00:00was that when fossil fuel prices are high then what you can do is adopt solar because that's
00:09the cheapest energy source now and that conserves more of your domestic fossil fuel production
00:14so then you can either save it or you can sell it and it's projected in the u.s co2
00:21emissions could
00:22go up 28 by 2030 if all the data centers were done as currently planned versus having no data centers
00:31at all
00:42welcome to tangelic talks your go-to podcast from tangelic where we dive into the vibrant world of
00:48clean energy development sustainability and climate change in africa we bring you inspiring stories
00:54insightful discussions and groundbreaking innovations from the continent making waves in the global
01:00community tune in and join the conversation toward a brighter greener future let's get started
01:08welcome to tangelic talks a podcast at the intersection of energy equity and empowerment
01:13with your co-host victoria cornelio and andres tamas and in today's episode we've got our very
01:18own charles welcome charles thanks for having me thank you for coming on can you please tell
01:24people who you are because we know you already charles is part of the tangelic team we're very excited
01:29to get to chat to him yeah i'm charles f stromeyer the fourth and i'm the principal frontier ai and
01:37investment strategy at tangelic and i have over 30 years experience and investing in an ai
01:47and i took my first two ai courses actually in 1992 when i was an undergrad so it's actually been
01:56around
01:56a lot longer than most people think yeah i think there was someone who was shocked the other day when
02:03i
02:04said algorithms are ai and they've been around for however long what was ai like in 19 in the 1990s
02:12what was it being used for uh it was used for things like trying to predict uh let's see
02:21market fluctuations like in quantitative finance they were using artificial neural networks
02:28but they were very primitive this is before what we call deep learning ai or the transformer architecture
02:36which came out of google in 2017 but was then launched as gpt in 2018 by open ai that's crazy
02:44and how did you find yourself in this in this world what attracted you to get into this field of
02:50work
02:51well the two courses i took were very interesting the two ones in ai at oberland college in 1992
02:58and then i was at a scientific conference in 2004 and i met this guy who had developed a new
03:06kind
03:07of ai system at mit and that's really when i got interested in it again okay so it tracks back
03:14then
03:15yeah why would it be useful to have before and after data well i'll give you two examples
03:22data centers as you know use a lot of energy and water and they emit co2 and generate noise
03:31so i was a mass challenge mentor for optic cloud and they're the first to use ai agents to optimize
03:39cloud
03:40computing you can think of it like your spam email there's a lot of digital waste in there they can
03:46find all the digital waste in cloud computing and re-optimize it so that's a software solution
03:55and then a hardware solution that i really like is a new startup from two mit researchers
04:01called ferverit and they have a liquid cooling solution for data centers that doesn't use any water
04:08software and it's much more energy efficient so you can get 35 percent more tokens out of an ai model
04:19and the neat thing about it is it could enable data centers in places that it wasn't previously feasible
04:26like africa the middle east or parts of america where there just isn't enough water for data center
04:32cooling yeah and so the before and afterward matter in this case if you wanted to demonstrate
04:40a reduction in co2 emissions from a data center you can measure before when the data center is running
04:47and then after the intervention either software and or hardware how does the data center directly emit co2
04:55uh it's because of electricity demand going up from the data center it uses all kinds of electricity like
05:03it'll it'll burn natural gas or coal and it's projected in the u.s co2 emissions could go up
05:1228 percent by 2030 if all the data centers were done as currently planned versus having no data centers at
05:20all
05:21can this like it can these satellite technologies also be used to measure for example i've seen stats
05:26about the the uh the farm that the data center ai data center that kevin o'leary wants to build
05:35and the the projected increases in both daytime and nighttime temperature in the local area would
05:44those also be able to be measured by satellite i think they would be yeah i'm not an expert on
05:50that though so i'd have to look into it but it seems like something that could also be it
05:56yeah the satellites can also measure deforestation and crop disease the ai would basically accelerate
06:03and help you understand what's going on a little bit better right and there's other ways that i can
06:10help like predictive ai with early warning systems for floods for example for better weather monitoring
06:18yeah there's also precision agriculture so i was once a mentor for a company called auto agronom
06:27and they developed technology that uses sensors and tensiometers to listen to the needs of the roots of
06:35plants and by optimizing when you feed and water them you use much less fertilizer and water so that would
06:43be
06:44one another is digital twins and urban modeling some you can model cities and see how they might respond
06:54to rising sea levels for example and then more generally global monitoring platforms like what i just
07:01mentioned with satellites that's really cool so it kind of helps us get ahead of the issues we might be
