AI's Energy Crisis: Why This Commodity Will 2x By 2040 | Edward Campbell
Summary
AI's Energy Demand: The rise of AI is causing a significant increase in electricity demand, with nuclear energy positioned as the primary solution to meet this need.
Nuclear Energy Investment: There is a projected 30% shortfall in uranium supply versus demand by 2040, highlighting investment opportunities in uranium mining and nuclear energy infrastructure.
Government Initiatives: The U.S. government is actively supporting nuclear energy development through tax credits and infrastructure projects, such as the proposal for AI data centers at Oak Ridge, Tennessee.
Private Sector Involvement: Major tech companies like Microsoft and Amazon are investing in nuclear energy projects, indicating a trend towards vertical integration and self-sufficiency in energy supply.
Challenges for Renewable Energy: Wind and solar energy face limitations due to intermittency and land use, making them less suitable for AI's continuous power needs compared to nuclear energy.
Investment Opportunities: Investors are encouraged to focus on uranium miners, nuclear utilities, and reactor tech startups, while being cautious of pure-play wind and solar companies without storage solutions.
Geopolitical Considerations: The shift towards nuclear energy is influenced by geopolitical factors, with countries like Canada and Australia poised to fill the gap left by reduced reliance on Russia and China for uranium.
Long-term Outlook: The nuclear renaissance driven by AI's energy demands presents a durable investment theme, with structural undersupply and policy support providing long-term tailwinds for the sector.
Transcript
AI has burst onto the scene. You have a step change in the demand for electricity. Right now there is uh a 30% shortfall in terms of uh uranium demand over uran uranium supply by uh 2040. >> Uh what will be the primary sources of energy that could power this additional demand? >> So it's mainly going to be nuclear. >> I'm joined today by Ed Campbell, founder of Red Hook Phoenix Investment Partners. We'll be talking about the future of energy and the commodities and resources needed for this AI transition. One of the biggest revolutions since the industrial revolution. What is the actual investment implication of this revolution? We'll talk about this with head. Edward, welcome to the show. Good to see you. >> Thank you, David. Thanks for uh thanks for having me. >> I'd like to start with this piece of news and get you to comment on this. This week, uh the Department of Energy has issued a proposal for AI data centers to be built at Oakidge, Tennessee. Now, this was the the site of apparently the founding of the original Manhattan project. They're calling this Manhattan Project 2.0. Uh this is the next step forward in the transformation from nuclear remediation site to nuclear renaissance hub, says the acting assistant secretary of environmental management. The department of energy is excited to partner with the best of American industry to bring this vision to reality. uh specifically the Oakidge reservation is one of the four sites identified by the department of AI uh department for AI infrastructure and generation projects on federal land. The DOE's office of science and um and office of environmental management are now seeking proposals from the US companies to potentially enter into one of one or more long-term leasing agreements at the site that will be solely funded by the applications. Um, I wonder how this fits into the $500 billion Stargate initiative that Trump launched earlier in the year and broadly how the government is um is planning to fund and build infrastructure for AI in our country. >> Yeah. Well, I think these are all uh examples of um the geopolitical imperative of um you know, us being um the major player in AI um energy sovereignty um well AI sovereignty uh equals energy sovereignty and you know the United States um has the biggest um installed base of uh of nuclear energy us along with Europe but we're really not keeping up with China um in terms of um building our base going forward. So AI is uh the next great general purpose technology um like the printing press, the steam engine, electricity or the internet. And um it's really more than that because it's a it's a meta technology that uh rewires how all existing technologies are used. So, it's it's really going to be the the biggest generalpurpose technology out there and that's going to create a tremendous amount of wealth and that wealth is going to be a critical component of national power. So, it's important for uh the United States to um keep its edge there. >> Infrastructure takes a lot of money, a lot of capital, a lot of time to build it out. Who's going to fund all these projects? Well, I think you know the vast majority of it is uh is going to be private capital, right? So, um I think that uh nuclear has a lot of good things going for it. Um if you think about it on on a on a legislative basis. So, a lot has changed since um the person in the White House has changed. If we go back to the Biden administration and the inflation reduction act, um there was really a lot of emphasis on wind and solar, um lots of subsidies. Uh currently green energy, especially wind, uh is going through a a bit of a default cycle here. One because the projects were uh maybe not economically feasible. Uh I know that here in New Jersey there were big uh plans for um offshore wind on the Jersey Shore which a lot of people weren't happy about. Well, those companies have gone bankrupt and there's a lot of stress in um in that industry um as the subsidies are pulled back. Meanwhile, the one big beautiful bill uh which pulled back those