André, founder and CEO

Asetek invented its closed water cooling circuits in 2003, and received its first OEM order from HP back in 2007. At that time, it was COMPLETELY new, but HP nevertheless took the lead and implemented it in its gaming PCs and commercial workstations.

Since that modest beginning, Asetek has sold more than six million(!) of these closed circuits for PCs, servers and data centers; it currently produces around 100,000  systems per month. Customers range from leading consumer brands to the big OEMs we all know, such as Acer, Asus, Dell, HP, Fujitsu, Intel etc.

We in Denmark and the EU are five to ten years behind in the use of water cooling for data centers. Asetek has several big data center installations in the USA, Singapore, Japan, Taiwan and elsewhere, and the customers range from US government customers such as the Department of Defense and Department of Energy to a number of universities, while Asetek also has several government and commercial customers in Asia. So yes – you could say the technology is mature!

Most of Asetek’s current end-customers use water cooling to save energy on cooling. District heating is not used in many places outside Denmark and the EU, so this implementation is not yet all that widespread, but, precisely because of our high consumption of district heating, we in Denmark have a unique opportunity to take the lead.

Tax and duty are a political matter that Asetek doesn’t concern itself with. From our point of view, the idea is that we should all do our best to cut CO₂ and save the environment. If the state wants to, it can simply charge the same duty on the heat as it does today, thus making the switchover tax-neutral.

Asetek would recommend keeping the duty but also allowing the data center industry to sell the heat. This would be a good incentive for the data center operators to make their data centers as green as possible. One can easily imagine a close collaboration between power stations/heat plants and data centers, so there are really no losers.

As Denmark and the EU are, unfortunately, lagging far behind, we at Asetek find ourselves encountering the same objections as we did when we first had to convince our big American customers many years ago. But we did in fact succeed. For example, it needed direct contact between Michael Dell (who founded Dell) and André Eriksen (who founded Asetek) for Dell to incorporate Asetek’s water cooling in their gaming PCs. Typical objections are:

• Water and electronics don’t mix. What happens if there is a leak? We all have water pipes running right alongside our electrical installations in our houses, cars and factories. The vast majority of data centers already have sprinkler systems in the ceiling. If they were to leak, the result would be the same. The industry does not question this. What’s more, many data centers already have water right next to the rack cabinets for what are known as rear door heat exchangers.
  

  Asetek’s patented technology has been developed in such a way that, should a leak occur, the damage will be very limited. The risk of leakage is an imaginary argument that doesn’t hold up; the best proof of this is that Asetek has sold more than six million cooling systems and continues to sell more and more. Wouldn’t the world’s biggest computer manufacturers have stopped doing business with Asetek if our technology wasn’t safe and reliable?

• Water cooling is complicated and expensive!? As far as whoever maintains the various servers and racks is concerned, it makes NO difference whether water cooling or air cooling is used. On the installation side of things, Asetek’s solution is simpler than traditional air conditioning, because the coolant is cooled by means of passive radiators, with the heat blown out into the open air or used for district heating. 

  The main difference between water cooling and air cooling is that, in a water-cooled data center, the installation side and the IT side must be integrated. Normally, the way the data center facilities (the buildings) are established is that an overall cooling requirement, independent of servers and racks, is specified. With water cooling, a little more coordination is needed.

  As regards costs, installation costs for buildings etc. will be the same in the worst case. Water cooling removes approximately 70% of all heat from the servers, so some air cooling is still needed, but far, far less equipment is required, which also brings the price down. On the other hand, some pipework is needed to connect to the district heating network.On the IT side, the servers and rack cabinets must have water cooling factory installed, as Fujitsu do for their customers, for example – in partnership with Asetek. The price of a water cooling system for a COMPLETE rack varies between 50 and 150 thousand Danish kroner for a rack cabinet FILLED with up to 100 servers. Depending on its purpose (supercomputer, commercial, social media etc.), a full server rack can cost several million kroner, so the additional cost of water cooling is minimal, and will in any case pay for itself in reduced energy costs within the first year of operation. So, no – it is neither expensive nor complicated!

• Heat pumps use electricity, so it still doesn’t make much sense to reuse waste heat! Correct – but Asetek’s solutions can deliver hot water at a constant 60 degrees, and they do NOT need costly and indeed energy-consuming heat pumps. ‘Our’ coolant can be output directly to the district heating network.

