Everything Podcasts (00:17.807)
It takes a certain kind of ambition to do what farmers do. Between the hours and the hard labor, to the public scrutiny and bureaucratic maze running, it's not an easy task. While the agricultural industry feeds millions, quietly tilling and producing behind the scenes, many forget that our food comes from the hands of real people with real stories.
Join us as we share stories from those with Boots on the Ground and unearth unique perspectives on agriculture's biggest conversations. It's time to grab your shovel and get to work. I'm Don Schaefer, and this is Digging In with UFA.
Everything Podcasts (01:05.367)
Artificial intelligence is here, moving at exponential speeds through communities and industries all over the world, and changing the way we do almost everything. As polarizing as the arguments for or against have been in shaping and shaking industries across Canada and the world, agriculture is facing this rapid shift in more ways than one.
Farmers have always embraced helpful advancements in technology. Between precision agriculture, GPS guidance, drones, and AI-assisted crop management, many real opportunities for better process and product have been utilized.
Now, as news about artificial intelligence dominates headlines across Canada, Alberta is quickly becoming part of that story. Recently, plans were announced for a massive hyperscale AI data center to be built northeast of Edmonton. If completed, it would be the largest AI data center in Canada and among the largest in the world. To support the enormous energy demands of the facility, three companies are also proposing the Greenlight Electricity Center, a large natural gas-powered generating station instead of
Sturgeon County's industrial heartland. Together, these projects represent a significant investment in Alberta's growing role in the digital economy and are raising important questions about energy, infrastructure, and the future of rural communities.
Meanwhile, farmers and communities are debating questions about land use, electricity demand, water consumption, municipal planning, and who ultimately benefits from these investments. Successful completion of this data center expansion will no doubt bring economic benefits to the province, which is an attractive location for such developments because of its cooler climate, deregulated energy production opportunities, and lots of natural gas. But there are still some questions to be answered.
Everything Podcasts (02:56.571)
Do the cost of land and energy use outweigh the voracious appetite of technological advancements? Do the farmers and producers of Western Canada have a realistic expectation or concern when it comes to their land and resources? Can we bridge the polarization and come to a place of equity where all perspectives and fears are addressed? Today in Part 1 of a Necessary Conversation, we're digging in on AI.
Data centers and agriculture.
Everything Podcasts (03:30.779)
Though there are still very big conversations to be had about the rapid advancements in AI.
Knowledge is the key. Knowing the ins and outs of such a fast-paced industry calfts the space for realistic and nuanced conversations, despite the opinions held at either end of them. Joining us to talk about the realities for the future of Western Canada and the future of agriculture, this is David Gerhardt, head of computer science at the University of Manitoba, and Robert Sake, professional agrologist and CEO of the T1 Technology Corporation. My name is David Gerhardt.
I'm currently the head of computer science at the University of Manitoba. I'm an academic. I teach computer science. I've done so for 20, 25, a million years. I'm very interested in the impact of technology on society broadly. So I've been studying sort of how this tech, AI particularly, but all sorts of technology has been changing over time and the way that it affects like regular people in their day-to-day lives. A lot of academic work can be a little bit ivory tower, a little bit removed from reality, and so I try to sort of bring that back into reality whenever I can.
I've done a lot of conversations with media and public on artificial intelligence over the last few years because people have a lot of questions and I try to sort of build a balanced set of answers that give people the information they need to make good decisions, but also do some thinking as well. Name is Robert Sake, professional agrologist, certified ag consultant, been in the industry a long, long time, built agritrend group of companies, exited to Trimble, that was twenty years. CEO of a robotics company with Dot out of Regina exited to Raven.
That was 20 months, CEO of T1 Technology Corporation, which has a product called Visor Pro AI. We are active in the heavy equipment sector of agriculture, forestry, construction, municipal in Australia, the United States, and Canada. Wrote two books, 2014 and 2019, both with a chapter on AI and a member of A360 with Peter Diamondas. So introduced to AI pretty early on in the journey.
Everything Podcasts (05:36.261)
The economic impact of data centers is undeniable. But the environmental impact should also be considered. In the same innovative ways that have shepherded such a technological boom, are there similarly innovative approaches that reduce resource consumption? As with any conversation around AI, the answer is incredibly complex. But when the technology is advancing so quickly, it's important to slow down and ask these questions.
So I can give a bit of background context. I mean, we've had data centers for a long time. The first data centers in Canada were in the 1960s. AI data centers are different, but not that different. They're a different kind of computer, they're a different kind of processing environment, but in general, it's the same idea. It's a big room, it's a box full of machines, full of computers, tied into power and network infrastructure that allow all of these AI applications to do their thing. The growth, recent growth is definitely quite rapid. Part of the challenge with the conversation of this is that the information that we use.
To assess the sort of environmental and energy and water impact comes from a lot of different sources and a lot of different directions. So the conversation around water, for example, some people will measure the water consumption of the power that drives the AI or the power that drives the data center as part of the water use of the data center. Where that isn't typically done in other industries, right? So the water that you evaporate to drive the turbines to generate the electricity, that is counted sometimes, and other times it isn't counted. It's a broad, difficult conversation. In general,
Yeah, power consumption is big, water consumption is big. There are ways to do it differently, there are ways to do it with less use of water, but it's definitely a scale that people are surprised at for sure. My view of AI data centers in Canada is very akin to the railway system in Canada. And can you imagine a Canada that resisted the development of the railways? The principle applies today, and really it's a question of whether Canada embraces or
Chooses to tap out. Just from the context of size, people think these things are really big. And to David's point, they are large, they're large boxes, but a world-class hyper-scaling data center is a hundred to two hundred acres in size for a really big one. We've got a hundred and sixty million acres of farmland in Canada. So, in terms of context,
Everything Podcasts (08:00.731)
When people start to think about and talk that these data centers are gonna overtake farmland, it's a pinprick and nothing compared to urban sprawl that takes a lot more farmland every year than these data centers would. So just from a contextual standpoint, we should realize that the data centers are foundational infrastructure and that Canada makes needs to make a decision whether it participates or not.