subsidies has created a lot of tax credits um for um for nuclear. Uh there was an executive order signed that has an aggressive plan for us uh building out nuclear plants along with tax credits and um so I think you know maybe some of the capital that would have gone to um wind and solar is now uh freed up for uh nuclear energy. So I I think it's going to be a combination of tax incentives as well as private capital. And another big thing with the um one big beautiful bill was um the expensing of depreciation back to uh 100%. So that uh certainly is uh something that uh incentivizes investment. >> I wonder if tech companies are going to become utility companies part-time as well. Take a look at this piece of news as well. This is coming from Microsoft. They're spending $1.6 billion to restart a nuclear reactor at 3M Island, which is the site of the worst nuclear accident in US history. was shut down in 2019. Um, yeah, they are signing a 20-year power purchase agreement last September. Um, and they've agreed to offtake the entire capacity of the plant's unit one reactor which has a capacity of 837 million watts. Uh, now uh yeah, they're calling the constellation is the uh the name of the project. What the constellation and Microsoft sorry um the company so um constellation acquired unit one of their plant in 1999 and they're partnering with Microsoft now what's what what does this signal to you that um tech companies are also acquiring and starting nuclear reactors? >> Yeah. So um the old model was hyperscalers uh bought from the grid like uh everyone else. Uh but the new model as you mentioned and as is mentioned in that article is uh vertical integration. So we have Microsoft backing the three- mile island restart. We have Amazon uh sitting data centers uh next to nuclear plants. We have Meta um engaging in a 20-year nuclear contract with uh Constellation. So, you know, I think here you have um SMRs turning electricity from a commodity into, you know, from an asset that you rent to one that you want to own or control. Uh when we think about um the bigger picture in terms of electricity consumption and and the long-term trends, well, if you think about the environment that we've been in for decades, grid planners were uh planning on declining load, right, associated with the decline in manufacturing. uh greater efficiency and now that AI has burst onto the scene you have a step change in the demand for electricity. So um you know for example let me cite some statistics here um you know data centers currently are 4% of the grid share as of 2024. Uh McKenzie estimates that they're going to be 12% by 2030. Now that that's just data centers, but if we um no, that's really the whole share of of data centers. But if you think about just the AI workloads alone, um they're currently about 3 to um 3 to 4% of of electricity. So when you get to 2030, um you know, you're going to run out of uh the grid is going to be completely clogged, right? So that's why uh you see a lot of these companies buying their own and locating them on on their facilities either as owners or uh with partners. Um basically you need a a behind the fence solution rather than um hooking up to the grid because the grid is going to be uh completely clogged. >> How much energy is needed by uh data centers in the future? Has anyone made well I'm presuming people have but what what have people made in terms of projections on how much energy is going to be needed in the next 5 to 10 years? Well, I don't I don't have the numbers for the next five to 10 years, but um you know, >> sure. In the future, yeah, >> data centers alone are going to be 12% of of of US electricity, right? Um and I don't have the the figures on on electricity by themselves, but as that becomes a bigger share, um you're going to need a lot more electricity than uh or on pace to be able to provide. >> I'll pull up some numbers later for the interview, but what does this mean for for commodities? uh what will be the primary sources of energy that could power this additional demand? So it's mainly going to be nuclear um because um you know it will involve gas um but uh nuclear has an advantage over gas uh in that uh it's carbon-f free right so there's still going still people who care about uh ESG as uh as investors so you know gas with carbon capture will play a role um but wind and solar are not suitable for uh the always on needs for AI. Um, wind and solar, they basically they can't scale AI. They're just too intermittent and uh too land intensive. Uh, for example, if you if you look at a density gap, right? In order to produce uh 100 megawws of um from a data center um for an SMR, you can do that on less than one acre of land. Whereas for solar that's going to take um 500 to 600 acres. Um for wind it's going to take 4,000 to 5,000 acres. So um both in terms of the land use and in terms of the intermittency, they're not simply they're simply not well suited for the AI age. um you need basically 24/7 reliable power and uh you basically can't uh wait for the sun up to come back up or wait for the wind to start blowing again. And even with improved battery storage um you know it's it's just not going to be able to scale in the same way nuclear can with the demands for AI. >> I just found something um as we were talking just now. So this is an article from Forbes. They're pushing uh AI is pushing the world toward an energy crisis. Catchy title. Uh good headline. At present, this accounts for 3%. So data centers and their associated transmissions account for 3% of global consumption emitting as much CO2 as Brazil. Now >> it's more than a lot of countries. Yes, >> it's yeah looking forward they're projecting or whoever has done the study