 |
"Next Year's News This Week" the trillion-dollar promise: ON data centers and deception By Evan Anderson _______ Why Read: The AI race has led to the construction of more data centers than even many of us in the tech sector ever imagined possible. At the same time, the promises being made by companies intending to build out AI tools are proving to be beyond dubious. Read on to find out just who is building all this capacity, how it's going, and where it will go from here. _______
We calculate that companies across the compute power value chain will need to invest $5.2 trillion into data centers by 2030 to meet worldwide demand for AI alone. We based this figure on extensive analysis and key assumptions, including a projected 156 gigawatts (GW) of AI-related data center capacity demand by 2030, with 125 incremental GW added between 2025 and 2030. This $5.2 trillion figure reflects the sheer scale of investment required to meet the growing demand for AI compute power - a significant capital commitment that underscores the magnitude of the challenge ahead [...]. - McKinsey Quarterly (4/28/25)
Sundar Pichai said while the growth of artificial intelligence (AI) investment had been an "extraordinary moment", there was some "irrationality" in the current AI boom. - BBC News (11/19/25)
Woah, there. The AI market boom is in full force. Billions of dollars are shifting regularly in funding rounds that value individual companies at astounding levels, even as they continue to lose money: OpenAI, the market leader in consumer-use LLMs, nearly $500 billion. Anthropic, which overtook OpenAI to increasingly lead the enterprise market, $350 billion. Xai is hoping to raise a new round at a $230 billion valuation. Meanwhile, the concept of artificial intelligence itself seems to have morphed. The term is now used almost synonymously with large language models (LLMs) - a tool that increasingly has proven to not always be particularly "intelligent." The problematic difference prompted AI visionary Yann LeCun to announce that he was departing Meta on November 19 to start his own company. According to the BBC: Prof LeCun has suggested LLMs will be less useful in attempting to create AI systems that can match human intelligence. Instead, he wants to pursue what he called "advanced machine intelligence". It trains AI models primarily by using visual learning - trying to replicate how a child or a baby animal learns. That differs to LLMs, which are fed vast amounts of existing data, and then asked to generate a result based on the data and a prompt. Meanwhile, the creeping feeling of incredulity that one gets seeing these kinds of huge financial numbers tossed out casually for LLM buildouts is accelerated by the other costs. Data centers are required to power all these newfangled tools; and those data centers are wildly costly.
They're so costly, in fact, that US$3 trillion is already lined up to be spent between now and 2030. The McKinsey report quoted above, focused on trends for use of compute, posits that $5.2 trillion will be needed to meet those needs. That is a lot of moolah. Such astronomical costs put this buildout on par with the greatest endeavors of human history. World War II is estimated to have cost the United States $340 billion in 1945 dollars - around $6 trillion in today's dollars. That makes sense in its context: fighting a multifront global war with millions of combatants over four years. For that kind of cash, one should expect a grand endeavor. And the captains of LLM leaders today claim that that's exactly what's happening. "Entire industries will be reinvented," we are told. "Many people may lose their jobs. But in the end, it will be an achievement of epic proportions." Let's call this the "trillion-dollar promise": the idea that something awfully important would be required to justify a trillion-dollar expense. Money is only the beginning. The true cost of this massive expansion of the world's data-center infrastructure is multifold. Yes, cash for construction will be essential, but so, too, will be a massive amount of water to cool them, GPUs to stock them with, electricity to run them, and land to build them on. Each of these also comes with an opportunity cost: a lot is sacrificed in devoting that level of resources to one industry. To understand what this means, let's dive into some of these other costs before taking a look at who is asking for these resources and how their projects seem to be going.
|