07:07facing down the line yeah hopefully hopefully is this a product of modern ai or is this just a product
07:14of like how how much how different is modern ai from like just deep learning and machine learning
07:20in the past is it more is it more dynamic and that's why it helps us more than traditional modes
07:26of like
07:26uh this sort of computing it's mostly due to the transformer architecture i don't want to get too technical but
07:33it's in gpt stands for generative pre-trained transformer now arguably there were large
07:41language models in the 1990s like they beat people at the game jeopardy okay you might remember that yeah
07:50but it wasn't really till the transformer architecture that things took off and the token usage by large
07:58language models is very important you can think of a token as like a bit of text or data okay
08:04and it's
08:05or as a unit of intelligence as far as these ai models go no no it was similar to that
08:11is is it because
08:12there is so much more data available that we can make these technologies i guess better at predicting
08:20or are we implementing them differently it's a mixture of both the big data or large data sets
08:25certainly helps with making more accurate predictions and then i guess tying that into for
08:31example esg tracking if you're able to do before and after accurate measures would that help improve
08:38for example the reputation of a company or help fight greenwashing in some way because you've got
08:44some hard evidence of how things go following the articles in the european union and things like that
08:50like is that one of the benefits of having a before and after yeah and but i don't want to
08:56sound pollyannish that finance or tech can solve all our climate problems uh lisa sachs at columbia
09:04university just had a paper out and she i don't remember the title of the paper but i could look
09:09it up she
09:11argues there's three main types of climate risks there's planetary which is things physical like rising sea
09:18levels there's economic which is reduced economic production or productivity for example and there's
09:26finance financial which would be like credit quality or stock sell-offs because of the impact so
09:36part of the problem with the banks committing earlier to zero carbon and then reneging is these
09:44different types of climate risks they didn't map well onto their objectives which is sort of it sort
09:50of defeats the point of of uh of making that commitment if you're going to take it back
09:56just because it doesn't align with your objectives that's sort of but it really really is a great
10:02indication of the cultural state of things yeah and there's an old saying that impact investing is kind
10:08of like a houseboat it's neither a great house nor a great boat this is because of the tension between
10:15wanting to make a positive impact versus needing to make a profit yeah is this a tension that we see
10:21a
10:22lot in climate tech in general yeah because initially with with climate tech there's various barriers to
10:31getting it to the communities that need it most and one of those is the high upfront capital costs of
10:37doing a big solar project or a wind farm for example there's the high initial price for consumers for
10:46certain of them like in africa you can do pay as you go solar and that works well but evs
10:51are expensive
10:53but because of the iran war there's a lot of interest in cheap evs from china going to other parts
10:59of asia and even africa and then there's an uneven playing field because fossil fuels have a lot of
11:08heavy subsidies and even indirect subsidies that support them so it's sort of yeah it's sort of
11:15very unfair although tech like how would how would the subsidies for green for example energy or
11:23uh alternatives differ from the ones for fossil fuel um well the you look at the leader in clean energy
11:31it's china and they do industrial subsidies and other techniques to favor and prop up their industry
11:40and for example from about mid-march to mid-april the total volume of solar panel exports from china
11:51roughly doubled a lot of that went to asia and africa led by nigeria and once uh i told the
12:01finance
12:02minister of nigeria that a good strategy that i thought of in 2008 but and i told her later at
12:09the
12:09africa business conference at harvard business school was that when fossil fuel prices are high
12:18then what you can do is adopt solar because that's the cheapest energy source now and that conserves
12:24more of your domestic fossil fuel production so then you can either save it or you can sell it at
12:31the
12:31higher export price so that's not a direct subsidy but it's a way that even clean energy can help conserve
12:39fossil fuels absolutely it can help you keep it as either an economic reserve or as uh as uh just
12:48emerge for in case of emergency right or you can use less at home so there's less pollution or you
12:54can
12:54sell it at the export price which is much higher than the domestic price this is for net exporters of
13:01fossil fuels like nigeria i guess a lot of people when they hear fossil fuels they think it's extremely
13:06contradictory to the green space and renewables and things like that is there something we're
13:14missing when it comes to these things is that sort of a knee-jerk reaction i don't think it's entirely
13:20an
13:21unjustified reaction because global emissions and temperatures are at relative highs and they're
13:27projected to go higher so but one thing with clean energy myths that people have about it is it's not
13:37a guaranteed 100 clean silver bullet like there's heavy resource extraction so for batteries and solar and