• Many of the servers in a data center will often spend some time ‘idling’, so you can’t control how much heat they release! It is true that many data centers have all too many servers sitting there consuming power for no reason. Asetek has small pumps in each and every server; if the processor is standing idle, the pumps are turned down so that the water released is always at 60 degrees regardless of the server load.

• Water cooling doesn’t extract 100% of the heat from the servers anyway, meaning air cooling is still required – so what’s the point? There is no obstacle, potentially, to extracting 100% of the heat with water cooling, but then it does get complex and expensive unless the servers are heavily custom designed.

  Asetek has chosen to focus on standard servers, and we can reduce one of the world’s biggest (and fastest-growing) environmental problems by 70%. We think that is a good start, and more than enough to be going on with. After all, the wind turbine and electric car industries had to start somewhere, too…

• After all, waste heat from a data center is pretty much just lukewarm water and cannot be reused without heat pumps. Heat pumps use electricity and have to be paid for. Correct, generally speaking. But Asetek’s water cooling can deliver hot water at a steady 60 degrees and does not need heat pumps.

• The water has to be a lot hotter to be usable for district heating. According to Professor Henrik Lund at Aalborg University (who researches district heating), the optimum temperature is 60 degrees. Asetek’s water cooling can deliver 60-degree hot water with no need for it to be heated up artificially.

• A data center isn’t a stable heat source, because the quantity of heat depends on the load on the servers in the data center. Correct—nor does a wind turbine deliver power when there is no wind. Unlike with windmills, though, with data centers a minimum quantity of year-round heat can quickly be established, so this will not be a problem in practice.

• If data centers are to supply district heating, won’t this take away the raison d’être of Denmark’s existing power stations/heat plants? We don’t believe this is the case. A heat plant already has a production price, and it would be natural for it to enter into a supply agreement with a data center – or perhaps the heat plants should think innovatively and locate data centers in the immediate vicinity of existing plants.

• We politicians should not interfere in the way private companies build data centers. Quite wrong. It is precisely politicians’ job to ensure that requirements are imposed on SUCH severely environmentally damaging companies, and this should include the potential for recovering energy that otherwise has to be generated in a different (and environmentally harmful) way. But the requirements must also be realistic, of course. And there is no reason to establish more costly offshore wind farms than necessary. More specifically, there is no need to establish more green energy sources than what is needed to support the needs of to data centers. There are plenty of other energy hogs to supply green energy to.

• Wind turbines and other sustainable energy initiatives are already up and running, though? The one doesn’t exclude the other. On the contrary, as water cooling reduces electricity consumption by up to 25% throughout a data center (by cutting in half the cooling energy needed), fewer wind turbines and less investment will be needed. Moreover, ‘green electricity’ does not solve the problem of all the heat that currently just goes to waste. Clearly, it’s easier to do everything ‘the usual way’. This is a massive problem that has been unnoticed until now. It is imperative that we lay down requirements and get a grip on the issue immediately, as has been done in several other industries, such as the automotive industry and in agriculture. It goes without saying that there are many other sectors to look at, too. But the data centers are indisputably huge, global and fast growing – and, more to the point, solutions already exist!

• Isn’t it crazy to let data centers into Denmark and give them all this ‘green electricity’? The value of this opportunity is not well understood. In fact, we have a unique opportunity to reuse energy from data centers for district heating, something we can only do thanks to our well-developed district heating network. As we have relatively low outdoor temperatures, data centers based in Denmark will cost less to cool. As well as being able to make use of the heat, Denmark and many parts of the EU are ideal locations for data centers, as seen from a global perspective. In Denmark, over 50% of all dwellings use district heating; in the EU as a whole – as far as we are aware – the figure is currently 12-13% and rising. So, there is great potential.

Yes, the data center industry is comparable with the automotive sector in many ways. It is important to understand that social media etc. are just one category (hyperscale), with Facebook, Google, PayPal, LinkedIn etc. having their own servers manufactured by ODMs in China and Taiwan. These hyperscale customers typically operate their data centers themselves.

However, commercial data centers and supercomputers typically consist of servers produced by known OEMs such as HP, Dell, Fujitsu etc., but again it is the end-customer himself who is in charge of operation.

In other words, there are a variety of manufacturers behind the servers installed in a data center. And it is ultimately the servers that use up all the electricity, both in operation and for cooling, so you need to go in and lay down requirements for the servers used in a data center, requiring them to be capable of using water cooling and able to recycle the heat in the form of hot water at 60 degrees. It sounds complicated, but Asetek’s standard solutions can cope with it (and there are also other producers in the world who can do it, too).

