Not content with upending the stock market, the workplace, academia, industry, and much of daily life, the AI revolution is also bringing about changes in the way we design and build infrastructure, in particular (but not exclusively) the vast data centres and compute farms that keep it all running. The era of hyperscale infrastructure is upon us.
We’ll get to the bits that apply more to cement and construction in a moment, but bear with me whilst we take a dive down the AI rabbit hole. For AI sceptics, there are concerns that the boom is all a vast bubble, fuelled by rampant speculation, much like the dotcom bubble of the early 2000s, with an economic sword of Damocles poised to fall just as soon as the world wakes up and sees through the illusion. And those concerns aren’t entirely without merit. For example: the AI compute giant, Nvidia, had a share price of ~US$0.50 back in 2015 when the company was primarily associated with developing graphics hardware for video games and niche industrial uses. Today, after a long-planned pivot towards AI, Nvidia’s stock price sits at over US$200, giving the company a market capitalisation of US$5.15 trillion, making it the world’s most valuable company and roughly equivalent to 16% of US GDP.
And if that wasn’t crazy enough, one only needs to look at the example of Allbirds. In a move that would make anyone with even a vague idea of what happened in October 1929 raise an eyebrow, this sustainable shoe company (yes, you read that correctly) made its own pivot to become an ‘AI compute infrastructure’ business. And the reaction from the market? An immediate 580% rise in the company’s stock price.
And then, of course, there’s the fact that many of the major AI services have yet to make a profit, despite huge uptake of their services from around the world. OpenAI, the company behind the ubiquitous ChatGPT, is valued at US$850 billion, yet made a US$14 billion loss this year despite revenues of US$20 billion. If I were to don my tinfoil hat briefly, I might be tempted to suggest that the long-term plan was to get everyone hooked on using the technology before massively ramping up the price once dependency has set in. One doesn’t become a tech billionaire by playing nicely, after all.
Anyway, that dependency does, I suppose, show how the concerns of an imminently bursting AI bubble might be overwrought. Unlike the dotcom era when a company simply announcing that it had a website suddenly made it stock market gold, the world is actually using AI and using it on a vast scale. Even Neom, Saudi Arabia’s much vaunted utopian city of the future, is now being reimagined as an economically self-sustaining data hub. All of which neatly leads me back to hyperscale infrastructure.
The huge infrastructure projects needed to support the world’s insatiable demand for AI are placing new challenges (and opportunities) at the feet of cement producers. Whilst these projects will need vast amounts of cement and concrete and represent valuable business, they will also test the expertise of suppliers, necessitating specialised concrete mixes and demanding the ability to operate under strict timelines while delivering consistency at unprecedented scales, all while meeting strict requirements for emissions reduction. It’s a brave new world for the cement industry.
