- 3 months ago
From livestock AI to textile waste, data center cooling and solar powered Bitcoin mining: New technologies promise efficiency but reveal growing pressure on water, energy, and resources.
Category
🤖
TechTranscript
00:07How much do you already use artificial intelligence, either at work or elsewhere?
00:12Many companies have started to depend on it.
00:15For some agricultural businesses, AI means fewer employees and allows them to keep track
00:21of systems and livestock.
00:27And these are our other topics on MADE.
00:30Why people are plunging AI data centres into the sea.
00:35How AI and new processes could help cut textile waste.
00:40And how solar power and waste heat make Bitcoin mining more environmentally friendly.
00:56Just two days old, a little miracle.
00:59The thoroughbred foal Danger Zone.
01:02And he's alive at all, is thanks to modern AI technology.
01:07His birth was dramatic.
01:08The colt arrived far too early, but artificial intelligence helped save his life.
01:13Janina Brinkmann is his owner.
01:16She trains young horses, looks after sick animals and regularly helps with the births in the
01:21stable.
01:23An AI-linked camera monitors her horses.
01:29During Danger Zone's birth, it spotted the emergency and alerted Janina, who freed the colt, just
01:36in time.
01:41The foal survived only because the camera alerted me at exactly the right moment.
01:47It was a difficult birth and the mayor didn't take care of him.
01:50But I got the alarm in time and was able to help.
01:58When Janina's smartphone rings, it means there's an alarm in the stable.
02:04The AI detects even the slightest behavioural changes.
02:08If a horse is lying flat on the ground or nervously stamping its hooves, it often signals
02:14colic and a potentially life-threatening intestinal blockage.
02:22For owners who've just fought through a colic and pulled their horse back from the brink with
02:27all the stress and vet spills, this is a real game-changer.
02:34Their horses are monitored 24-7 and even the smallest sign of colic triggers an alarm.
02:43Janina relies more and more on smart technologies.
02:46Even before training sessions, she checks an app to see how much a horse has eaten or drunk,
02:53even how long it has slept.
02:56For professional riders preparing for competitions, this kind of data is invaluable.
03:04The software was developed by the Hamburg-based company Acaris.
03:10Ten employees train the AI, and for six years they've been collecting video footage from horse stables.
03:18They draw their data from a staggering amount of recordings.
03:30What we did was train the AI on 650 years' worth of video material, so it could understand normal horse
03:37behaviour – eating, drinking, moving, lying down.
03:40Then we added examples of problematic behaviour.
03:44That way the system can tell in real time whether a horse is fine or whether something is wrong.
03:52At the beginning, it was a risky venture.
03:55Dräger couldn't find investors, so he financed the start-up with his own money and help from his family.
04:02Today, he stores the valuable video data securely on servers in the EU.
04:07The worst thing, he says, would be if they were hacked.
04:12Acaris is now transferring its expertise from horses to cows on large dairy farms with hundreds of animals.
04:19What farmers once learned through experience and close contact is now handled by data and sensors.
04:27With about 100 to 160 cows, we still knew each one, but as we grew, that personal connection disappeared.
04:35We don't know their names anymore, they don't even have names now.
04:41Today, the animals are identified by numbers.
04:45Neckband sensors track their feed intake, resting times and movement – all crucial indicators of the cattle's health.
04:55But the AI goes even further.
04:59It uses video data to track each animal's movements from above.
05:04And it can tell whether a cow is doing well.
05:09All that information feeds into the central control room.
05:16We wouldn't even have a chance to manage this without the software.
05:19It's essential for us.
05:21Everything comes together in one program where it's analysed centrally.
05:25Without it, we'd be sunk.
05:31In the control room, the AI shows that cow number 073 is limping.
05:38Hoof care is needed.
05:40Otherwise, she'll be in pain and produce less milk.
05:49The AI not only helps detect illnesses early, it also improves animal welfare and ultimately boosts milk production.
06:01And with every new video frame, the system keeps learning.
06:06AI training in the cowshed is already moving into the next phase.
06:16Data centres are mushrooming and the more we depend on AI, the more of them we'll need.
06:22All around the world.
06:23And they in turn consume a lot of energy.
06:25Much of which is used to cool the centres down and stop them overheating.
06:29Are there better ways to do this?
06:33What happens when you ask ChatGPT to reword an email?
06:37Your prompt goes to a data centre and triggers a wave of computations on hundreds of servers.
06:44As those servers get hit with requests from thousands of users at once, they start to heat up.
06:50To keep them from overheating, data centres typically cool in with water, fans and air conditioners.
06:56Cooling has always been a core part of how data centres work.
07:00Even one of the first computers in the 1940s had vacuum tubes which were cooled by fans and ventilation.