is projecting that uh in the US alone the increase in power demand due to data centers is expected to rise from 200 terowatt hours as of 2022 to 260 in 2026 equivalent to 6% of all power use across the country expected demand expected to double by 2030 that's only in a couple years so um we've talked about nuclear energy what about the grid do we have the load at our current grid to service this kind of demand. >> Uh I don't have the numbers. Um you know, we're servicing it now. Uh fine, but the problem is um you know, as we scale and um electricity uh production is not just is not going to be able to keep up with uh electricity demand outside of a much bigger footprint for nuclear. >> So what's the thesis around uranium investing then? If uh nuclear is going to be the primary primary energy source uh in the future for servicing additional AI data center demands then uh the power source of nuclear reactors. Uranium certainly should be very high in demand. Uh what's your play there? >> Yeah. So you know we like we like the commodity uh we like uh uranium miners. Um there's basically you know right now there's uh a 30% shortfall in terms of uh uranium demand over uran uranium supply by uh 2040 that's uh that's going to double. Um I think that there's an opportunity uh for Canada. Uh Canada is poised to be uh a global energy powerhouse. Um basically uh so if you look at Camco and um their next generation Rook one that's uh set to be the world's largest uh uranium mine by uh 2030. And if you think about things in uh from a geopolitical sense, right? We're no longer in a world where efficiency and globalization is the main thing. We're breaking into a world of uh smaller blocks and um the um US aligned world is uh not going to be able to rely on Russia and uh China for Russia was a big player in in uranium. So uh places like Canada and Australia are going to have to uh fill that gap. So if we can't rely on um China and um Russia for critical minerals, uh places like Canada have a have a have an opportunity to uh to step up here. So you know in terms of who are you know who are the winners from uh an investment perspective it includes uranium miners uh it includes nuclear utilities and operators uh reactor tech startups um people who build SMRs uh new scale is is uh is one name that comes to mind if if you want to think about an ETF um there are thematic ETFs that focus on um both uranium as a commodity but also the the uranium miners. Uh the Sprout Uranium Trust is uh is one that comes to mind. And in terms of losers, uh I think we're looking at uh pure play wind and solar without storage. Fossil fuel generators um without uh a decarbonization pathway are going to have a tough time. and uh grid constrained utilities that are uh bypassed by um some of the PPAs that um you know nuclear operators are um able to generate. >> This is a chart from one of your reports uh the um >> AI nuclear nexus July report. So it's showing different countries from different parts of the world with their nuclear capacity right now. Red bar showing operating reactors, blue bar under construction and green bar planned. It looks like the US, if you add all the bars, is actually going to be behind China, assuming they actually go ahead with a planned nuclear uh construction. Uh what is RO by the way? >> Uh rest of the world. >> Oh, rest of the world. Okay. Um everyone else. >> Yeah. So I I mean as I was saying before you can see that the US and Europe has the biggest installed base but China is moving rapidly and um the you know Europe in particular had been going the other way in terms of uh phasing out their nuclear plants. uh now they're uh starting to reverse course but uh the US and and Europe are nowhere near moving at the same pace that uh China is and you know that risks us being left behind right so that's China as a as a uh you know uh command and control economy they can move quick um I guess this is you know one of the drawbacks of uh of a constitutional republic that uh you know you can't move as as quickly as uh the Chinese communist government can. >> You're right that winners deliver dense reliable carbon-f free power, losers do not. You also are right that pure play wind and solar developers without storage face limits as AIdriven demand prioritizes dense always on power. Uh so tell us why you're bearish on pure play wind and solar. Well, uh they're not well suited to uh provide for uh the demands of AI. You know, we already talked about uh the land density issue and the inter intermittency. Uh also uh public they're currently going through a difficult default cycle and uh the rug is being pulled out from under them in in terms of in terms of the subsidies. So um yeah there's just uh they're basically they will still play a role in decarbonization and decarbonization goals but uh I do see their role um you know becoming less and they're basically going to miss the AI mega trend. Let's talk about uh the next page where you disclose that you have a um uranium miners trade Euro ETF that's up about 70% as of June 2025. Uh you write that current primary production of uranium falls short of demand by roughly 30%. This is expected to get worse with demand growing outpacing supply. The shortfall is expected to double by to 60%. Uh by the time um planned nuclear facilities become operational 2040ish. And so, uh, the thesis for uranium miners now, um, are they are they, um, are they just a leverage play on uranium itself, or is there a different story? >> It's basically a leverage play on, uh, on on uranium itself. Uh, okay. Yeah, we and we see this as as a as a long-term tailwind um, for for these companies. Yeah. So, it's not really particular to uh, it's really just a mega trend