Warren Buffett's Berkshire Hathaway identifies the bottleneck of the AI revolution, and it's not chips.
Exclusive information $BRK.B $GOOGL
Greg Abel, Berkshire Hathaway's new CEO and Warren Buffett's successor, points to a fact that the market is increasingly beginning to understand:
The expansion of AI data centers could make electricity one of the biggest growth drivers of the decade.
According to Greg: we all already "see and feel" the direct impact of artificial intelligence, and he even defined Google as a key player in the AI market. (Buffett's Berkshire is heavily invested in Google)
But the more interesting story is actually behind the scenes,
AI consumes electricity at a monstrous rate.
Data centers running AI models require enormous amounts of electricity, and as models become larger and more complex, and as technology companies build more and more computing infrastructure, the demand for electricity continues to rise.
For Berkshire, this is a great business opportunity.
Buffett's successor says that over the years he believed that electricity supply would become a significant "limiting factor" for AI growth.
In other words, it's possible that in the future there won't be a lack of computing power, but simply a lack of electricity to power it.
Figures from the state of Iowa in the United States illustrate how much the phenomenon is already here:
About 8% of the state's total electricity consumption last year came from data centers,
Consider the implication: almost a tenth of the state's electricity consumption already comes from infrastructure that operates data centers.
If the demand for AI continues to grow at a high rate, this figure could become much more significant.
For Berkshire Hathaway's energy and electricity division, Berkshire Energy, which operates electricity infrastructure in Iowa, the implication is clear:
More data centers require more electricity, and more electricity requires investments in generation and transmission infrastructure.
And this is exactly where Berkshire can win,
The race for AI usually revolves around headlines about Nvidia, chips, cloud, and models...
But Berkshire is looking at the layer beneath all of that,
You can build more data centers, you can order more chips, you can launch more models.
But without electricity to power all of this, this whole dream stands still.
Therefore, for Berkshire, the lack of electricity can become an economic asset for whoever holds the necessary infrastructure to solve it.
The digital revolution needs chips, the artificial intelligence revolution needs electricity in quantities we haven't known before.
And therefore, it's possible that the winner in the AI era won't be the one who built the smartest model or the most advanced chip,
but rather the one who will be able to supply it with the electricity to make it work.