07:06Despite being literally trillions of times more powerful, the data centres of today still look kind of similar.
07:13The big difference is they're using a lot more energy.
07:17Nearly 50% of their entire energy use can go towards keeping the CPUs that processor prompts, also known as
07:23microchips, from overheating.
07:24This Google data centre in the US state of Oregon is large enough to run computations needed for AI applications.
07:32A typical data centre covers an area of about 9000 square metres.
07:36This one is 13 times bigger.
07:38It's called a hyperscale data centre.
07:41The scale of the AI data centres, which we call hyperscalers, is so much larger than the data centres that
07:54we have been living with pretty much without even noticing them for the past decade plus.
08:01Google's hyperscaler in the Dales, Oregon, used nearly a third of its city's water supply in 2021, according to public
08:08records obtained by the state's largest newspaper.
08:11Relying on water for cooling is a common approach.
08:14That's why when you look closely at this map, you see big clusters of data centres here, close to fresh
08:19water.
08:20Nearly a tenth of the world's data centres are in this part of the US.
08:23The region is called the Great Lakes and is home to one fifth of the world's surface fresh water, much
08:29of it naturally cold.
08:30Using it lets data centres spend far less electricity on cooling.
08:34The water gets sprayed into rooftop cooling towers, allowed to cool naturally through evaporation, and then used to keep servers
08:41from overheating.
08:43Fewer fans and less electricity, the method is called evaporative cooling.
08:49Now let's look at the other side of this trade-off.
08:53Google's data centre in Mesa, Arizona, sits in one of the driest regions in the United States.
08:59So they went with air cooling.
09:01Giant fans pushing air across the servers.
09:04No water, but a lot more electricity.
09:09But as new AI models drive data centres to pack in more powerful chips and denser servers,
09:14they're generating more heat than ever, meaning cooling needs to work even harder to keep up.
09:19Or, it can work smarter.
09:22One solution is to directly cool the microchips, where most heat is generated, instead of the whole server room.
09:29Direct-to-chip cooling uses a water-based liquid coolant.
09:32This circulates across metal plates, typically made from copper or aluminum, stuck onto the chips.
09:37Companies say using liquid to absorb heat from chips in a recirculating closed-loop system cools more effectively than chilling
09:44the air.
09:44Liquids move heat faster than air does.
09:48Air is very light, and liquids are very dense.
09:51So the same volume moving through the system, liquid can remove way more of the heat.
09:58But what if you could take that idea even further and dunk an entire server into cooling liquid?
10:05I know what you're thinking. Tossing an AI server rack into a cold water bath is a sure way to
10:10fry the system.
10:11So, for immersion cooling to work, data centre providers use special oily liquids that won't conduct electricity or cause metal
10:18to rust.
10:22Gigabyte, a data centre services provider in Taiwan, says cooling its servers in this bubbling liquid allows them to be
10:28stacked more densely.
10:29With the direct-to-chip cooling method we explored earlier, some fans are still needed.
10:35With immersion cooling, there's no air around the server, removing the need for fans altogether.
10:41In the frosty harbors of Denmark, one operator is taking cooling to a whole new level.
10:47By plunging data centres directly into the frigid waters of the North Sea.
10:55And by placing them subsea, we eliminate the electrically driven cooling.
11:00So we see about a 40% reduction in the power that's consumed and then about 40% decrease in
11:07the carbon emissions.
11:07There's Maxi again. Her company's shipping container-like capsules weigh 46 tons.
11:13Inside, high-density server racks hold over 1,100 processors.
11:17Suspended in, you guessed it, a non-conductive oily coolant goo.
11:22The heat this coolant absorbs from the computing components quickly emanates outward into the frosty waters just outside.
11:28The modules can be connected, for example, to offshore wind turbines to power the computations, though no electricity is needed
11:36for cooling.
11:37No fans, no lights, and no pumps for circulating the gooey coolant.
11:41This means that 99% of electricity goes directly to computing power, according to the company.
11:47There are problems with underwater data centres, though.
11:50Every day, there's some GPU that fails due to plenty of different reasons.
11:55If it's in the ground, we can easily replace them or fix the errors on these GPUs or some other
12:04hardware.
12:05But if it's undersea, then it's kind of hard to maintain.
12:10Surveys show that over half of facilities have experienced at least one outage in the past three years.
12:15Though most of these are minor hiccups, it shows that problems can come from all angles, like memory errors, network
12:22glitches, power issues, or even natural disasters.
12:25But there's one more data centre frontier, and this one goes far beyond the limits of Earth itself.
12:34Operators are exploring the idea of shooting data centres into space to orbit around our planet.
12:41The idea is they can rely on solar power for their computing, while the cold of space would help to
12:46naturally keep the servers cool.