13:47wind you have to have lithium cobalt and rare earth metals and 90 percent of the rare earth metals come
13:54from
13:55china and there's complex supply chains and there's infrastructure updates yeah so the production of
14:02it isn't necessarily scot-free yeah but that said another myth is that evs are powered by dirty electric
14:10grids because they're powered by natural gas or coal but in the long run transitioning to evs is much less
14:18carbon intensive than maintaining internal combustion engine vehicles right okay oh the good news about
14:26this is that there's something called moore's law which says that the power of semis rough conductors or
14:34computer chips roughly doubles every year and the cost comes down proportionally well with solar it's been
14:44even quicker than that and with batteries it's been even quicker than solar so the idea that renewables
14:51cost more just isn't feasible anymore because solar's the cheapest global energy source and we now have
14:57smart grids and batteries that can handle it at least with sophisticated electric grids they know how
15:03to incorporate it is that specifically talking at a mass scale like a company or a country or household
15:11finances like having solar in your house it depends on if there's um net metering so
15:19if i like in my neighborhood if i well we're on an independent power company but if i was on
15:27a grid
15:27like here in the northeast of the u.s and i had solar and i generated excess electricity i could
15:33sell
15:33it back to the utility and save money that way okay so it is becoming more affordable overall
15:40yeah and it goes back also in the developing world to what you said about esg benefits for example
15:47when a household in africa gets solar for the first time or clean cooking then the women and girls don't
15:55have to spend as much time collecting firewood they can study more they can work more they get more free
16:02time yeah and that's one of the things that in the black and white conversation of energy
16:08kind of get lost sometimes of it's not just about you know being able to turn on the lights is
16:15the
16:15knockoff effect of having access to these you know be empowered into the grid or having some energy
16:21independence is how it frees up your social life as well i want to focus a little bit on finance
16:28from a perspective of climate technology sounding very promising but then where does the funding
16:35and scalability come in because like you said it's hard to get it to the people that need it most
16:39but isn't that true of most charity projects right getting them funded and scaled well the reality is most
16:48startups and non-profits fail and so part of it is that scalability issue how do you climb the valley
16:56of death for clean tech startups there's depending on the type of startup for example with an impact
17:04or circular economy startup in africa you have to scale it with both public and private sector resources
17:12it's not sufficient to scale it just privately unless it's truly some kind of breakthrough technology
17:19that just pays for itself okay but that's the general problem especially in the u.s is the clean
17:25tech valley of death where you get some funding you build a prototype it works but then scaling it
17:31economically um in the clement space is a priority then environmental impact or commercial viability it
17:40depends on the region that's one of the barriers to getting clean energy to people who need it is
17:45uncertainty and fragmented regulation and trade tariffs all right okay yeah because
17:55with uncertainty it's harder to finance a long-term project and as you can imagine trump in the u.s
18:03is
18:03now promoting more coal so but thankfully a recent project in the u.s just raised 3.5 billion for
18:11solar
18:12and batteries so it's still doable and the really good news here in the u.s it was widely covered
18:19on
18:19the news yesterday for the first time ever consumers are now getting more electricity from solar than from
18:26coal oh wow that is good news not in the u.s and i didn't figure that that's very interesting
18:36yeah
18:39and the other thing is there's more emphasis now on nuclear now the leader in nuclear in the u.s
18:46is the u.s and eu but china's catching up they recently are building or have built six new large
18:53nuclear reactors so it's predicted by the end of the decade they'll be the leader in nuclear power
18:59but global warming's a global issue so every little bit helps yeah when it comes to i believe by year
19:07they're on track to to beat out everyone else when it comes to nuclear and that that'll be a that'll
19:14be a
19:15pretty pretty good chunk of their advantage over the entire world and having very very stable uh
19:22um clean energy yeah here in the u.s there's some emphasis on larger nuclear but there's a lot more
19:30recently on small modular nuclear reactors and i was once a mass challenge mender for oklo which now
19:40trades on the stock market under the symbol oklo and one thing they're doing is they did they just
19:48announced a 1.7 billion private center for recycling spent nuclear waste into new nuclear fuel right
19:56because newer newer tech uh reactors can use all of the fuel like it can use the the primary fuel
20:04and
20:04then the waste from it to keep itself sort of going right yeah it's also they might have the potential
20:10to create tritium which is needed by commonwealth fusion systems they're the most prominent
20:18commercial fusion project in the u.s because like i feel like uh when it comes to when it comes
20:25to
20:25nuclear like the the i'm very close to i love nuclear energy and um the this is like sort of
20:34the silver