that they're going to be able to surf. Is it true that currently in North America uh most producers aren't actually, you know, in production or at least not in the volume that they should be? >> Definitely not in the volume that uh that they should be. Um, you know, I I will point out that when when you think about um nuclear as a uh as a clean energy source, uh not only does it uh produce energy with no carbon emissions, uh uranium mining is also a lot cleaner compared to uh you know some of the alternative things like um lithium and copper that uh would would be respons would be play a big role in um in uh in the green energy place. >> What about construction companies involved in the construction of new plants? Um is that primarily done by the private sector? And if so, is that investable? >> Construction companies I'm not too familiar with, but um a lot of the infrastructure buildout is uh is going to be driven uh by private investment. you know, for example, some some of the things that I'm focused on um in in my work in private equity and um and venture funding is on uh the infrastructure, right? So land purchases to build data centers powered by uh SMRs um classified ready sites uh that have a national security angle and uh yeah the public market doesn't really like to get involved on uh in long-term projects like this until until the cash flows are are very visible. So, uh I do see a lot of the uh investments related to this theme um more of a play in private markets than public. I mean, once those assets are up and running and cash flowing, they'll get spun out into a into an ETF or a data center read. Um but until then, it's it's mainly going to be in the domain of private uh sector invest private investing. >> You you're right also that SMR small modular reactors are first of a kind project. So it's going to take some time before uh they prove themselves for financing to come in. Um now ultimately the uh conclusion of this report is that the atomic engine of the AI um century is upon us. So the rise of AI is reshaping the physical economy in ways that are only beginning to be understood. Tell us about that Ed. How's our economy going to be changed forever? Well, uh, data centers are the factories of, uh, of the 21st century and, uh, you know, it's it's more than just a digital story, right? Uh the scaffolding involves um you know physical um structures that uh you know require things like concrete and um so it it's it's going to be more than just in in in the digital space. It's it's going to be about the physical scaffolding that that is needed for things like uh chips and um data centers and SMRs and all of the equipment uh related to that. So, uh in terms of the bottom line of the report, uh AI's exponential energy demand uh makes nuclear not optional but essential. Uh uranium and nuclear equities uh have durable tailwinds. uh they're structurally under supplied. Uh they have policy support AI demand and I think that's fueling a nuclear renaissance and it's not a short-term theme. It's uh it's the atomic engine of the AI century as as you mentioned >> the opposition to this plan. What is um the push back on a nuclearpowered society? >> Well, there are public perceptions. Uh obviously there are things that happened you know like um Fukushima um Three Mile Island you know if you if you go all the way back uh Chernobyl um but um the new gen you know I mean a lot of those things um you know Chernobyl required you know massive uh incompetence on on on a huge scale in order for something bad to happen. And I, you know, I don't want to minimize these things, but if you look at the, you know, what the actual consequences of Three-Mile Island and and Fukushima were, they they weren't that great. And as we come up with new uh generations of the technology, especially the ASMRs, uh they get safer and safer. So, and I do think the public perception is uh is changing. Uh we're seeing that uh policy has become much more supportive. Uh certainly Trump's executive order streamlines um timelines. So I think the direction of travel is clear but uh the path is going to be jagged for investors. So inspect investors should expect progress but uh it's it's not going to be a straight line and you know then we do have these choke points with uh with uranium right. So unless we have a lot more uranium, um it's it's going to be hard for us to scale uh in in the way that we need to in order to meet AI demand. So I think investors should focus on on those choke points and and look for opportunities there. >> With a 70% climb on URA over the past couple of months, is that fairly valued already? You think? >> Um I don't really see that. Um, you know, there's I I'm of the view as an investors that valuation only matters when it's at extremes. Uh, and I don't think um we see extreme valuations um in these companies basically because of of the history and uh and and the prior bearishness. Uh sure, it will get overbought from time to time and and there there'll be shakeouts, but I think these are uh positions that you want to hold for the long run and when they do have shakeouts at more. >> All right. Excellent. Thank you very much. We appreciate uh we appreciate your insights today. Uh Ed, where can we find your work? >> Um well, you can I publish thematic research for um Rosenberg Research Associates, partner with the with Dave Rosenberg. So, uh subscribe to Rosenberg Research and you can get uh my insights as well as the rest of the team. >> All right. Thank you very much, Ed. We'll uh follow you there. Appreciate your time. We'll speak again soon. Take care. >> All right. Thank you, David. Thanks for having me. Thank you for watching. Don't forget to like and subscribe.