12:48By staying close to Earth's surface, data could travel back and forth using optical lasers without too much lag time.
12:55In 2026, Elon Musk's rocket launch company SpaceX merged with his artificial intelligence company XAI with the goal of launching
13:03AI data centres into space.
13:05But researchers are sceptical.
13:08Data centres require a lot of cooling.
13:12That's actually really hard to do in space.
13:14It seems like, oh, space is cold, infinite heat sink capacity, but it's actually really hard to get the heat
13:22to leave, right?
13:23When the sun is shining, all that heat builds up.
13:26Some of these new cooling technologies are more feasible than others, but they all have the same goal.
13:31Making AI use less water and energy.
13:34But that might not be enough.
13:36It could also make sense to reduce the need for cooling in the first place.
13:40Something as simple as shifting workloads to cooler hours or water-secure regions can significantly reduce environmental impact.
13:47There's also a school of thought in AI called Small Data, Big Tasks.
13:52This approach relies on changing the way we train AI models altogether by teaching it in small, deliberate steps, rather
13:59than brute force programming them with millions of possible scenarios.
14:03This reduces the mass of computations that keep data centres running, hot and thirsty, around the clock.
14:10And then we, the users of AI, can also play our part.
14:14You can opt for less water-intensive LLMs or select which model they use when entering props.
14:19Another solution, of course, is choosing not to use a chatbot to draft every email or automate every moment of
14:25friction in our lives.
14:28Because behind every quick AI response is something very physical.
14:32Water, energy, and communities already stretched thin.
14:42Do you love buying and owning new clothes?
14:45You're certainly not alone.
14:47Big chains like Primark, Zara, and H&M have been doing a roaring trade for years.
14:53And now Chinese online retailers like Shein and Teemu are shaking things up with fast fashion at rock bottom prices.
15:00With collections being overhauled more quickly than ever, our closets fill up faster and faster.
15:07But what happens to the clothes we no longer want?
15:16The world is being flooded with unwanted clothing, with an estimated 90 million tons more piling up every year.
15:24Every 60 minutes we generate clothing waste weighing the same as 18 fully loaded A380 aircraft at takeoff.
15:31And that's a conservative estimate.
15:34What happens to all that material?
15:37Most of it is incinerated or dumped, often in countries in the global south.
15:43Chile has become the worst dumping ground for used clothes.
15:49Even in Europe, less than 1% is actually recycled to make new clothes.
15:55That's despite recent innovations that could significantly boost that figure.
15:59In this episode we explain what needs to happen for those technological advances to finally be put to use.
16:07Fast fashion involves a relentless cycle of mass production that keeps prices low and consumption high
16:13while intensifying a trend that began in the 1990s.
16:17Cheap clothes production in poorer countries and cheap prices for shoppers in industrial nations.
16:24This model factory in the German city of Augsburg shows how recycling could help to conserve our precious resources.
16:33Regardless of whether the recycling process is chemical, mechanical, biological or whatever else,
16:39you always have to sort first.
16:40And the better the sorting, the better the outcome.
16:47And what we're doing new here is to use AI for that.
16:54Stefan Schlichter is the founder of the factory.
16:57The AI can detect non-fiber elements such as buttons or zippers
17:00and identify the type of fabric and clothing in question.
17:05We're still partly at the basic research stage, but already moving toward real-world application.
17:11We expect the first systems to reach the market within the next two years.
17:15So technology can help, but the current situation is a sobering one,
17:20because the majority of used clothes never even make it as far as a sorting facility.
17:26In Europe, most old clothes and shoes end up in household waste
17:30and are therefore dumped or thermally recovered, that is, incinerated.
17:35And of those items that are collected separately, over a quarter are destined to essentially end up in the trash.
17:41So what is the European Union doing?
17:44Among the new requirements for producers, a responsibility to finance sorting and recycling infrastructure.
17:51And clothing production has to be greener, as laid down by the Eco Design for Sustainable Products regulation.
17:58But will the plan work as intended?
18:00This can and will work.
18:03Right now, this situation is tricky, due to the complex nature of the issue.
18:07And that complexity raises a question.
18:10Which legislation will actually take effect, and when?
18:18Nicole Hoon also works for the model recycling factory in Augsburg.
18:25She says that the fibers fashioned here from used clothes are now market-ready.
18:36We need binding recycled content requirements for new clothes.
18:40Only then will the market regulate itself, so that buyers are actually willing to purchase recycled fibers.
18:49That's the only way to prevent these massive amounts of clothing being incinerated.
18:53And instead, reintegrate that waste into new products.
19:02A recycled content figure of 5% would be a decent start, she says.
19:06The researchers in Augsburg use mechanical recycling, which they say is the most advanced of the options.
19:16Mechanical recycling involves the fabrics being shredded after sorting until the material comprises individual fibers.
