Interview: Demand Is Exceeding Our Capacities Thirtyfold — Aroosh Thillainathan


Northern Data AG is one of the leading players in the high-performance computing (HPC) market and is building the world’s largest HPC facility in Texas.

Based on contracts already signed, the stock-market-listed company (German Stock Exchange, XETRA: NB2, ISIN: DE000A0SMU87) has recently issued guidance for the year 2020, with sales expectations in the region of EUR 120 million to EUR 140 million, and EBITDA expectations from EUR 45 million to EUR 60 million. The tremendous upsurge in demand for HPC, which has been turbo-charged by the corona crisis, has accelerated Northern Data’s growth beyond precedent. We spoke with CEO Aroosh Thillainathan.

Q: Mr. Thillainathan, in a few days you will be commissioning the world’s largest data center in Texas. How does a comparatively unknown listed company from Germany come to own a data center like this, in the USA of all places – the home market of the IT giants?

Aroosh Thillainathan (AT): Our holding company is indeed based in Germany, but we are a truly international player. We have our roots in the blockchain business. Blockchain was the first commercial application of HPC and, over the past seven years, this niche market has given us valuable experience in how to scale the use of high-performance computers in a cost-effective and energy-efficient way. For example, the highly specialized chips needed for blockchain applications have extreme electricity and cooling requirements – we needed to address these challenges many years ago, while our competitors in the HPC sector are only uncovering them now. The reality is that, only three years ago, there were no large-scale applications for HPC outside of blockchain. Autonomous driving and artificial intelligence weren’t a significant market back then. So, while Amazon and Google might have deep pockets, they don’t have the same knowledge we do.

Q: You offer solutions for high-performance computing. What exactly does high-performance computing mean?

AT: High-performance computing (HPC) represents the future of computing. HPC offers many times the computing power that normal desktop PCs generate. We are talking about thousands of high-performance computers processing billions of units of data in real time. As a result, HPC can be used for tasks that would be impossible for regular computers or would take an impractically long time to complete. In recent years, HPC has mainly been used in scientific contexts to run complex models. Recently, however, more and more practical fields of application have been entering the market, making HPC a massive growth market. The application areas are exploding.

Q: What are the potential applications of HPC?

AT: These include artificial intelligence, autonomous driving, blockchain technologies and the life sciences sector. In the field of life sciences in particular, HPC and its immense computing power is revolutionizing the discovery and development of new drugs and therapies.

Q: Does this mean that you are benefiting from current issues, such as the coronavirus pandemic?

AT: In two respects, actually: First, due to time pressures, pharmaceutical companies and research institutions around the world are using HPC to perform epidemiological and bioinformatical calculations and simulations that would take months or even years on traditional platforms. What’s more, the COVID-19 pandemic is also causing a massive overload of existing data infrastructure. As a result of this, streaming services like Netflix and Amazon Prime have already been forced to significantly reduce their data rate – and thus transmission quality – in order to avoid overloading the networks. So, demand for our HPC solutions and capacities is skyrocketing.

Q: What are the particular challenges involved in HPC?

AT: Due to the extreme computing power, HPC systems require many times more electrical energy and cooling than current computer systems. This presents enormous challenges for the entire IT industry, since the existing infrastructure of conventional data centers is in no way prepared to meet this increased need for power and cooling. Today, data centers consume power in the low double-digit megawatt range and are usually located in locations where they compete with established industry for energy supplies. In Texas, we’re building a capacity of 1.0 gigawatt – or 1,000 megawatts – by the end of the year, and from there we will scale up to 3.6 gigawatts. By way of comparison: Frankfurt am Main is the leading data center location in Europe, with more than three dozen operators sharing the market. But the entire Frankfurt am Main metropolitan area – along with industry, including the chemical plants in Höchst and the international airport – weighs in at a peak load of around 800 megawatts. The network capacities in Frankfurt would not even be enough to realize HPC projects of our caliber. Not to mention that the cost of electricity in Frankfurt is several times higher than at our locations in Texas or Norway.

Q: What HPC solutions does Northern Data offer that other vendors do not? Or, more succinctly: what is your USP?

AT: We have a wide range of unique and leading hardware and software solutions for HPC. For example, we have modular high-tech data centers that can be deployed practically anywhere in the world, so we can go wherever energy is available in abundance and at low cost. In addition, as mentioned before, in Texas we are building the largest data center in the world, where we will be able to offer HPC on a whole new scale. We’re doing so with our proprietary software which uses AI to control and operate hundreds of thousands of high-performance computers in parallel. Large-scale HPC is incredibly complex to do reliably, but we have cracked it. This is one of the reasons why we have been signing contracts with large blue-chip corporations.

Q: How quickly can you scale your business?

AT: Very quickly – but not as quickly as demand is skyrocketing.

Q: Who are your customers?

AT: Our first customers for the Texas facility are large multi-billion-dollar corporations. Our customer pipeline is mostly comprised of very large corporates because – from the customer’s point of view – these projects involve total budgets in the triple-digit millions.

Q: How do you expect the HPC market to develop in future?

AT: The HPC market has been in its infancy for the past few years but is currently skyrocketing across a range of diverse application areas. The demand for HPC exceeds supply many times over. To give you a feeling for this: According to our calculations, in the next 18 to 24 months, the demand for HPC capacity in blockchain computing alone will be thirty times higher than the resources we can build up in a year. And there is no player on the market that is currently building more capacity than we are. The fact that the market is exploding in many different areas is also illustrated by a statement by Eric Yuan, CEO of Zoom Video Communications, in an interview last week. When asked if he ever expected that a pandemic could change the way people work digitally today, he replied, “If I had known that, we would have reserved a hundred times more capacity.” That hit the nail on the head. We live in data-intensive and CPU-intensive times; demand is quite simply exploding.

Q: How is this development reflected in your figures?

AT: Only a few days ago, we published our guidance for 2020. In the current year, we expect revenues of EUR 120 million to EUR 140 million with EBITDA of between EUR 45 million and EUR 60 million. This is a very conservative estimate because it just includes contracts we have already firmly concluded. So, this is just an initial flavor of what we can expect in the coming years. After all, our data center is only just going into operation, and the rest of the year will also be marked by the ramp-up of capacity. We plan to have completed 1 gigawatt of capacity in Texas by the end of 2020 and can expand to up to 3.6 gigawatts on the site. Preparations for expansion at other important locations in Canada and Scandinavia are in currently in full swing.

Q: How much turnover do you make per gigawatt?

AT: Between EUR 400 and 500 million with a gross profit margin of 40 to 50 percent. This is because we can achieve an unprecedented level of automation thanks to the proprietary artificial intelligence we programmed. Our software takes care of operation, control and maintenance activities for the hundreds of thousands of high-performance computers. This enables us to deploy just 20 full-time employees per gigawatt and keep our fixed costs extremely low.

Q: Where do you see your company in the years ahead?

AT: We are in an extremely fast-growing market and have the opportunity to further expand our leading position in the coming years. The tremendous demand from our blue-chip customers give us a solid basis and sufficient tailwind to build up a multi-billion-euro business in the HPC segment that will become a global leader in this emerging industry. This is what I have been working towards for years. Together with my colleagues, I have had to overcome many technical obstacles in recent years and solve many challenges that initially seemed quite unsurmountable. Our current successes show me that we are on the right track. My goal is to seize this opportunity with our team and let our shareholders join us on an unprecedented journey to success.


Source link


Please enter your comment!
Please enter your name here