And secondly, from a footprint, now we can get into you know energy and water in a minute, but from a footprint standpoint, it's really not that big of a deal. No, I agree entirely that I
Did the math if you're looking at the 40-ish data centers proposed for Alberta, that's maybe 10 sections altogether in in a you know 10,000 sections or more, right? So the amount of land is is is tiny. But if it's like your neighbor's land that's right over there and you know that they've been, you know, doing doing farm and doing wheat and canary seed there for a million years, and then suddenly it's got a big box on it, like this happens in other industries as well. We see, you know, farmland, like you say, go away to urban sprawl, we see it it go away to other things. So
the question of whether it has to be on this land or whether it can be somewhere else. And this is the nice thing about data centers is they can't really be anywhere. As long as you can run the wires there, as long as you can get the power and the compute and the network transmission stuff there. They can be almost anywhere. But when an industry comes in and they want to build something, they'll find the land that's, you know, that's cheap, they'll find the land that's for sale, and they'll build it based on their needs as opposed to the broader needs of the province of the country.
David, I I I think there is something that we can talk about with respect to where these data centers are located. Class one and class two agricultural farmland is very rare. Canada's blessed with a lot of it. And you don't necessarily need to put data centers on class one, class two farmland. So I think there could be some debate around where these things are located, but again.
Everything Podcasts (10:06.149)
People getting freaked out. I just went past a greenhouse that wasn't there five years ago, and it's basically a full quarter section of land. And that's a lot of light, a lot of energy utilization. Yes, it's agriculture, but if you live next door to that greenhouse, that's an impact as well. Like any major technological shift, the rise of artificial intelligence is generating both excitement and curiosity. As new AI projects take shape.
Many people are interested in understanding what this could mean for jobs, energy demand, economic growth, and everyday life. Having informed conversations about these opportunities and considerations will be an important part of navigating this rapidly evolving landscape. So so for me, there's sort of two the the fear comes in two directions. One is resource consumption, right? Like water use, power use. This is something that we've talked about a lot. Many of these sites need to sort of bring their own power with them.
as opposed to tie into the grid, which is unfortunate because Canada's grid is quite clean. Pulling hydropower from the north would be much better than burning natural gas right beside the power plant. But water is also the other thing that we talk about an awful lot. And many of the studies that are looking at water, again, like we talked about before, measure it in weird ways, right? they measure water consumption in a lot of everything surrounding the data center rather than the center itself. And if you look at, you know, at the data center in Louisiana for example, where first of all it's much hotter than here, right?
And the regulations are much more lax here. And so a company, if they wanted to, could plug into the municipal water supply and just spray water over the machines to cool it, that would be an enormous waste of water. And it would need a lot more than here in Canada. Here you would need less water. And also like closed loop cooling is a thing. We don't have to, you know, use up the water to cool it. So the water consumption debate is really frustrating because these things do use a lot of resources, but they don't use the level of resources that is sort of part of the conversation in the public.
Discourse these days because those studies are from different contexts, not from here. Yeah. So the water consumption demand keeps declining as the technology evolves. The other thing is why do we use water? Well, for cooling.
Everything Podcasts (12:28.614)
And when you consider where we should have data centers, you know, I would argue that Alberta, Manitoba, Saskatchewan is a heck of a lot more efficient for data centers than Arizona, where the amount of energy required to cool these data centers is through the roof. So again, from a standpoint of water, I think we need to put that into context. And I think that there are
citing concerns if water is a limited resource in an agricultural area, especially maybe if you're competing against water utilization for irrigation, then I think that's a discussion that requires a discussion regarding trade-offs. And also it requires adequate calculation to your point, David. Are you calculating the water use everything from the the beginning of the cycle or are you calculating the water use
Within the containment of the data center itself. So I think those things are important. It would be better, I think, if we were talking about cooling requirements rather than water use, because water the use of water in data centers is for cooling. And to your point before, like there's a data center in the basement of a building in the middle of downtown Winnipeg that heats the building half of the year, right? And so not only is the cooling requirements not as much as people say, but also it it it adds a benefit because that excess heat that gets generated.
Like half of the year we need excess heat in Canada, and so it's not as much of a drawback as it is considered to be. It's not nothing, right? But it's not as as severe of a concern as a lot of folks are making it out to be. But to me, the other sort of constraint is in terms of the benefit to the people that are that are there, right? So if the water consumption, yes or no, but will people get anything out of it, right? Like if if a data center is built and then the economic revenue of that goes to some centralized, multinational technological oligarchy.
That doesn't sit as well as if that investment can be brought back into the local economy and invested in the schools and hospitals that we need it, right? Data centers aside, the swell of artificial intelligence integration within most of the world's industry sectors has become relatively commonplace. Agriculture is no different. When you look at the projections of the impact of AI on different sectors of the economy, the impact of AI
Everything Podcasts (14:52.916)
To disintermediate large chunks of the agricultural sector are relatively small. AI would be akin to how big of an impact we'll have on construction, how big of an impact will have in agriculture is small compared to let's say white-collar accounting or legal firms or administrative functionality. Nonetheless, agriculture produces tremendous amounts of data. I was speaking at the Canadian
semiconductor conference in Vancouver, wondering why I was there. But the reality is that agriculture produces a lot of data, most of which goes unused. And so my contention is that AI will help to aggregate data from farms to help provide perspective on forecasting what might happen and then layering
That farmer or agronomist judgment over top of that data to complete the decision-making circle. So I think there's two sides to the impact of agriculture with AI. One is right now we literally have a proliferation of data that's not being utilized. The second piece is when you consider precision agriculture, remote sensing, IoT, Internet of Things, when you consider
Grain marketing strategies, financial fintech, when you consider the possibility of blockchain contracts to help a farmer in Saskatchewan blockchain that chickpea contract with an end user.
There's lots of opportunity for AI to impact agriculture. Again, I don't think we want to necessarily be sitting on the outside looking in on this. The federal artificial intelligence strategy that was announced last week, a couple weeks ago, lists a few key sectors that that the federal government wants to sort of focus on encouraging development in. And agriculture is one of them, right? Growing food and the especially the changes that's gonna happen to agriculture as as the climate shifts, allowing
Everything Podcasts (17:01.005)
Farmers to be more adaptive, allowing farmers to make these kind of changes as they need to. Again, to your point, always data informed, always with the human at the top of the loop, not just in the loop, but at the top of the loop, actually making those final decisions. In the university, in the academy, we have lots of people who are working on applications of AI to the agricultural sector in many different directions. We have this sort of terrible tendency to not have that end up in the hands of the people actually doing the work, right? There's a lot of big thinking. We could do this, we could do this, this would be great. And then nothing happens. And so finding a
Way to get those ideas, those innovations actually down to the people that are that are doing the actual work is gonna be a big part of it. But I'm consistently amazed at the kind of applications and extensions that people can find with this technology. So I think the actual impact is gonna be limited only by the kind of questions people can ask, the kind of data that they can collect, and the kind of reasoning that they can deliver to make change in almost any industry. I mean, these tools are risky and challenging and problematic.