19:25The camera can now identify material components using near-infrared.
19:30So it can determine how much polyester and how much cotton is in the material.
19:38After being blended with new material, those fibers can then be spun into new yarn.
19:44The advantage here is that the original material, used clothing, does not have to be strictly pre-sorted according to
19:52individual fiber type.
19:54Take polyester cotton blends.
19:56Waste management people assume you first have to separate and process them individually, and then combine them again.
20:03But here in the recycling workshop, we have shown that you can absolutely recycle them as a blend.
20:09The polyester even retains an extremely high quality and survives recycling with less fiber shortening than the cotton.
20:17So, we do still have plenty of ways of making viable products.
20:25But does starting small work here?
20:28The McKinsey management consultancy says that recycling only pays off if you start on a large scale.
20:35For the EU, it estimates this would mean investing up to 7 billion euros by 2030.
20:41But the dividends would not be long coming, with a new recycling industry potentially generating a profit of over 2
20:49billion euros annually in the EU.
20:57Bitcoin was the world's first cryptocurrency and is suddenly a success story.
21:02But mining Bitcoins consumes vast amounts of energy, roughly the same as the whole of Switzerland, per year.
21:09That's terrible for the environment, creating massive CO2 emissions and draining groundwater supplies.
21:16A pioneering German entrepreneur powers his data center with his own green electricity.
21:21A role model for the big players?
21:25Hard drives submerged in oil bought for one purpose, mining Bitcoins with surprisingly good odds of success.
21:32That's what this German entrepreneur is doing.
21:36Incredible. The machine started up and we joined the network for the first time.
21:42And he insists that almost any company in the world could do it.
21:46And fairly easily, in fact. But is that really true?
21:52Christian Kläger's Bitcoin mining setup looks tiny.
21:56Especially next to his major competitors in the US, who operate with far more computing power.
22:03It seems like a race he doesn't stand a chance of winning.
22:07Just like the big players, we join what are called mining pools.
22:11Basically a betting pool where people work together to construct the next block and earn the reward.
22:17Whether we contribute a lot of computing power, or just a little, the payout is proportional, so the odds are
22:23similar.
22:25But the far more important question is, where does the electricity for all this come from?
22:35In his case, the electricity is practically free. A major advantage in Bitcoin mining.
22:41The power comes from solar panels on the company's roof.
22:47They were originally installed to supply the company's main business.
22:52Producing equipment for spraying liquids.
23:01When you think it through, it's obvious. There are weekends, holidays, times when business slows and you're on reduced hours.
23:08But the sun keeps shining and producing electricity.
23:12And when we feed that power into the grid, last year for example, we earned less than two cents per
23:17kilowatt hour on average.
23:23That's what led him to Bitcoin mining.
23:26And to make the system even more efficient, he uses the heat generated by the computers as well.
23:32It warms the water bath used to test spray cans, which are also filled at his company.
23:37At around 50 degrees Celsius, the cans are checked for leaks.
23:46If you knew all the challenges beforehand, you probably wouldn't do it.
23:49That means sourcing the hardware, mostly from China, and then integrating everything in the software side.
23:55And it's not just the data center, it's also the heating system and the hydraulics.
24:00And finding electricians or heating contractors who don't run the other way when you mention Bitcoin is another challenge.
24:10The expertise he's gained is now also being used in Northern Europe, in Finland on a larger scale.
24:17Working with partner companies, Kläger uses only green electricity for mining.
24:22The waste heat from the servers is used to warm nearby homes.
24:28And the best part? One of our machines in Finland actually solved the puzzle for a new Bitcoin block.
24:34Because the blockchain is transparent, we could clearly see that it was our machine that did it.
24:43The reward for solving the complex calculation? 3.125 Bitcoins, currently worth around 185,000 euros.
24:52All good? Not entirely. In 2025, Bitcoin mining consumed 138 billion kilowatt-hours worldwide.
24:59Mostly green electricity, but electricity that could also replace countless coal-fired power plants.
25:07People say Bitcoin mining uses electricity that could power entire countries.
25:12But that comparison doesn't help.
25:15It's all about the benefit.
25:18AI data centers also use huge amounts of energy.
25:22And we don't claim their power should go toward world peace.
25:26That's just an exaggerated example.
25:31But what about the finances of his German operation?
25:34Since launching in November 2022, it has generated roughly the value of one Bitcoin per year,
25:40currently worth just under 60,000 euros.
25:43It's not a huge sum, but still far better than selling surplus solar power into the grid for much less.
25:51And that's all from us at MADE.
25:53This time, we looked at how AI can assist in the world of work,
25:57from farming to textile production.
26:01See you next time.
26:02See you next time.
26:06See you next time.
26:09See you next time.
Comments