Simon Naylor, SVP – Asia Pacific, Oper8 Global Group
In a recent post, we wrote about how data centres are helping to win races – literally – but there is a much bigger race against time going on right now: something the World Economic Forum is calling the data centre “gold rush”.
This gold rush is in response to the explosive growth of the digital economy, driven by the rise of AI. Investing in data centres and in the digital economy isn’t just delivering the computing infrastructure needed to power AI and associated capabilities; it’s also attracting further investment in complementary industries, bringing in capital, technology, skills and greater employment opportunities.
The industry is already shifting the needle.
Gartner estimated that worldwide semiconductor revenue in 2024 was US$626 billion, with that figure projected to increase to US$705 billion in 2025. Gartner attributes the key drivers for growth in the chip sector as the GPUs and AI processors used in data centre applications to meet the rising demand for AI and GenAI workloads. Maintaining that CAGR would see chip sector revenues more than doubling by 2032.
Reinforcing that, by 2032 the global data centre industry is also projected to more than double with a projected worth of over US$584 billion.
Lessons from history
Why is a gold rush a race against time? Using a real example helps to explain. Beginning in 1848, the California Gold Rush attracted more than 300,000 people, tripling the state’s population at the time. While it had an incredible impact on the economy and associated businesses at the time, most late arrivals made little or wound up losing money. The best claims had already been staked out, the gold was becoming more difficult to find and extract, and the prices for food, materials and all the equipment you needed to mine gold were a lot higher.
Late arrivals to the data centre AI gold rush will be competing for diminishing supplies of space, water and energy.
The International Energy Agency projects that electricity demand from data centres worldwide is set to more than double by 2030 to around 945 TWh, more than the entire electricity consumption of Japan today. That is exacerbated by AI. Alex de Vries-Gao, the founder of Digiconomist, has calculated that AI systems will account for nearly half of data centre power consumption by the end of this year.
On current trends, water is also going to be in short supply. Cornell University estimates that data centres powering AI applications could use up to 6.6 billion cubic metres of water globally by 2027, the equivalent of half of the UK’s current annual consumption.
Scarcity adds up to greater cost. Real estate speculators are already increasing the cost of land for data centres – up to six times their value, according to some observers.
McKinsey & Company estimates that by 2030, data centres will require US$6.7 trillion worldwide to keep pace with the demand for compute power. Data centres equipped to handle AI processing loads are conservatively projected to require US$5.2 trillion in capital expenditures – for land, materials, and site development; power generation and transmission, cooling, and electrical equipment; and chips and computing hardware.
McKinsey agrees that it’s a race, noting that companies that proactively secure critical resources — and, materials, energy capacity, and computing power — could gain a significant competitive edge.
Transformative potential of AI
In another study, McKinsey found that 92% of companies globally plan to increase their AI investments with the firm’s research sizing the long-term AI opportunity at US$4.4 trillion in added productivity growth potential from corporate use cases.
While we can’t ignore AI’s growth and its transformative potential for business, it’s clear that current data centre growth trends are unsustainable.
To win the AI race, while we can’t influence the real estate market, our advice is to design your data centre infrastructure and utilise advanced technologies so they have a direct impact on reducing your consumption of the three things that will be in short supply: space, water and energy.
That means more efficient power and cooling solutions; higher density racks and smaller overall data centre form factors to reduce their footprint; and operational support that optimises the performance of the facilities over the longer term.
Be smart in your deployment of smaller, more tactical and localised edge computing for AI applications that rely on low latency response or require more secure or private operations; utilise cheaper existing hyperscale platforms for more generic AI functions; and look to newer AI models and chip designs that require less computing resources and therefore reduced power and cooling.
The big winners in the AI race will be those organisations who can be smart and nimble in leveraging both their own technology platforms and hyperscale resources to be the best and fastest adopters of AI in their industries. Contact us now to find out how Oper8 Global can assist.