But they're incredibly capable, which is why there's continual investment in this field, even though the public discourse is fraught and people have these concerns and all this negativity. The stuff is really, really cool. It works really, really well. When you think about Canada from an agricultural perspective, we feed 500 million people beyond our own population. We are an exporting nation. We have tremendous agricultural capacity. Even aggressive data-centered
Development doesn't really impact, like we talked about, doesn't impact the footprint, but it can by leveraging AI in agriculture, it can address some of those food security issues, especially when you consider, you know, whether you call it weather or climate change, supply chain durability, supply chain resilience. And the last one would be: are we going to invest in agriculture or not? And so I get concerned when you start.
To consider a binary argument around whether AI is good or not good for agriculture. It's a strong current, it's a wind that is blowing. And to take down our sales when the wind is favorable, I think would be a tremendous loss of opportunity for the country. So I don't think it has to be a binary yes or no decision. I think there's a way to enjoy this and still make good decisions on harnessing.
Everything Podcasts (19:27.598)
The power of AI for the country and and for agriculture in particular. Artificial intelligence has existed for decades. Its rapid appearance in everyday software, education, entertainment, and the workplace has made it feel as though it arrived overnight. Companies increasingly present AI not as an optional tool, but as an inevitable part of our future. That combination of extraordinary promise and growing uncertainty has left many people uneasy.
Questions about employment, creativity, authorship, and the concentration of technological power are no longer abstract. They're becoming part of everyday life. How can we maintain control of a technology whose development outpaces the speed of regulation? Are these fears warranted?
There was an Angus Reed survey a few weeks ago that asked Canadians how they feel about this very question. And most Canadians felt that there should be regulation on our artificial intelligence. And at the same time, most Canadians felt that it was very difficult for governments to appropriately regulate the technology because it was moving so quickly. This is the dichotomy we live in, is that is that people have concerns, and these concerns are legitimate, and we should be able to speak to those concerns, and the governments should be able to put frameworks in place, primarily so that companies, business,
Businesses, small and medium enterprises can invent new things that then have a market, right? If I build a thing with AI, it's gonna be great. Everything AI is wonderful, I'm gonna build a thing with AI. If I then take it to market and the public start to protest me because I used AI to build it, I don't have a market, right? Public confidence, public trust. Whether or not people are informed properly or whether or not we are talking about even the same thing, if the public doesn't trust this stuff, they're not gonna buy this stuff, and then I don't have a market to sell this stuff.
My perspective, being in agriculture my whole life, the terminology AI has been around a long time. It usually referred to in the dairy sectors and cattle as artificial insemination. You know, when you put it in context, what kind of AI are we talking about? But you know, in terms, David, in terms of framework, increasingly as I use AI, I'm really using it as augmented intelligence. I'm using it to augment my intelligence to make better decisions.
Everything Podcasts (21:43.36)
And when you talk about AI and you look at what's going on in the Frontier Labs, I mean, we just had OpenAI drop five point five, and then we have Opus from Claude coming out with you know four point six, four point seven, now four point eight. What Robert's referring to are iterative versions of developed large language models. OpenAI, the company that owns Chat GPT.
Released a GPT model called 5.5, which is its latest version. Anthropic, the company that owns Claude, has a flagship collection of LLMs called Opus with new iterations, essentially and simply newer versions of a product. This thing is moving very, very fast, and I'm not a big proponent of regulation from the standpoint of just creating bureaucracy.
But I I think that having some guardrails would be important. And I also think that with those guardrails and and some regulatory certainty comes more confidence in the investment community and perhaps more confidence in the citizens that are are seeing this impact and wondering what it means for their lives. So when we have a complete void of the discussion around the issues of AI.
And there's just kind of no direction, I think that's uncomfortable. I'd rather have some regulation, some direction. And this is coming from a person who's got a pretty red neck, but I'd rather have some regulation than in in this area, because it's moving very, very quickly. You can imagine highways with no regulation, and there you've got a problem.
The direction that that many of these jurisdictions are moving in is to sort of voluntary certifications, where a company, a Frontier Lab or or an AI company would would voluntarily say,
Everything Podcasts (23:40.942)
I am in compliance with this certification that says that I, you know, used the appropriate data and used the appropriate training and whatever it is that we're concerned about, if there were a certification that a company could put onto it, again, voluntarily, so this wouldn't necessarily be regulation more red tape, but to say that I understand, I as a developer understand the issues that people are concerned about and have attempted to address them in these ways, that can go a long way to build confidence. The widespread uncertainty, I think, is doing a lot of damage to the market. Like you look at anthropics.
For example, right? They released this massively powerful model called Mythos, which apparently can find cybersecurity holes in any software ever written. Anthropic also recently released Mythos, an AI frontier model recognized for its proficient execution of finding software vulnerabilities, launched as a model specifically for cybersecurity and biology research.
And then a few months later, right away they say, Okay, we're releasing this. but we're not releasing it because it's too dangerous. We're decided to keep it for ourselves. but we'll release it to some people, but not others. And then a few weeks later, the federal government identifies that as a what do they call it? Like a military they put a specification on it to prevent it from being exported because it was so valuable to the American economy. So how is a small business owner supposed to make any sort of strategy, any decision in terms of how I'm gonna proceed if
Maybe I have this model, maybe it gets taken away from me, maybe it's great, maybe it's not. It's very difficult to know how to build any kind of long-term plan for the use of these tools in this kind of uncertain environment, this anxiety-driven environment. I think it'll be really difficult, David, for governments to grapple with this given the speed and
Do we really want to regulate this thing to the point where another area such as China takes a lead on North American technology and advancement? So I'm glad I'm not in the shoes of the regulators right now, but I but I do think some would be good. I think your analogy to the highway is a great example, because like the frontier labs will often say like AI are the brakes on innovation. We don't want to put brakes on innovation. We want to keep moving, we want to, you know, open up the, you know, step on the accelerator and keep things moving.