AI's Energy Crisis: Why This Commodity Will 2x By 2040 | Edward Campbell
Summary
Transcript
AI has burst onto the scene. You have a step change in the demand for electricity. Right now there is uh a 30% shortfall in terms of uh uranium demand over uran uranium supply by uh 2040. >> Uh what will be the primary sources of energy that could power this additional demand? >> So it's mainly going to be nuclear. >> I'm joined today by Ed Campbell, founder of Red Hook Phoenix Investment Partners. We'll be talking about the future of energy and the commodities and resources needed for this AI transition. One of the biggest revolutions since the industrial revolution. What is the actual investment implication of this revolution? We'll talk about this with head. Edward, welcome to the show. Good to see you. >> Thank you, David. Thanks for uh thanks for having me. >> I'd like to start with this piece of news and get you to comment on this. This week, uh the Department of Energy has issued a proposal for AI data centers to be built at Oakidge, Tennessee. Now, this was the the site of apparently the founding of the original Manhattan project. They're calling this Manhattan Project 2.0. Uh this is the next step forward in the transformation from nuclear remediation site to nuclear renaissance hub, says the acting assistant secretary of environmental management. The department of energy is excited to partner with the best of American industry to bring this vision to reality. uh specifically the Oakidge reservation is one of the four sites identified by the department of AI uh department for AI infrastructure and generation projects on federal land. The DOE's office of science and um and office of environmental management are now seeking proposals from the US companies to potentially enter into one of one or more long-term leasing agreements at the site that will be solely funded by the applications. Um, I wonder how this fits into the $500 billion Stargate initiative that Trump launched earlier in the year and broadly how the government is um is planning to fund and build infrastructure for AI in our country. >> Yeah. Well, I think these are all uh examples of um the geopolitical imperative of um you know, us being um the major player in AI um energy sovereignty um well AI sovereignty uh equals energy sovereignty and you know the United States um has the biggest um installed base of uh of nuclear energy us along with Europe but we're really not keeping up with China um in terms of um building our base going forward. So AI is uh the next great general purpose technology um like the printing press, the steam engine, electricity or the internet. And um it's really more than that because it's a it's a meta technology that uh rewires how all existing technologies are used. So, it's it's really going to be the the biggest generalpurpose technology out there and that's going to create a tremendous amount of wealth and that wealth is going to be a critical component of national power. So, it's important for uh the United States to um keep its edge there. >> Infrastructure takes a lot of money, a lot of capital, a lot of time to build it out. Who's going to fund all these projects? Well, I think you know the vast majority of it is uh is going to be private capital, right? So, um I think that uh nuclear has a lot of good things going for it. Um if you think about it on on a on a legislative basis. So, a lot has changed since um the person in the White House has changed. If we go back to the Biden administration and the inflation reduction act, um there was really a lot of emphasis on wind and solar, um lots of subsidies. Uh currently green energy, especially wind, uh is going through a a bit of a default cycle here. One because the projects were uh maybe not economically feasible. Uh I know that here in New Jersey there were big uh plans for um offshore wind on the Jersey Shore which a lot of people weren't happy about. Well, those companies have gone bankrupt and there's a lot of stress in um in that industry um as the subsidies are pulled back. Meanwhile, the one big beautiful bill uh which pulled back those subsidies has created a lot of tax credits um for um for nuclear. Uh there was an executive order signed that has an aggressive plan for us uh building out nuclear plants along with tax credits and um so I think you know maybe some of the capital that would have gone to um wind and solar is now uh freed up for uh nuclear energy. So I I think it's going to be a combination of tax incentives as well as private capital. And another big thing with the um one big beautiful bill was um the expensing of depreciation back to uh 100%. So that uh certainly is uh something that uh incentivizes investment. >> I wonder if tech companies are going to become utility companies part-time as well. Take a look at this piece of news as well. This is coming from Microsoft. They're spending $1.6 billion to restart a nuclear reactor at 3M Island, which is the site of the worst nuclear accident in US history. was shut down in 2019. Um, yeah, they are signing a 20-year power purchase agreement last September. Um, and they've agreed to offtake the entire capacity of the plant's unit one reactor which has a capacity of 837 million watts. Uh, now uh yeah, they're calling the constellation is the uh the name of