself. If it really wanted to spy on you, I guess it could break through all kinds of firewalls. But I think the fact that this law exists is an issue of concern.
Dale Eisler: I get the data-sovereignty argument and the importance of that, but if we could just set that aside, my purpose is to understand the economic implications. If we were not to move on these data centres – not to build as big as the demand might be for them right at this moment – if we said, no, we’re not going to have them in Canada, at least not to this scale, that means we would be dependent on other data centres elsewhere in the world. Would that somehow limit our access as a nation to AI, setting aside the sovereignty issue? If we concede that we don’t have control of that, would it inhibit our access to AI in Canada because we don’t have these domestic data centres? That’s what I don’t understand. Aside from that data-security thing, what are the economic implications of not doing this?
Chris Waddell: Yes, that’s the argument some people make, at least. I’ve heard it. I don’t know if anyone listens to Rory Stewart on the British podcast The Rest Is Politics, but he certainly makes the argument that you have to have your own, or you’re going to be cut out of the use of some of the AI tools that you want to use otherwise.
Peter Nicholson: If you do have your own data centre, you can at least have a resident foundation model on it with the ability to actually respond to local queries. And that will be perfectly adequate for a lot of purposes. It is still vulnerable, of course, to being cut off in the sense that you don’t get the updates of the model and you could be cut off from the global network. That global network, I think, is going to become more and more important in the application areas as AIs become more and more systems of agents that are gathering information from data sources, other data centres and other AI systems all over the world.
So it is important, for resilience purposes, to be sure that we have, in the event of an emergency, enough data-centre capacity with the existing models already loaded so that you can continue to function for most purposes. That’s the first point.
Second, I definitely think that the government – let’s say the military and the security services—probably needs to have its own dedicated system. So in that sense, Chris, that would be a subsidy, in the sense that the government is buying the service directly.
With respect to sheer geography, there are certain applications that depend on low latency, and so having a data centre over in the UAE or Australia is not practical for some applications. It is for many others. For certain applications there is a geographic-distance factor, but I wouldn’t put that high on the list because generally you can get close enough to a U.S. data centre.
With respect to data sovereignty – and this I don’t know, but it would be really interesting to know – you may have data stored on a U.S. server physically resident there, but if you have the encryption key, there’s no way that the American source, even with a government order, can decode that. So an issue here is to what extent effective encryption, and the legal structures surrounding it, obviate this argument about data residency.
I would say, Dale, in answer to your question, that increasingly Canada and every other country is going to have to have some resident capacity for resilience and national security, probably for individual critical businesses and critical infrastructure – the Bank of Canada, the banking system, utilities, etc. You just cannot afford, whether for political reasons or because of some kind of kinetic interruption, to be rendered blind in a world that’s going to become more and more AI-dependent.
Chris Waddell: Picking up on that, what should our politicians and political system do in response? It seems to be, for some people, less about whether we actually do these or build these or don’t build these. Some people have suggested that data centres should have to rely only on renewable energy, as that might make a little bit of difference in terms of the electricity they use. Are there things anyone thinks we should be regulating, or debating whether we should regulate, that would address people’s concerns?
Miville Tremblay: We said at the beginning, and I guess we should go back to this, that the real issue is AI itself. It’s not really the centres, although there is a minimum of regulation to build something big anywhere. So that’s not where my concern is. The backlash that is fairly strong in the U.S. and rising in Canada leaves me with contrary feelings. On one hand, I understand that this is a major technological development and that we can’t do without it. I use it more and more in my work. So I think there are benefits, even if they are further down the road.
But so far, there is a terrible lack of governance of that industry, and I truly mistrust the bunch of people who run it in Silicon Valley. I don’t know if you’ve heard the outlandish interview with Elon Musk by The Economist last month. Well, do you trust this guy to run something this important? He’ll be hidden away on Mars.
So I think the Americans are paying the price for their unwillingness to consider serious governance. My hope is that the objections will stall them enough and make them lose enough money that they switch. But then again, it’s not just a question of regulating or not regulating. I think it’s tied up, like everything else in the U.S., with the highly polarized environment of extremely right-wing discourse and views, and increasingly also hard-left views – relatively speaking, for the U.S., of course. I think it runs deeper, and part of it is clearly a backlash against the “tech bros” in the U.S., the fact that they are making amazing fortunes while other people have difficulties going through their own lives, feel that the cost of living is too high, and fear for their jobs. So I think it’s deeply political. The issue is not technology; the data centres are just a pretext.