Everything Podcasts (25:51.512)
My analogy is that regulations are the road markings. I'm happy to go as fast as I can as long as I know that the guy coming towards me is gonna stay on the other side of the yellow line. If there's no yellow line, who knows what's gonna happen, right? This is what we're missing. There is no yellow lines, there's no pavement, there's no nothing. We're just all driving towards each other at at high speed, hoping for the best. Which I think is why people get concerned. And the again, the public sentiment part of this I think can't be overstated because if you build an AI company, so you're building an AI company, when you market this to your customers, are you
Marketing this as like an AI company or as a heavy machinery company that happens to use AI in this particular whatever? The latter. And I think people need to understand that. That when we when we talk about ourselves in the marketplace, we call it Visor Pro AI. But really, we're a data company. And this is something that I think people need to understand as well. When you look into the future of farming, I'll use farms as an example. AI will eventually be the operating system of a farm.
Yeah.
The degree to which the AI will impact the farm positively or negatively will depend on the data. And the data comes into the farm in AI language, they refer to it as connectors. Are you going to connect your AI to your Google Gmail? Are you going to connect it to your calendar? Are you going to connect it to these other programs that you're using? On a farming side, we'll think about AI as an operating system with pipelines from that farm into the AI system. And some pipelines
would include weather stations, soil moisture probes. You could think about bin monitoring, bin sensors. You could think about hooking your farm to the markets like the Chicago Board of Trade. You could think about connecting your farming operation to the telematics inside of your equipment. And you know, there's so much agronomically, there's so much information that potentially could be harnessed that
Everything Podcasts (27:47.864)
Today we're really not. So I I think about the operating system as being the AI. I think about our company being more of a pipeline aggregator where we're bringing in information
From the original equipment manufacturers, we're bringing information from the legacy work order systems, we're bringing information from tribal knowledge at the dealership, we're bringing in information from the standard operating procedures and bringing all those together, then using AI to bring that information in and extract information out.
All of this data we've been discussing needs somewhere to live, which brings us back to the topic at hand. Data centers exist on farmlands, oftentimes in more rural communities, utilizing resources. And for the farmers and producers affected by these new data investments and the large scale propositions to come.
The realities rightfully instill fear. Given the speed and spread of data center necessity for AI development, the effects on the rural communities who share this land seem inevitable. I don't believe so. Getting back to the opening statements, you know, a data center, a large data center is a hundred to two hundred acres. I think that my question would be: should we allow that to happen on class one?
Or class, you know, two farmland. But you when I bring I I'm not gonna just say data centers, should we allow subdivisions on class one farmland? Should we allow highways? Should we allow solar farms? Should we allow wind turbines on class one?
Everything Podcasts (29:19.428)
Farmland. That's really a very value-driven question based on the finite resource of Class One farmland. Then again, David, if you're the owner of Class One farmland, somebody's coming along with a big check for a data center and you're on the edge of Fort Saskatchewan or you're in LaSalle, Manitoba, and
You know, somebody's offering that to you. Now, will that impact farmland in general? I I don't believe it will. Again, it is a relatively small fraction of the overall acreage that could be affected by a bid for a data center.
The other impact I think so I I agree a hundred percent. I think the total amount of land we're talking about is tiny compared to the overall amount of land. Even the overall amount of class one land is still fairly tiny. There is the potential for an impact of water sources, right? Overland water sources. If we find ourselves in a position where an AI company can build a data center and use whatever water they want from wherever they want, like I don't think we're that way. I think that our our regulatory structure would prevent a lot of these kind of concerns, or maybe should. But if, you know, if I build a data center next to a river and I divert the whole river,
Into the data center for cooling and it's all gone and nobody downstream gets anything for agriculture. That's also a potential impact on other stuff downstream. But but the same could be true for a pig operation or or or anything else, right? It's all shared resources compared to the individual chunk that is the thing that's happening on that land specifically. Yeah, well, whoever approved that development without adequate impact assessment should be held to account. And that's why regulatory framework works. That's why.
matters. The the answer to risk is accountability structures, not prohibition. We've got to be able to understand what the impacts are, but that understanding what the impacts are does not mean a no. Yeah, that's right. It means a reasonable conversation about risks, impacts, and then how those impacts are going to be mitigated. I want to go back to to farmland and farmland values have been actually appreciating strongly. It's one of the best asset classes in the last two decades. And
Everything Podcasts (31:27.45)
I I don't again anticipate that data centers are going to have a profound effect on farmland values. The interesting question, a yin and a yang, again, is are we over restricting what landowners can do with it? So you've got, you know, me arguing on one side of my mouth saying we should be careful about class one farmland. And the other one is should we be over regulating? A valid question is given Canada's large, large footprint.
Given the fact that cooling matters to AI data centers, can these data centers be located in more marginal or more out-of-the-way areas? We've got natural gas that can be pipelined into these areas. A lot of these areas could be in the north where there is more water, and a lot of these areas could be in colder climates. Just think what that would mean to the tax base for a town like Flynn Flon or
Thompson, Manitoba, or Melfort, Saskatchewan, or La Range even. So think about what that could be in terms of of an economic benefit. Forget about the negative. Think about what it could be for municipal taxation, country benefits, sit jobs for people. I I kind of think about that. With a topic as big as AI, it only makes sense to give it some room to breathe and keep the conversation going in two parts.
Stick around for the continued conversation next time while we continue to dig in on AI, data centers, and agriculture.
Everything Podcasts (33:15.756)
At the end of the day, we look to our resiliency within ourselves and our communities to make our collective futures as bright as possible. All it will take is conversations like these, some vulnerability, and a little digging.
Thanks to the support of UFA Cooperative, we're able to share stories from those who live and breathe agriculture. Thanks to Robert and David for giving us insight into the realities of data centers and a conversation around such a polarizing topic. For more information on today's episode, please visit UFA.com. With listeners like you, we'll continue to dig a little deeper. Here on Digging In with UFA. I'm Don Schaefer. Thanks for listening.
Another Everything Podcast production. Visit Everything Podcast dot com, a division of Patison Media. Subscribe wherever you get your podcast.