After all, the car industry didn’t invent expensive catalytic converters, particle filters, diesel exhaust fluid and other environmental innovations that have made their cars more expensive without good reason. The did it ONLY because it became a requirement!

Data centers are currently under no requirements regarding either energy efficiency or reuse of residual heat. Until that happens, the environmental impact will not be eliminated.

• Dell, HP and the rest make their living selling servers, among other things. Because no environmental requirements are politically imposed on data centers at this time, and because there is therefore no customer demand for it, it makes no sense for them to implement it. HP in the USA tested Asetek’s data center water cooling several years ago and knows that it works, but, with no demand, they aren’t adopting it. A company like HP has nothing to gain by implementing water cooling. Just as BMW had nothing to gain by putting particle filters in their cars, unless it becomes a requirement or there is a demand for it. Companies think about profit, not the environment.

• Facebook, Apple and the rest want to cultivate a green image. They can have one in Denmark because here wind turbines can supply ‘green electricity’. Unprompted, though, these companies have no interest in recycling waste heat. They see themselves as social media, not as a power station or heat plant. The winner with water cooling is the environment, and unless requirements are set, Facebook, Apple etc. ‘couldn’t care less’. Remember, too, that they have many data centers around the world, and it is easiest for them to make them all identical. Many parts of the world do not yet have district heating, so, if the companies can avoid thinking about residual heat etc. in Denmark and the EU, that makes it easiest for them. Companies do whatever is easiest; they meet the minimum requirements and typically do no more.

• It is a bit of a ‘chicken and egg’ situation. In Denmark and the EU, we have well-developed district heating, and we need it. The IT industry is very American, and little thought is given to district heating there. In Denmark, on the other hand, we lag far behind in the use of water-cooled data centers, so things won’t move without a ‘jump start’. It is not a question of technology. It is question of whether we want to do the best for the environment.

As previously stated, Asetek has already sold to a significant number of customers, including data centers throughout the world. Many of these are publicly accessible.

• Here is a video from the University of Tokyo, showing a data center being built using Asetek’s water cooling. The data center also won a prize as one of the greenest data centers - if not the greenest - in Japan:
  https://www.asetek.com/newsroom/media-resources/videos/2017/installing-thehighest-performing-supercomputer-in-japan/?referer=0
  
  
• Here is a video from the University of Tromsø, which tested and used Asetek’s water cooling in their data center, with heat recycling, back in 2016: https://www.asetek.com/newsroom/mediaresources/videos/2016/asetek-recycles-waste-heat-at-university-oftromsoe/?referer=0
  
  
• Here is a video from Berkeley in Palo Alto, California, the first to test Asetek’s water cooling, as long ago as 2013. Lawrence Berkeley National Lab is recognized as a world leader in the field. Here, Henry Coles gives a presentation following a 6-month study: https://www.asetek.com/newsroom/mediaresources/videos/2013/asetek-direct-liquid-cooling-demonstration-presented-bylbnl/?referer=0
  
  
• Here is an official report by the US Department of Energy’s National Renewable Energy Laboratory (NREL), demonstrating the potential and stating that the system is easy to use and that, after 18 months in an operational environment, they had encountered no problems despite the technology then being at an early stage: https://www.nrel.gov/esif/partnerships-asetek.html

Since we in Denmark and the EU unfortunately lag behind in research and implementation in this area (although many believe otherwise), Asetek has no Danish testimonials. Henrik Lund, Professor of Energy Planning and the head of Innovation Fund -- Denmark’s 4G District Heating Research Center at Aalborg University, is very interested in conducting a research project on data centers and district heating. Professor Lund is happy to talk about the potential, and can verify the statements made regarding the potential. He cannot talk about Asetek’s technology as he is not yet familiar with it. However, the above technology testimonials from customers should be more than sufficient. Henrik Lund may be contacted on 9940 8309 / lund@plan.aau.dk.

Even though in Denmark we are behind with implementing water cooling in data centers, we still have a unique opportunity to be first movers in implementing significant recycling of residual heat from data centers, and we could be world leaders in this field. And, from being considered a necessary evil, our impressions of data centers could be turned into something truly positive, so that we would want to have more of them in Denmark.