Chris Waddell: Remember, when social media started out, we didn’t realize what could potentially happen with it. We didn’t regulate it, for instance. We didn’t apply the same standards to social media that we apply to existing news organizations when you’re doing news or spreading news. And now you can’t regulate it; we’re seeing people run into problems when they try to do that. And we’re now seeing some of the less positive things about social media. Could it be that on this issue some people, at least, are thinking we shouldn’t get fooled again, because we don’t know what’s going to come out of AI? And as Miville says the people involved have not turned out to be necessarily the most trustworthy people around.
Dale Eisler: Somebody mentioned how rapidly this is all happening in terms of these data centres. We’ve had them in the past, but they had a very low profile, and suddenly these big, massive centres are just popping up all over. There’s a local example here in Saskatchewan, where Bell is building, I think, a 300-megawatt data centre. It’s in a rural municipality right on the edge of the city, and the first broad public notice of this was when the municipal council approved the development. Right around the same time, three councillors plus the reeve resigned with no explanation. This centre is a huge development, supposed to generate $12 billion for the economy, and there was no public process leading up to this decision point that I was aware of, or it certainly wasn’t very visible if there was. Have any of you seen the same kind of rapidity in terms of how quickly these projects get announced without any due diligence leading up to the decision point?
Peter Nicholson: That’s a great point, Dale. And the answer is yes. It’s been rampant in the U.S., where approvals have been given, let’s say in Texas, in a matter of a week. And many of these deals are subject to non-disclosure agreements between the data-centre sponsors and the municipal governments. Miville, you mentioned to me the fantastic Ezra Klein podcast episode with Jasmine Sun. I recommend that to everybody. This woman spent the last several months touring around Michigan and Wisconsin, talking to everybody across the whole spectrum. She gives a fabulous account of both the Silicon Valley view and the Main Street view of this problem.
These non-disclosure agreements were a real flashpoint, until now the companies themselves have realized this was stupid on their part, and at least in the U.S. I don’t think they’re a feature anymore. But one other thing – Chris, this is to echo what you said – I think the experience with social media has certainly, if not soured people, raised a cautionary note for so many people, particularly in North America and probably Western Europe. It seems to be less so in Asia.
The other factor that really plays against the data centres – and AI more generally – is that the benefits are still very abstract to most people. Yes, they can play with their chatbot, but they don’t see the really big benefit that would justify almost any cost at all.
So in cost-benefit terms, it’s pretty easy to convince yourself rationally that this should be opposed. It’s quite different than if you’re building, let’s say, an electricity-generating plant, telephone transmission lines, or even a solar farm. In those cases you can see the benefit pretty immediately, both in space and time. But we’re not going to see the kind of benefits from AI that are driving this humongous amount of investment, for quite some time yet. So that represents a very abstract notion, and it’s treated more like a toy, I think, and even a potentially dangerous toy, sort of like Chucky in the horror movies. Under those circumstances the politics really become difficult. How can you bring forward the benefit in time to justify whatever disruption there is in putting in place the infrastructure that will enable that benefit to be realized eventually? That’s the political-economy problem, it seems to me.
Chris Waddell: It reminds me a little bit of the free-trade debate in the late 1980s, when it was easy to point to all the problems and all the people who would lose their jobs, but it was much more difficult to point to what jobs would be created and what opportunities might be developed. We’ve got a little bit of that again.
David Dodge: Yes, and even if people generally can see the benefit, it’s very hard to see the benefit to your own local community. It’s not like any other one of these other things. So people in Port Hope really don’t want to see another nuclear plant sitting in Port Hope, although we can all see the benefit of having more electricity in the province. So there’s a question of whether there’s something local that you can really put your finger on, as opposed to something more abstract.
Miville Tremblay: The benefits that are already here are diffuse and are in daily use in services that have marginally improved. But the big promise, of a cure for cancer for example, is way down the road, if it’s there at all, because normally you’ve got to find new molecules. Maybe this is now faster, but you still have to test them the old-fashioned way with rats, and then humans, and do all kinds of studies that take years and years. So the benefits are long-term, and the worries are short-term.
There’s also a qualitative issue, at least if you try to follow that industry as I try to, in that it moves so fast that about every two weeks you get a new generation of models and new consequences that you hadn’t thought of before. It runs far faster than anyone can follow it, and so it creates all kinds of natural insecurity.