It takes a certain kind of ambition to do what farmers do. Between the hours and the hard labor, to the public scrutiny and bureaucratic maze running, it's not an easy task. While the agricultural industry feeds millions, quietly tilling and producing behind the scenes, many forget that our food comes from the hands of real people with real stories.
Join us as we share stories from those with Boots on the Ground and unearth unique perspectives on agriculture's biggest conversations. It's time to grab your shovel and get to work. I'm Don Schaefer, and this is Digging In with UFA.
Everything Podcasts (01:05.367)
Artificial intelligence is here, moving at exponential speeds through communities and industries all over the world, and changing the way we do almost everything. As polarizing as the arguments for or against have been in shaping and shaking industries across Canada and the world, agriculture is facing this rapid shift in more ways than one.
Farmers have always embraced helpful advancements in technology. Between precision agriculture, GPS guidance, drones, and AI-assisted crop management, many real opportunities for better process and product have been utilized.
Now, as news about artificial intelligence dominates headlines across Canada, Alberta is quickly becoming part of that story. Recently, plans were announced for a massive hyperscale AI data center to be built northeast of Edmonton. If completed, it would be the largest AI data center in Canada and among the largest in the world. To support the enormous energy demands of the facility, three companies are also proposing the Greenlight Electricity Center, a large natural gas-powered generating station instead of
Sturgeon County's industrial heartland. Together, these projects represent a significant investment in Alberta's growing role in the digital economy and are raising important questions about energy, infrastructure, and the future of rural communities.
Meanwhile, farmers and communities are debating questions about land use, electricity demand, water consumption, municipal planning, and who ultimately benefits from these investments. Successful completion of this data center expansion will no doubt bring economic benefits to the province, which is an attractive location for such developments because of its cooler climate, deregulated energy production opportunities, and lots of natural gas. But there are still some questions to be answered.
Everything Podcasts (02:56.571)
Do the cost of land and energy use outweigh the voracious appetite of technological advancements? Do the farmers and producers of Western Canada have a realistic expectation or concern when it comes to their land and resources? Can we bridge the polarization and come to a place of equity where all perspectives and fears are addressed? Today in Part 1 of a Necessary Conversation, we're digging in on AI.
Data centers and agriculture.
Everything Podcasts (03:30.779)
Though there are still very big conversations to be had about the rapid advancements in AI.
Knowledge is the key. Knowing the ins and outs of such a fast-paced industry calfts the space for realistic and nuanced conversations, despite the opinions held at either end of them. Joining us to talk about the realities for the future of Western Canada and the future of agriculture, this is David Gerhardt, head of computer science at the University of Manitoba, and Robert Sake, professional agrologist and CEO of the T1 Technology Corporation. My name is David Gerhardt.
I'm currently the head of computer science at the University of Manitoba. I'm an academic. I teach computer science. I've done so for 20, 25, a million years. I'm very interested in the impact of technology on society broadly. So I've been studying sort of how this tech, AI particularly, but all sorts of technology has been changing over time and the way that it affects like regular people in their day-to-day lives. A lot of academic work can be a little bit ivory tower, a little bit removed from reality, and so I try to sort of bring that back into reality whenever I can.
I've done a lot of conversations with media and public on artificial intelligence over the last few years because people have a lot of questions and I try to sort of build a balanced set of answers that give people the information they need to make good decisions, but also do some thinking as well. Name is Robert Sake, professional agrologist, certified ag consultant, been in the industry a long, long time, built agritrend group of companies, exited to Trimble, that was twenty years. CEO of a robotics company with Dot out of Regina exited to Raven.
That was 20 months, CEO of T1 Technology Corporation, which has a product called Visor Pro AI. We are active in the heavy equipment sector of agriculture, forestry, construction, municipal in Australia, the United States, and Canada. Wrote two books, 2014 and 2019, both with a chapter on AI and a member of A360 with Peter Diamondas. So introduced to AI pretty early on in the journey.
Everything Podcasts (05:36.261)
The economic impact of data centers is undeniable. But the environmental impact should also be considered. In the same innovative ways that have shepherded such a technological boom, are there similarly innovative approaches that reduce resource consumption? As with any conversation around AI, the answer is incredibly complex. But when the technology is advancing so quickly, it's important to slow down and ask these questions.
So I can give a bit of background context. I mean, we've had data centers for a long time. The first data centers in Canada were in the 1960s. AI data centers are different, but not that different. They're a different kind of computer, they're a different kind of processing environment, but in general, it's the same idea. It's a big room, it's a box full of machines, full of computers, tied into power and network infrastructure that allow all of these AI applications to do their thing. The growth, recent growth is definitely quite rapid. Part of the challenge with the conversation of this is that the information that we use.
To assess the sort of environmental and energy and water impact comes from a lot of different sources and a lot of different directions. So the conversation around water, for example, some people will measure the water consumption of the power that drives the AI or the power that drives the data center as part of the water use of the data center. Where that isn't typically done in other industries, right? So the water that you evaporate to drive the turbines to generate the electricity, that is counted sometimes, and other times it isn't counted. It's a broad, difficult conversation. In general,
Yeah, power consumption is big, water consumption is big. There are ways to do it differently, there are ways to do it with less use of water, but it's definitely a scale that people are surprised at for sure. My view of AI data centers in Canada is very akin to the railway system in Canada. And can you imagine a Canada that resisted the development of the railways? The principle applies today, and really it's a question of whether Canada embraces or
Chooses to tap out. Just from the context of size, people think these things are really big. And to David's point, they are large, they're large boxes, but a world-class hyper-scaling data center is a hundred to two hundred acres in size for a really big one. We've got a hundred and sixty million acres of farmland in Canada. So, in terms of context,
Everything Podcasts (08:00.731)
When people start to think about and talk that these data centers are gonna overtake farmland, it's a pinprick and nothing compared to urban sprawl that takes a lot more farmland every year than these data centers would. So just from a contextual standpoint, we should realize that the data centers are foundational infrastructure and that Canada makes needs to make a decision whether it participates or not.
