What’s happening? For one, Amazon, Alphabet (Google’s parent), Meta and Microsoft are spending tens of billions of dollars per year on artificial intelligence computing infrastructure, a number that is expected to reach $650 billion by 2026, according to a recent report citing corporate projections, most of it on data centers, graphics processing units (GPUs) and networking equipment.
Are we in the middle of an industrial revolution, or a massive bubble? Let me try to make some sense of this. (And yes, I’ll add some color that you won’t find on a quarterly earnings call). We are not talking about a couple of Excel spreadsheets with extra zeros.
We are talking about an AI infrastructure arms race – the equivalent of whole cities for AI, filled with whirring computers, GPUs stacked up like skyscrapers, and terawatts of electricity flowing through mega campuses.
That is what the Big Tech companies are building, and they are using Wall Street-scale money to do it.
Consider this projection: global tech spending is expected to reach $6 trillion this year, with data center systems alone – the key ingredient in the infrastructure needed for AI computing – expected to reach $650 billion.
This isn’t a niche business anymore. It is the underlying infrastructure for most of the AI-based services that are being built today.
Amazon’s stock price fell after it announced plans to spend $200 billion on capital expenditures – much of it tied to AI and supporting technologies – despite stronger-than-expected sales from its cloud computing and retail businesses.
Alphabet’s shares tanked after its parent company warned that its spending could double year-over-year as it spends more on data centers, AI chips and related technologies, even as revenue rose at Google’s search and cloud business.
“Why would anyone complain about spending?” you might ask. Well, it’s possible, but of course, to build AI capacity and invest in data centers, cloud computing and the rest. That could be a great thing!
We need more AI training and inference, for medical research, climate modeling, more intelligent customer service chatbots and the like. But there is also a real concern of an AI bubble.
You know, like a case in which there’s way more hype and capital than return on investment, a la the internet bubble in the late 1990s. Now it’s time for me to chime in.
On the one hand, all this infrastructure buildout is necessary, and companies are so hungry to be first to market that they are throwing money around like Monopoly money.
On the other hand, we know from experience that companies can’t just spend money without a clear path to making money back. (That’s how bubbles form.) This presents a quandary in the market, and this is what you’re seeing when these companies’ shares rise and fall.
Let’s dive a bit deeper. In addition to the dollar figures, all this spending is also shifting the global balance of power for components such as memory chips, AI accelerators and high-speed networking gear – not to mention power and cooling.
Memory chips, accelerators, networking gear, and more are now at the center of international disputes, supply chain blockages and price swings, never mind the energy and cooling to power these facilities.
So what does this mean? If you’re wondering whether these investments will bring us AI solutions the likes of Uber, or an AI bust the likes of Y2K, you’re not alone.
We’ll have to wait and see whether these companies can translate all this AI capacity into products and actual demand. But one thing is for sure. Big Tech is not pulling back on its investments anytime soon.
Whether this pays off or not, this will be an exciting couple of years for investors, developers and pretty much anyone who uses AI.