the project. What the constellation and Microsoft sorry um the company so um constellation acquired unit one of their plant in 1999 and they're partnering with Microsoft now what's what what does this signal to you that um tech companies are also acquiring and starting nuclear reactors? >> Yeah. So um the old model was hyperscalers uh bought from the grid like uh everyone else. Uh but the new model as you mentioned and as is mentioned in that article is uh vertical integration. So we have Microsoft backing the three- mile island restart. We have Amazon uh sitting data centers uh next to nuclear plants. We have Meta um engaging in a 20-year nuclear contract with uh Constellation. So, you know, I think here you have um SMRs turning electricity from a commodity into, you know, from an asset that you rent to one that you want to own or control. Uh when we think about um the bigger picture in terms of electricity consumption and and the long-term trends, well, if you think about the environment that we've been in for decades, grid planners were uh planning on declining load, right, associated with the decline in manufacturing. uh greater efficiency and now that AI has burst onto the scene you have a step change in the demand for electricity. So um you know for example let me cite some statistics here um you know data centers currently are 4% of the grid share as of 2024. Uh McKenzie estimates that they're going to be 12% by 2030. Now that that's just data centers, but if we um no, that's really the whole share of of data centers. But if you think about just the AI workloads alone, um they're currently about 3 to um 3 to 4% of of electricity. So when you get to 2030, um you know, you're going to run out of uh the grid is going to be completely clogged, right? So that's why uh you see a lot of these companies buying their own and locating them on on their facilities either as owners or uh with partners. Um basically you need a a behind the fence solution rather than um hooking up to the grid because the grid is going to be uh completely clogged. >> How much energy is needed by uh data centers in the future? Has anyone made well I'm presuming people have but what what have people made in terms of projections on how much energy is going to be needed in the next 5 to 10 years? Well, I don't I don't have the numbers for the next five to 10 years, but um you know, >> sure. In the future, yeah, >> data centers alone are going to be 12% of of of US electricity, right? Um and I don't have the the figures on on electricity by themselves, but as that becomes a bigger share, um you're going to need a lot more electricity than uh or on pace to be able to provide. >> I'll pull up some numbers later for the interview, but what does this mean for for commodities? uh what will be the primary sources of energy that could power this additional demand? So it's mainly going to be nuclear um because um you know it will involve gas um but uh nuclear has an advantage over gas uh in that uh it's carbon-f free right so there's still going still people who care about uh ESG as uh as investors so you know gas with carbon capture will play a role um but wind and solar are not suitable for uh the always on needs for AI. Um, wind and solar, they basically they can't scale AI. They're just too intermittent and uh too land intensive. Uh, for example, if you if you look at a density gap, right? In order to produce uh 100 megawws of um from a data center um for an SMR, you can do that on less than one acre of land. Whereas for solar that's going to take um 500 to 600 acres. Um for wind it's going to take 4,000 to 5,000 acres. So um both in terms of the land use and in terms of the intermittency, they're not simply they're simply not well suited for the AI age. um you need basically 24/7 reliable power and uh you basically can't uh wait for the sun up to come back up or wait for the wind to start blowing again. And even with improved battery storage um you know it's it's just not going to be able to scale in the same way nuclear can with the demands for AI. >> I just found something um as we were talking just now. So this is an article from Forbes. They're pushing uh AI is pushing the world toward an energy crisis. Catchy title. Uh good headline. At present, this accounts for 3%. So data centers and their associated transmissions account for 3% of global consumption emitting as much CO2 as Brazil. Now >> it's more than a lot of countries. Yes, >> it's yeah looking forward they're projecting or whoever has done the study is projecting that uh in the US alone the increase in power demand due to data centers is expected to rise from 200 terowatt hours as of 2022 to 260 in 2026 equivalent to 6% of all power use across the country expected demand expected to double by 2030 that's only in a couple years so um we've talked about nuclear energy what about the grid do we have the load at our current grid to service this kind of demand. >> Uh I don't have the numbers. Um you know, we're servicing it now. Uh fine, but the problem is um you know, as we scale and um electricity uh production is not just is not going to be able to keep up with uh electricity demand outside of a much bigger footprint for nuclear. >> So what's the thesis around uranium investing then? If uh nuclear is going to be the primary primary energy