And secondly, from a footprint, now we can get into you know energy and water in a minute, but from a footprint standpoint, it's really not that big of a deal. No, I agree entirely that I
Did the math if you're looking at the 40-ish data centers proposed for Alberta, that's maybe 10 sections altogether in in a you know 10,000 sections or more, right? So the amount of land is is is tiny. But if it's like your neighbor's land that's right over there and you know that they've been, you know, doing doing farm and doing wheat and canary seed there for a million years, and then suddenly it's got a big box on it, like this happens in other industries as well. We see, you know, farmland, like you say, go away to urban sprawl, we see it it go away to other things. So
the question of whether it has to be on this land or whether it can be somewhere else. And this is the nice thing about data centers is they can't really be anywhere. As long as you can run the wires there, as long as you can get the power and the compute and the network transmission stuff there. They can be almost anywhere. But when an industry comes in and they want to build something, they'll find the land that's, you know, that's cheap, they'll find the land that's for sale, and they'll build it based on their needs as opposed to the broader needs of the province of the country.
David, I I I think there is something that we can talk about with respect to where these data centers are located. Class one and class two agricultural farmland is very rare. Canada's blessed with a lot of it. And you don't necessarily need to put data centers on class one, class two farmland. So I think there could be some debate around where these things are located, but again.
Everything Podcasts (10:06.149)
People getting freaked out. I just went past a greenhouse that wasn't there five years ago, and it's basically a full quarter section of land. And that's a lot of light, a lot of energy utilization. Yes, it's agriculture, but if you live next door to that greenhouse, that's an impact as well. Like any major technological shift, the rise of artificial intelligence is generating both excitement and curiosity. As new AI projects take shape.
Many people are interested in understanding what this could mean for jobs, energy demand, economic growth, and everyday life. Having informed conversations about these opportunities and considerations will be an important part of navigating this rapidly evolving landscape. So so for me, there's sort of two the the fear comes in two directions. One is resource consumption, right? Like water use, power use. This is something that we've talked about a lot. Many of these sites need to sort of bring their own power with them.
as opposed to tie into the grid, which is unfortunate because Canada's grid is quite clean. Pulling hydropower from the north would be much better than burning natural gas right beside the power plant. But water is also the other thing that we talk about an awful lot. And many of the studies that are looking at water, again, like we talked about before, measure it in weird ways, right? they measure water consumption in a lot of everything surrounding the data center rather than the center itself. And if you look at, you know, at the data center in Louisiana for example, where first of all it's much hotter than here, right?
And the regulations are much more lax here. And so a company, if they wanted to, could plug into the municipal water supply and just spray water over the machines to cool it, that would be an enormous waste of water. And it would need a lot more than here in Canada. Here you would need less water. And also like closed loop cooling is a thing. We don't have to, you know, use up the water to cool it. So the water consumption debate is really frustrating because these things do use a lot of resources, but they don't use the level of resources that is sort of part of the conversation in the public.
Discourse these days because those studies are from different contexts, not from here. Yeah. So the water consumption demand keeps declining as the technology evolves. The other thing is why do we use water? Well, for cooling.
Everything Podcasts (12:28.614)
And when you consider where we should have data centers, you know, I would argue that Alberta, Manitoba, Saskatchewan is a heck of a lot more efficient for data centers than Arizona, where the amount of energy required to cool these data centers is through the roof. So again, from a standpoint of water, I think we need to put that into context. And I think that there are
citing concerns if water is a limited resource in an agricultural area, especially maybe if you're competing against water utilization for irrigation, then I think that's a discussion that requires a discussion regarding trade-offs. And also it requires adequate calculation to your point, David. Are you calculating the water use everything from the the beginning of the cycle or are you calculating the water use
Within the containment of the data center itself. So I think those things are important. It would be better, I think, if we were talking about cooling requirements rather than water use, because water the use of water in data centers is for cooling. And to your point before, like there's a data center in the basement of a building in the middle of downtown Winnipeg that heats the building half of the year, right? And so not only is the cooling requirements not as much as people say, but also it it it adds a benefit because that excess heat that gets generated.
Like half of the year we need excess heat in Canada, and so it's not as much of a drawback as it is considered to be. It's not nothing, right? But it's not as as severe of a concern as a lot of folks are making it out to be. But to me, the other sort of constraint is in terms of the benefit to the people that are that are there, right? So if the water consumption, yes or no, but will people get anything out of it, right? Like if if a data center is built and then the economic revenue of that goes to some centralized, multinational technological oligarchy.
That doesn't sit as well as if that investment can be brought back into the local economy and invested in the schools and hospitals that we need it, right? Data centers aside, the swell of artificial intelligence integration within most of the world's industry sectors has become relatively commonplace. Agriculture is no different. When you look at the projections of the impact of AI on different sectors of the economy, the impact of AI
Everything Podcasts (14:52.916)
To disintermediate large chunks of the agricultural sector are relatively small. AI would be akin to how big of an impact we'll have on construction, how big of an impact will have in agriculture is small compared to let's say white-collar accounting or legal firms or administrative functionality. Nonetheless, agriculture produces tremendous amounts of data. I was speaking at the Canadian
semiconductor conference in Vancouver, wondering why I was there. But the reality is that agriculture produces a lot of data, most of which goes unused. And so my contention is that AI will help to aggregate data from farms to help provide perspective on forecasting what might happen and then layering
That farmer or agronomist judgment over top of that data to complete the decision-making circle. So I think there's two sides to the impact of agriculture with AI. One is right now we literally have a proliferation of data that's not being utilized. The second piece is when you consider precision agriculture, remote sensing, IoT, Internet of Things, when you consider
Grain marketing strategies, financial fintech, when you consider the possibility of blockchain contracts to help a farmer in Saskatchewan blockchain that chickpea contract with an end user.
There's lots of opportunity for AI to impact agriculture. Again, I don't think we want to necessarily be sitting on the outside looking in on this. The federal artificial intelligence strategy that was announced last week, a couple weeks ago, lists a few key sectors that that the federal government wants to sort of focus on encouraging development in. And agriculture is one of them, right? Growing food and the especially the changes that's gonna happen to agriculture as as the climate shifts, allowing
Everything Podcasts (17:01.005)
Farmers to be more adaptive, allowing farmers to make these kind of changes as they need to. Again, to your point, always data informed, always with the human at the top of the loop, not just in the loop, but at the top of the loop, actually making those final decisions. In the university, in the academy, we have lots of people who are working on applications of AI to the agricultural sector in many different directions. We have this sort of terrible tendency to not have that end up in the hands of the people actually doing the work, right? There's a lot of big thinking. We could do this, we could do this, this would be great. And then nothing happens. And so finding a
Way to get those ideas, those innovations actually down to the people that are that are doing the actual work is gonna be a big part of it. But I'm consistently amazed at the kind of applications and extensions that people can find with this technology. So I think the actual impact is gonna be limited only by the kind of questions people can ask, the kind of data that they can collect, and the kind of reasoning that they can deliver to make change in almost any industry. I mean, these tools are risky and challenging and problematic.