I’m coming back to the boys in Silicon Valley who run this. They are in a rat race, presumably with China, although I’m not so sure that this is real. China is trying to move fast. They were a few months behind. They are probably a few weeks behind. Would it make a big difference to be three months behind or ahead? I’ve got doubts about this. China will get there anyhow, and there may be a period of time when they are better than the Americans, and then the other way around. So I’m not so sure whether it is that material.
I think what the proponents are saying is about the famous “singularity”, which has many meanings depending on whom you listen to or read. But one of those is the capacity for the models to improve themselves on their own, which could bring them onto an exponential curve where you can lose control, or where it’s harder to catch up.
Chris Waddell: Miville’s comment opens the door to something that Ed Greenspon did. Ed asked ChatGPT to compare what the United States is doing with data centres and what China is doing with them, and ChatGPT said, that first of all, China has been forced to do more with less computing power because the U.S. has prevented China from getting the most recent chips, China has been forced to focus on drastically reducing the amount of computing power needed to produce an answer, and Chinese companies, ChatGPT says, are very good at optimization. If you’ve got the best chips, you can solve your problem by throwing more computing power at it. But China has had to try to optimize what its less effective chips can actually do.
And China has a major electricity advantage. It has enormous generating capacity through traditional generation and renewables. It can build large centres where power availability is not a problem. The U.S. problem is that centres are being built faster than grids and generation capacity can be expanded. Chinese models appear to be more efficient relative to U.S. models, but that might not mean they use less electricity because, if they’re more efficient, it may encourage more activity.
ChatGPT says the U.S. has better chips, enormous data centres, a huge electricity supply, and increasingly large models, whereas China has less powerful chips, cleverer algorithms, greater efficiency and comparable capabilities at a lower cost. So ChatGPT says the battle between China and the U.S. is a race between semiconductors and electricity. How much electricity do DeepSeek, ChatGPT, and Gemini use per query, and what does that mean for the U.S. power grid versus China’s?
I wonder, though, if part of the problem in Canada picks up a little bit on the U.S. issue about our grids and generation capacity being able to adapt to these large centres, if they’re built in Canada, and how quickly that can happen. That, I think, may be the concern some people have about electricity.
Peter Nicholson: It shouldn’t matter, if the data centre brings its own power, which is what the Meta one is doing. That’s increasingly what’s happening in the U.S. If we’ve got surplus capacity in the grid, great, use it. But we don’t, because we haven’t expanded generating capacity and grid capacity much in Canada in the last 40 or 50 years. So I don’t think electricity is going to be the gating problem in that regard. But the point that you just made, I think, has a really interesting counterpoint to what Miville just said, which is that the U.S. strategy here, at least among the major model companies – the so-called hyperscalers – is to maximize the leading edge of capability. That follows from a belief, whether it’s right or wrong, that there really is a sort of winner-take-all dynamic here: that if you achieve what’s called “recursive self-improvement” (RSI), where a system can begin to improve itself, or even artificial general intelligence, two or three months ahead of the Chinese or your commercial competitor, then they will never catch up. There will be an exponential explosion.
I’m not convinced by that argument by any means, but I think it is what they believe. The Chinese philosophy is basically to deliver the capacity as cheaply as possible and to focus on the more rapid diffusion of use to individuals, businesses, etc., throughout the economy. I think, with respect to a technology like AI, the more users you get, the better your systems are going to get, because they are going to learn from that process.
So the Chinese, it seems to me, are locked into the right recursive loop here. And yes, the U.S. may conceivably achieve RSI. But if the U.S. does not get there, I think the Chinese strategy is going to be the winning one. That has to do with the very particular characteristics of AI as a technology.
Miville Tremblay: One thing we haven’t mentioned, which I think is relevant, is this: currently, it seems that the biggest use is to train the new models, and these could be far away because speed is not as important as proximity and latency. Whereas inference – the actual daily use – requires much less. Clearly, Canada is not involved in developing the leading, cutting-edge models, but is focused more, in a Chinese way, on practical use for businesses. That’s what I hear and see.
Chris Waddell: Let me ask one last question. This is an issue that does seem to be catching on among the general public. So what should our political classes do at the federal, provincial, or municipal level? What would you recommend as their best strategy in response to what seems to be a public – even if it’s not a particularly well-informed public – demand for a response, action, for something to happen?
David Dodge: We talked about the issue that you can’t be covert about what you’re doing. So, to me, the municipality opens up the debate. There are pros and cons, just like there are pros and cons for the municipality of an Amazon distribution centre, a generating plant, or whatever. So it seems to me that the issue is local in that regard, and I think you have to allow the local municipality to argue it out. As long as it’s open and disclosed, it seems to me that that’s all you can really ask.