source uh in the future for servicing additional AI data center demands then uh the power source of nuclear reactors. Uranium certainly should be very high in demand. Uh what's your play there? >> Yeah. So you know we like we like the commodity uh we like uh uranium miners. Um there's basically you know right now there's uh a 30% shortfall in terms of uh uranium demand over uran uranium supply by uh 2040 that's uh that's going to double. Um I think that there's an opportunity uh for Canada. Uh Canada is poised to be uh a global energy powerhouse. Um basically uh so if you look at Camco and um their next generation Rook one that's uh set to be the world's largest uh uranium mine by uh 2030. And if you think about things in uh from a geopolitical sense, right? We're no longer in a world where efficiency and globalization is the main thing. We're breaking into a world of uh smaller blocks and um the um US aligned world is uh not going to be able to rely on Russia and uh China for Russia was a big player in in uranium. So uh places like Canada and Australia are going to have to uh fill that gap. So if we can't rely on um China and um Russia for critical minerals, uh places like Canada have a have a have an opportunity to uh to step up here. So you know in terms of who are you know who are the winners from uh an investment perspective it includes uranium miners uh it includes nuclear utilities and operators uh reactor tech startups um people who build SMRs uh new scale is is uh is one name that comes to mind if if you want to think about an ETF um there are thematic ETFs that focus on um both uranium as a commodity but also the the uranium miners. Uh the Sprout Uranium Trust is uh is one that comes to mind. And in terms of losers, uh I think we're looking at uh pure play wind and solar without storage. Fossil fuel generators um without uh a decarbonization pathway are going to have a tough time. and uh grid constrained utilities that are uh bypassed by um some of the PPAs that um you know nuclear operators are um able to generate. >> This is a chart from one of your reports uh the um >> AI nuclear nexus July report. So it's showing different countries from different parts of the world with their nuclear capacity right now. Red bar showing operating reactors, blue bar under construction and green bar planned. It looks like the US, if you add all the bars, is actually going to be behind China, assuming they actually go ahead with a planned nuclear uh construction. Uh what is RO by the way? >> Uh rest of the world. >> Oh, rest of the world. Okay. Um everyone else. >> Yeah. So I I mean as I was saying before you can see that the US and Europe has the biggest installed base but China is moving rapidly and um the you know Europe in particular had been going the other way in terms of uh phasing out their nuclear plants. uh now they're uh starting to reverse course but uh the US and and Europe are nowhere near moving at the same pace that uh China is and you know that risks us being left behind right so that's China as a as a uh you know uh command and control economy they can move quick um I guess this is you know one of the drawbacks of uh of a constitutional republic that uh you know you can't move as as quickly as uh the Chinese communist government can. >> You're right that winners deliver dense reliable carbon-f free power, losers do not. You also are right that pure play wind and solar developers without storage face limits as AIdriven demand prioritizes dense always on power. Uh so tell us why you're bearish on pure play wind and solar. Well, uh they're not well suited to uh provide for uh the demands of AI. You know, we already talked about uh the land density issue and the inter intermittency. Uh also uh public they're currently going through a difficult default cycle and uh the rug is being pulled out from under them in in terms of in terms of the subsidies. So um yeah there's just uh they're basically they will still play a role in decarbonization and decarbonization goals but uh I do see their role um you know becoming less and they're basically going to miss the AI mega trend. Let's talk about uh the next page where you disclose that you have a um uranium miners trade Euro ETF that's up about 70% as of June 2025. Uh you write that current primary production of uranium falls short of demand by roughly 30%. This is expected to get worse with demand growing outpacing supply. The shortfall is expected to double by to 60%. Uh by the time um planned nuclear facilities become operational 2040ish. And so, uh, the thesis for uranium miners now, um, are they are they, um, are they just a leverage play on uranium itself, or is there a different story? >> It's basically a leverage play on, uh, on on uranium itself. Uh, okay. Yeah, we and we see this as as a as a long-term tailwind um, for for these companies. Yeah. So, it's not really particular to uh, it's really just a mega trend that they're going to be able to surf. Is it true that currently in North America uh most producers aren't actually, you know, in production or at least not in the volume that they should be? >> Definitely not in the volume that uh that they should be. Um, you know, I I will point out that when when you think about um nuclear as a uh as a clean energy source, uh not only does it uh produce energy