But they're incredibly capable, which is why there's continual investment in this field, even though the public discourse is fraught and people have these concerns and all this negativity. The stuff is really, really cool. It works really, really well. When you think about Canada from an agricultural perspective, we feed 500 million people beyond our own population. We are an exporting nation. We have tremendous agricultural capacity. Even aggressive data-centered
Development doesn't really impact, like we talked about, doesn't impact the footprint, but it can by leveraging AI in agriculture, it can address some of those food security issues, especially when you consider, you know, whether you call it weather or climate change, supply chain durability, supply chain resilience. And the last one would be: are we going to invest in agriculture or not? And so I get concerned when you start.
To consider a binary argument around whether AI is good or not good for agriculture. It's a strong current, it's a wind that is blowing. And to take down our sales when the wind is favorable, I think would be a tremendous loss of opportunity for the country. So I don't think it has to be a binary yes or no decision. I think there's a way to enjoy this and still make good decisions on harnessing.
Everything Podcasts (19:27.598)
The power of AI for the country and and for agriculture in particular. Artificial intelligence has existed for decades. Its rapid appearance in everyday software, education, entertainment, and the workplace has made it feel as though it arrived overnight. Companies increasingly present AI not as an optional tool, but as an inevitable part of our future. That combination of extraordinary promise and growing uncertainty has left many people uneasy.
Questions about employment, creativity, authorship, and the concentration of technological power are no longer abstract. They're becoming part of everyday life. How can we maintain control of a technology whose development outpaces the speed of regulation? Are these fears warranted?
There was an Angus Reed survey a few weeks ago that asked Canadians how they feel about this very question. And most Canadians felt that there should be regulation on our artificial intelligence. And at the same time, most Canadians felt that it was very difficult for governments to appropriately regulate the technology because it was moving so quickly. This is the dichotomy we live in, is that is that people have concerns, and these concerns are legitimate, and we should be able to speak to those concerns, and the governments should be able to put frameworks in place, primarily so that companies, business,
Businesses, small and medium enterprises can invent new things that then have a market, right? If I build a thing with AI, it's gonna be great. Everything AI is wonderful, I'm gonna build a thing with AI. If I then take it to market and the public start to protest me because I used AI to build it, I don't have a market, right? Public confidence, public trust. Whether or not people are informed properly or whether or not we are talking about even the same thing, if the public doesn't trust this stuff, they're not gonna buy this stuff, and then I don't have a market to sell this stuff.
My perspective, being in agriculture my whole life, the terminology AI has been around a long time. It usually referred to in the dairy sectors and cattle as artificial insemination. You know, when you put it in context, what kind of AI are we talking about? But you know, in terms, David, in terms of framework, increasingly as I use AI, I'm really using it as augmented intelligence. I'm using it to augment my intelligence to make better decisions.
Everything Podcasts (21:43.36)
And when you talk about AI and you look at what's going on in the Frontier Labs, I mean, we just had OpenAI drop five point five, and then we have Opus from Claude coming out with you know four point six, four point seven, now four point eight. What Robert's referring to are iterative versions of developed large language models. OpenAI, the company that owns Chat GPT.
Released a GPT model called 5.5, which is its latest version. Anthropic, the company that owns Claude, has a flagship collection of LLMs called Opus with new iterations, essentially and simply newer versions of a product. This thing is moving very, very fast, and I'm not a big proponent of regulation from the standpoint of just creating bureaucracy.
But I I think that having some guardrails would be important. And I also think that with those guardrails and and some regulatory certainty comes more confidence in the investment community and perhaps more confidence in the citizens that are are seeing this impact and wondering what it means for their lives. So when we have a complete void of the discussion around the issues of AI.
And there's just kind of no direction, I think that's uncomfortable. I'd rather have some regulation, some direction. And this is coming from a person who's got a pretty red neck, but I'd rather have some regulation than in in this area, because it's moving very, very quickly. You can imagine highways with no regulation, and there you've got a problem.
The direction that that many of these jurisdictions are moving in is to sort of voluntary certifications, where a company, a Frontier Lab or or an AI company would would voluntarily say,
Everything Podcasts (23:40.942)
I am in compliance with this certification that says that I, you know, used the appropriate data and used the appropriate training and whatever it is that we're concerned about, if there were a certification that a company could put onto it, again, voluntarily, so this wouldn't necessarily be regulation more red tape, but to say that I understand, I as a developer understand the issues that people are concerned about and have attempted to address them in these ways, that can go a long way to build confidence. The widespread uncertainty, I think, is doing a lot of damage to the market. Like you look at anthropics.
For example, right? They released this massively powerful model called Mythos, which apparently can find cybersecurity holes in any software ever written. Anthropic also recently released Mythos, an AI frontier model recognized for its proficient execution of finding software vulnerabilities, launched as a model specifically for cybersecurity and biology research.
And then a few months later, right away they say, Okay, we're releasing this. but we're not releasing it because it's too dangerous. We're decided to keep it for ourselves. but we'll release it to some people, but not others. And then a few weeks later, the federal government identifies that as a what do they call it? Like a military they put a specification on it to prevent it from being exported because it was so valuable to the American economy. So how is a small business owner supposed to make any sort of strategy, any decision in terms of how I'm gonna proceed if
Maybe I have this model, maybe it gets taken away from me, maybe it's great, maybe it's not. It's very difficult to know how to build any kind of long-term plan for the use of these tools in this kind of uncertain environment, this anxiety-driven environment. I think it'll be really difficult, David, for governments to grapple with this given the speed and
Do we really want to regulate this thing to the point where another area such as China takes a lead on North American technology and advancement? So I'm glad I'm not in the shoes of the regulators right now, but I but I do think some would be good. I think your analogy to the highway is a great example, because like the frontier labs will often say like AI are the brakes on innovation. We don't want to put brakes on innovation. We want to keep moving, we want to, you know, open up the, you know, step on the accelerator and keep things moving.
