The fate of purpose-built AI Data Centers is the same as Launch Pad 39. The Launch Pad 39 was purpose-built in Cape Canaveral to send the Saturn 1 to the Moon. After Apollo 17 it became a relic. AI data centers are Moonshots.
AI workloads are presently at the training stage of all data that has been created. This stage will pass and computing loads will go down to train only newly created data.
Land groups, developers, and PE funds do not want to say: We will need less computing power, fewer giant AI data centers, and less electricity because this spoils the boom. They want to give the impression that the present situation will last forever so the millions of dollars keep coming.
I will give you another example. When the Internet became interactive and 3G was born, a boom of huge expectations was created that attracted many non-telco players. Lots of IT people, who did not know telecoms, entered the field - they are still there- and it was easy to fool people.
But reality has this nasty habit of hitting a boom. Investors discovered that the R&D invested would not be recovered. One by one the players dropped out but the roadside.
Lucent, Nortel, Panasonic, and companies kept dropping -Siemens, Alcatel- until just Ericsson and Nokia remained.
The competition we see, paid by billions of dollars is to see: Which company will be standing and who will drop by the roadside?
Google and Facebook are scared of losing advertisement revenue and have no alternative but to keep pumping money.
Microsoft having lost social media to Google and Facebook sees the opportunity to steal their market share. And all of them fear a newcomer entering from a blind spot and running away with the money.
Any smart player would be seeking, under the radar, locations where it is cheap to train AI as it has excess electricity. Let the big boys throw wads of money at the problem spend.
The whole thing got a life of its own when Sam Altman threw a $7 trillion figure when pitching ChatGPT to the UAE.
His pitch was picked up by the WSJ in early February 2024.
People in the industry and the media started using that $7 trillion figure without stopping to analyze the real size of it.
$7 trillion is almost a third of the US GDP and can be compared with FY2022, the federal government spent $6.3 trillion.
Another comparison. As of April 2025, the combined market capitalization of the Magnificent Seven tech companies (Apple, Microsoft, Nvidia, Alphabet, Amazon, Meta Platforms, and Tesla) was approximately $15.4 trillion |