20:34lining to the mass implementation of ai data centers in the united states that they've reactivated the
20:40conversation and this year updated a lot of the legislation about nuclear right um do you feel
20:47like this will get america more back on track it's like the timelines what what used to happen with
20:53nuclear in the united states is that the the timelines and the regulations would make it so that it just
21:01wasn't viable uh in a way in a proper way right um so now that we're seeing all of these
21:10uh small
21:11form factor modular nuclear reactors sort of being the main discussion where do you think they're going
21:18to be implemented or like do you see actually see it realistically like having multiple reactors going online
21:25maybe like one to uh i guess it's more on the timeline of four to five years from now yeah
21:32that's the current plan and they get activated on government and military sites initially uh
21:42but that's a bit of a very important debate now is how economically feasible will scaling up small
21:50modular reactors be versus traditional large nuclear power plants and i'm not convinced anyone really
21:58knows yet is nuclear on energy source that could be adopted in other places in the world it could be
22:05but
22:06there was a recent report that to give you an example if we installed more renewables that would do
22:14more for climate change than any direct air capture scheme because that's not very economical either and the
22:22advantage africa has the whole continent of africa they have 60 of the world's best solar potential
22:30but they only have one percent of installed capacity so i mentioned the startup fervor it what you could do
22:38in
22:38theory we don't know yet to scale but you can make a solar power data center for africa or the
22:45middle east
22:46and it wouldn't need any water for cooling with the liquid cooling alternative that fervor it has
22:52how does the liquid cooling alternative work do you know like i did yeah they submerge it in a special
22:57kind
22:58of liquid and unlike other liquid cooling systems it has much smaller bubbles which are much more efficient
23:05for heat transfer this is emerging that's on the surface of the servers themselves this is weather
23:12specific to where the center is yeah traditionally you couldn't build a large data center in the middle
23:19east africa or say arizona because there's no water to cool it with yeah i was wondering the reason i
23:25ask
23:26is because a lot of climate solutions are developed and then one of the issues i've seen with scalability
23:33is the context specific discussion of these technologies so for example you can't import the
23:38same technology to every place right a lot of the solutions have to be specific to context right it
23:44depends on the locality for example in the leading science journal nature there's recently a news article
23:52about there's going to be a lot more gullies in africa these are where they're there's not a lot of
24:00underground infrastructure and the rain suddenly forms and creates a gully in the ground so all these
24:07things these weather and climate and environment specific things have to be planned for before you
24:13do a large clean energy project and people are looking at what's called parametric insurance
24:20to hedge clean energy projects from climate change and physical risks but parametric insurance can be
24:27expensive and then we get into the finance tension again of needing funding i want to close off
24:33thinking about sort of in the next decade do you think we have a climate technology today that could
24:39be truly transformative in the way that it's being implemented at the moment or that we're close to
24:46breaking through i'm not an expert on nuclear fusion but commonwealth fusion systems just published five
24:52scientific papers claiming they'll be able to produce more energy than they use that would be a huge
25:00breakthrough if it works and can scale commercially i feel i'm always very skeptical when i hear that
25:07because that that that sort of claim comes around every five years saying it's five years from now
25:14yeah i agree and it really depends on the region the the best overall technology for climate change
25:21is solar because it's the cheapest and there's so many places where there's plenty of sunlight
25:27and the myth that solar doesn't work when the sun doesn't shine or the wind doesn't work when the
25:34wind doesn't blow that's not really true because of improvements in smart grids and batteries
25:40so for example in england where it's rainy 80 percent of the year we could still have solar
25:46yeah they have some solar in england and then they also have offshore wind power they could do
25:51geothermal potentially geothermal looks very promising as well yeah okay so there's a lot to be excited
25:59about geothermal seems to be working really well for like greenland if i remember correctly yeah i don't
26:05remember which one either good okay well thank you charles for helping us understand all of that
26:12you know as always you can check out the blog at tangeliclife.org slash tangelictalks where we'll
26:19have some research um that charles has mentioned and some links to some of the things we've talked
26:24about and some extra questions because we're going to stay on for a little bit but we'll catch you guys
26:28on the next one thank you
26:36let's talk power let's talk change for rural lights to brighter days equity rising voices strong we're
26:46building tomorrow where we all belong tangelictalks energy equity pride in power in the world side
26:56by side a spark becomes a fire a vision that's true together we rise it starts with you
Comments