Everything Podcasts (25:51.512)
My analogy is that regulations are the road markings. I'm happy to go as fast as I can as long as I know that the guy coming towards me is gonna stay on the other side of the yellow line. If there's no yellow line, who knows what's gonna happen, right? This is what we're missing. There is no yellow lines, there's no pavement, there's no nothing. We're just all driving towards each other at at high speed, hoping for the best. Which I think is why people get concerned. And the again, the public sentiment part of this I think can't be overstated because if you build an AI company, so you're building an AI company, when you market this to your customers, are you
Marketing this as like an AI company or as a heavy machinery company that happens to use AI in this particular whatever? The latter. And I think people need to understand that. That when we when we talk about ourselves in the marketplace, we call it Visor Pro AI. But really, we're a data company. And this is something that I think people need to understand as well. When you look into the future of farming, I'll use farms as an example. AI will eventually be the operating system of a farm.
Yeah.
The degree to which the AI will impact the farm positively or negatively will depend on the data. And the data comes into the farm in AI language, they refer to it as connectors. Are you going to connect your AI to your Google Gmail? Are you going to connect it to your calendar? Are you going to connect it to these other programs that you're using? On a farming side, we'll think about AI as an operating system with pipelines from that farm into the AI system. And some pipelines
would include weather stations, soil moisture probes. You could think about bin monitoring, bin sensors. You could think about hooking your farm to the markets like the Chicago Board of Trade. You could think about connecting your farming operation to the telematics inside of your equipment. And you know, there's so much agronomically, there's so much information that potentially could be harnessed that
Everything Podcasts (27:47.864)
Today we're really not. So I I think about the operating system as being the AI. I think about our company being more of a pipeline aggregator where we're bringing in information
From the original equipment manufacturers, we're bringing information from the legacy work order systems, we're bringing information from tribal knowledge at the dealership, we're bringing in information from the standard operating procedures and bringing all those together, then using AI to bring that information in and extract information out.
All of this data we've been discussing needs somewhere to live, which brings us back to the topic at hand. Data centers exist on farmlands, oftentimes in more rural communities, utilizing resources. And for the farmers and producers affected by these new data investments and the large scale propositions to come.
The realities rightfully instill fear. Given the speed and spread of data center necessity for AI development, the effects on the rural communities who share this land seem inevitable. I don't believe so. Getting back to the opening statements, you know, a data center, a large data center is a hundred to two hundred acres. I think that my question would be: should we allow that to happen on class one?
Or class, you know, two farmland. But you when I bring I I'm not gonna just say data centers, should we allow subdivisions on class one farmland? Should we allow highways? Should we allow solar farms? Should we allow wind turbines on class one?
Everything Podcasts (29:19.428)
Farmland. That's really a very value-driven question based on the finite resource of Class One farmland. Then again, David, if you're the owner of Class One farmland, somebody's coming along with a big check for a data center and you're on the edge of Fort Saskatchewan or you're in LaSalle, Manitoba, and
You know, somebody's offering that to you. Now, will that impact farmland in general? I I don't believe it will. Again, it is a relatively small fraction of the overall acreage that could be affected by a bid for a data center.
The other impact I think so I I agree a hundred percent. I think the total amount of land we're talking about is tiny compared to the overall amount of land. Even the overall amount of class one land is still fairly tiny. There is the potential for an impact of water sources, right? Overland water sources. If we find ourselves in a position where an AI company can build a data center and use whatever water they want from wherever they want, like I don't think we're that way. I think that our our regulatory structure would prevent a lot of these kind of concerns, or maybe should. But if, you know, if I build a data center next to a river and I divert the whole river,
Into the data center for cooling and it's all gone and nobody downstream gets anything for agriculture. That's also a potential impact on other stuff downstream. But but the same could be true for a pig operation or or or anything else, right? It's all shared resources compared to the individual chunk that is the thing that's happening on that land specifically. Yeah, well, whoever approved that development without adequate impact assessment should be held to account. And that's why regulatory framework works. That's why.
matters. The the answer to risk is accountability structures, not prohibition. We've got to be able to understand what the impacts are, but that understanding what the impacts are does not mean a no. Yeah, that's right. It means a reasonable conversation about risks, impacts, and then how those impacts are going to be mitigated. I want to go back to to farmland and farmland values have been actually appreciating strongly. It's one of the best asset classes in the last two decades. And
Everything Podcasts (31:27.45)
I I don't again anticipate that data centers are going to have a profound effect on farmland values. The interesting question, a yin and a yang, again, is are we over restricting what landowners can do with it? So you've got, you know, me arguing on one side of my mouth saying we should be careful about class one farmland. And the other one is should we be over regulating? A valid question is given Canada's large, large footprint.
Given the fact that cooling matters to AI data centers, can these data centers be located in more marginal or more out-of-the-way areas? We've got natural gas that can be pipelined into these areas. A lot of these areas could be in the north where there is more water, and a lot of these areas could be in colder climates. Just think what that would mean to the tax base for a town like Flynn Flon or
Thompson, Manitoba, or Melfort, Saskatchewan, or La Range even. So think about what that could be in terms of of an economic benefit. Forget about the negative. Think about what it could be for municipal taxation, country benefits, sit jobs for people. I I kind of think about that. With a topic as big as AI, it only makes sense to give it some room to breathe and keep the conversation going in two parts.
Stick around for the continued conversation next time while we continue to dig in on AI, data centers, and agriculture.
Everything Podcasts (33:15.756)
At the end of the day, we look to our resiliency within ourselves and our communities to make our collective futures as bright as possible. All it will take is conversations like these, some vulnerability, and a little digging.
Thanks to the support of UFA Cooperative, we're able to share stories from those who live and breathe agriculture. Thanks to Robert and David for giving us insight into the realities of data centers and a conversation around such a polarizing topic. For more information on today's episode, please visit UFA.com. With listeners like you, we'll continue to dig a little deeper. Here on Digging In with UFA. I'm Don Schaefer. Thanks for listening.
Another Everything Podcast production. Visit Everything Podcast dot com, a division of Patison Media. Subscribe wherever you get your podcast.
