As Big Tech reported its latest quarterly results to Wall Street last week it became very clear the ‘swing factor’ for investor comfort has become the performance of the respective cloud businesses of Amazon, Google and Microsoft. Cue some gnashing of teeth at Google’s Mountain View headquarters as poorly received cloud business revenues pushed the Google share price down almost 10% at one panicky point. It wasn’t that Google’s $8.4 billion of quarterly revenues were bad. It’s just that market expectations are very very high, and it’s all about beating those lofty expectation.
Microsoft did just that, posting intelligent cloud revenues of $24.2 billion. That big dollar number isn’t even the driver. It’s the growth, a whopping 28% increase on last year’s quarter compared to Google’s ‘disappointing’ 22%. This fired Microsoft’s share price 6% higher, joining Amazon, another cloud-driven 6% positive share price mover, in the cloud-happy zone. However, don’t let cloud services, ahem, “cloud” the big picture. The cloud is the big tech hosting environment for all of our business and personal digital apps and services but there’s a new kid in town. An intelligent one, or almost. Artificial Intelligence(AI) has its fear-mongers thanks to the rapid ‘creative’ impact of generative AI applications like Dall-E and ChatGPT but AI data science has been around since the 1950s. The key to the recent rapid evolution of AI has been big tech companies leveraging scale to reduce costs and huge advances in semiconductor chips’ computing power. This has had a massive knock-on effect on the data centre industry and its infrastructure footprint.
Data centres, like AI, are not new. Originally, they were the hubs used by telecommunications companies in the 1990s for hosting the internet. Then the iPhone and smartphone revolution came along and data generation exploded. Data centres became critical for storing this data and for processing capacity which brough big technology companies into the mix. The data centres became larger, housing state-of-the-art cooling systems, redundant power supplies and hi-tech security systems. Resilience was key, and SilverBack assisted in the ‘mission-critical’ construction of these data centres as the backbone of the digital economy. However, what we are witnessing in more recent projects for Microsoft in Gävle, AWS in Västerås(and Eskilstuna + Katrineholm), and Eco Data Centre in Falun suggest a step-change in the pace of data centre infrastructure build. The clue as to why was probably a few months after ChatGPT had raced to 100 million users in less than half a year.
In May 2023, the newly crowned trillion dollar poster child of AI, Nvidia Corp, upgraded their projections for demand (revenues) for their market-leading AI model training chips – known as Graphic Processing Units (GPUs) – by a whopping 50%. Yet, the bigger story according to Schroders’ analysts was “hiding in plain sight”. Nvidia in their call with Wall Street’s analysts explained the uplift to their projections by mentioning ‘data centres’ no less than 56 times. Clearly, the destination for these advanced GPU chips needed to be premium performance, secure and stable data centre environments. The cloud and data centres have become interchangeable proxies for the same growth story, AI. The size of the growth is up for debate given we are in the early stages of AI but there are estimates of data centre power consumption in the US alone, doubling by 2030 to 35 gigawatts. We will return to power consumption a bit later but let’s first deal with construction.
Microsoft has put aside $13.5 billion for the building of data centres to support cloud and AI growth. And the headlines keep coming. Check out the following:
Green breaks ground on two data centres in Zurich – Reuters
Japan Govt To Subsidize Half Of Y40-60 Billion Cost For Softbank’s Data Centre Construction – Nikkei
Amazon To Build Three Data Centres In Dublin – The Irish Times
Arguably, data centres are the most obvious digital shovels for the $15.7 trillion AI ‘gold rush’ contribution to the global economy by 2030(Source: PWC). However, in a climate emergency, the huge power consumption of these data centres is rightly becoming a sustainability issue. Thankfully, big tech is meeting the challenge head on with increased focus on using renewable energy sources and upgraded efficiencies. The key metric for efficiency is PuE or Power Usage Effectiveness and the average for data centres globally is 1.6. The highest possible efficiency is a PuE of 1.0 but Microsoft’s data centre in Gavle, Sweden, is hitting very impressive levels of 1.2. In fact, Microsoft has a goal of transitioning to 100% renewable power by 2025, and a further goal to becoming carbon negative by 2030. Other initiatives to partner with local power utility, Vatenfall, to increase efficiencies and a separate project to digitize Gavle’s port operations are further demonstrations of a regional government building on a sustainable future for data centres. Dublin is not going to miss out either.
Recent announcements by Amazon to build three data centres in Dublin have been accompanied by encouraging data on the impact of Amazon’s cloud business, AWS, on Ireland already. The report card looks pretty good:
AWS has contributed €11.4 billion to Irish economic output since 2012, and €2.4 billion in 2022 alone.
4,200 direct jobs with AWS and a further 10,000 jobs nationally, supporting skills training and partnerships with third level educational institutions.
Creation of world-class ecosystem. AWS data centres provide emplyoyment for 3,000 working with 500 AWS suppliers. Many of these firms have expanded abroad as market leaders in data centre construction materials and services.
Community education: The AWS Think Big Space promotes STEAM related learning and careers. Since launch, 7,000 students and 425 teachers have engaged with the programme.
AWS is on track to power its operations 100% with renewable energy by 2025 and is committed to be water positive by 2030. Currently AWS is managing to cool its data centres for 95% of the year without using any water, and just using outside air cooling systems.
It would appear that AI, data centre sustainability and community builds are all enjoying the right direction of travel. And, just to juice those positives further, what about “mind blowing” new chip processing speeds? Only last week, Davide Castelvecchi for Nature reported, “ A brain-inspired computer chip that could supercharge AI by working faster with much less power has been developed by researchers at IBM in San Jose, California. Their massive NorthPole processor chip eliminates the need to frequently access external memory, and so performs tasks such as image recognition faster than existing architectures do – while consuming vastly less power.” Scientists have described the energy efficiency as “just mind-blowing”. If so, it won’t just be Nvidia and Microsoft recording blow-out quarters. The global roll-out of data centre facilities might only be in its second innings. Telecoms first, internet next…… Skynet anyone?