Chris Waddell: But also, David, with a generating centre or even an Amazon centre, it’s possible to calculate more directly what the cost to the community might be, whether it’s the number of trucks that are running in and out or that sort of thing. But for a data centre, it’s a little more nebulous, which may actually make it a little more difficult.
Chris Ragan: But you also want to make those arguments. The local arguments, my guess is, will mostly be in opposition, and the benefits are going to be at the national, aggregate level, especially if they’re security-type benefits, right? So there is going to be an onus on the proponents of these data centres to make those local opponents feel that there is a bigger picture here. There’s a national goal here, and we need these data centres for the following reasons, and they’ve got to go someplace. And so you can be part of this national project, exactly like nuclear-waste storage.
Chris Waddell: I was going to make that comparison myself, although nuclear-waste storage hasn’t had many winning sales pitches so far.
Chris Ragan: No, but in fact one actually did go through with a bunch of local support, in northwestern Ontario in 2024, once they realized that, okay, this has to happen someplace. Yes, there are pros and cons, but as David said, you can’t be covert about it. You’ve got to be clear about what you think the benefits are and why this is in the national interest, if indeed it is, because otherwise local NIMBYism is going to sink everything.
Peter Nicholson: First of all, if you’ve got strong local opposition, I think you’re just beating your head against a wall trying to convince people that it’s good for them. Fortunately, there will be lots of places in Canada that are more than happy to host a data centre, and the benefit for the municipal unit is often a hell of a lot of tax revenue, or even an upfront bribe, if you want to call it that. So I think you let it play out in those communities that already see a positive value proposition.
What I think you really shouldn’t do is impose any national or provincial hard stop. You simply impose whatever ordinary environmental and due-process rules of the game you would impose on any other industrial activity. But for heaven’s sake, don’t single out data centres for special adverse treatment. That would be my policy advice.
Dale Eisler: Yes, but we can’t forget – and I think this point was made at the outset – that this is less about data centres and more about AI and the threat of AI, right? So people are using data centres as the mechanism to block AI. They have their environmental concerns, yes, indeed, but underlying that is the threat of artificial intelligence. People are hearing about massive job losses and a threat to humanity in some regards, and that’s what’s driving the sentiment on this. That’s going to be the big challenge, to my mind. I think these other issues can be managed and dealt with, but when it’s driven by this much larger fear of what this is going to do to us, you’ve got a huge challenge for governments and politicians in how they manage it.
Peter Nicholson: And it’s a hell of a lot bigger challenge for the Americans than it is now for us. Look, my thesis depends, Dale, on there being communities in Canada willing, for their own economic interests, to say, ‘Yes, this is fine,’ and municipal councils that are quite prepared, depending on where they are, to put their political future on the line. I think you’ll find, over the next few years, more than enough opportunities. And once you see that at least the usual arguments against data centres are frankly not availing, then we are back to this more existential argument about AI itself. Maybe by that time some of the benefits will start to be more apparent.
Dale Eisler: Yeah, maybe we’ll be into the “world of abundance” as Musk talks about.
(Laughter)
Chris Waddell: The one issue I’d add to Peter’s point is, of course, that those communities have to be in the right geographic proximity to where the industry wants to be in order to have access, which may limit the number of people who are interested. I don’t know. We’ll have to see.
Peter Nicholson: There’s not too much of a latency problem. There would be between St. John’s and Victoria, right? But in most places, we’re talking a couple of milliseconds.
Chris Waddell: Thanks, everybody, for an interesting discussion, as always. We’ll see you in a couple of weeks, with another idea to talk about.
If you enjoyed this, you may also like from Sage:
A lot of this commentary seems to miss the point.
Water not a concern? In fact it is and the recycled loops do not tell the entire story.
Canadian security? How do you square that with the META installation that is Mark Zuckerberg's project. The same META that used Cambridge Analytica to (allegedly) mess with the Brexit vote. Some might have heard tale of the separatist referendum being forced by the Smith Govt on Albertans.
Another concern specifically, and Dodge must know this, is that META is starting up before its self supporting methane gas consuming energy power is built.
I appreciate the emphasis on emissions that some have noted; four megatons of CO₂ per year for META compared with oil sands generating about 80 or 90 megatons is not insignificant.
We got lots of land in Canada so siting these behemoths is a silly concern? Could you be more dismissive of rural Canada and all the life that lives there? Sounds a lot like he LOO swaggering around telling us how much 'dirt' Canada has.
Sorry - this convo misses, big time.