with no carbon emissions, uh uranium mining is also a lot cleaner compared to uh you know some of the alternative things like um lithium and copper that uh would would be respons would be play a big role in um in uh in the green energy place. >> What about construction companies involved in the construction of new plants? Um is that primarily done by the private sector? And if so, is that investable? >> Construction companies I'm not too familiar with, but um a lot of the infrastructure buildout is uh is going to be driven uh by private investment. you know, for example, some some of the things that I'm focused on um in in my work in private equity and um and venture funding is on uh the infrastructure, right? So land purchases to build data centers powered by uh SMRs um classified ready sites uh that have a national security angle and uh yeah the public market doesn't really like to get involved on uh in long-term projects like this until until the cash flows are are very visible. So, uh I do see a lot of the uh investments related to this theme um more of a play in private markets than public. I mean, once those assets are up and running and cash flowing, they'll get spun out into a into an ETF or a data center read. Um but until then, it's it's mainly going to be in the domain of private uh sector invest private investing. >> You you're right also that SMR small modular reactors are first of a kind project. So it's going to take some time before uh they prove themselves for financing to come in. Um now ultimately the uh conclusion of this report is that the atomic engine of the AI um century is upon us. So the rise of AI is reshaping the physical economy in ways that are only beginning to be understood. Tell us about that Ed. How's our economy going to be changed forever? Well, uh, data centers are the factories of, uh, of the 21st century and, uh, you know, it's it's more than just a digital story, right? Uh the scaffolding involves um you know physical um structures that uh you know require things like concrete and um so it it's it's going to be more than just in in in the digital space. It's it's going to be about the physical scaffolding that that is needed for things like uh chips and um data centers and SMRs and all of the equipment uh related to that. So, uh in terms of the bottom line of the report, uh AI's exponential energy demand uh makes nuclear not optional but essential. Uh uranium and nuclear equities uh have durable tailwinds. uh they're structurally under supplied. Uh they have policy support AI demand and I think that's fueling a nuclear renaissance and it's not a short-term theme. It's uh it's the atomic engine of the AI century as as you mentioned >> the opposition to this plan. What is um the push back on a nuclearpowered society? >> Well, there are public perceptions. Uh obviously there are things that happened you know like um Fukushima um Three Mile Island you know if you if you go all the way back uh Chernobyl um but um the new gen you know I mean a lot of those things um you know Chernobyl required you know massive uh incompetence on on on a huge scale in order for something bad to happen. And I, you know, I don't want to minimize these things, but if you look at the, you know, what the actual consequences of Three-Mile Island and and Fukushima were, they they weren't that great. And as we come up with new uh generations of the technology, especially the ASMRs, uh they get safer and safer. So, and I do think the public perception is uh is changing. Uh we're seeing that uh policy has become much more supportive. Uh certainly Trump's executive order streamlines um timelines. So I think the direction of travel is clear but uh the path is going to be jagged for investors. So inspect investors should expect progress but uh it's it's not going to be a straight line and you know then we do have these choke points with uh with uranium right. So unless we have a lot more uranium, um it's it's going to be hard for us to scale uh in in the way that we need to in order to meet AI demand. So I think investors should focus on on those choke points and and look for opportunities there. >> With a 70% climb on URA over the past couple of months, is that fairly valued already? You think? >> Um I don't really see that. Um, you know, there's I I'm of the view as an investors that valuation only matters when it's at extremes. Uh, and I don't think um we see extreme valuations um in these companies basically because of of the history and uh and and the prior bearishness. Uh sure, it will get overbought from time to time and and there there'll be shakeouts, but I think these are uh positions that you want to hold for the long run and when they do have shakeouts at more. >> All right. Excellent. Thank you very much. We appreciate uh we appreciate your insights today. Uh Ed, where can we find your work? >> Um well, you can I publish thematic research for um Rosenberg Research Associates, partner with the with Dave Rosenberg. So, uh subscribe to Rosenberg Research and you can get uh my insights as well as the rest of the team. >> All right. Thank you very much, Ed. We'll uh follow you there. Appreciate your time. We'll speak again soon. Take care. >> All right. Thank you, David. Thanks for having me. Thank you for watching. Don't forget to like and subscribe.