I was looking at the AI-stock trade expecting the obvious answer: chips. Then I checked what has to happen after the chip gets sold.

Nvidia’s latest quarter was huge — $96.2B revenue, with Data Center revenue hitting $89B, up 117% year over year.

Hold up.

The less obvious constraint may be electricity. The IEA says data-center power use jumped 17% in 2025, while AI-focused facilities grew even faster. Its current base case has global data-center electricity consumption roughly doubling from 485 TWh in 2025 to 950 TWh by 2030.

That changes how I look at the AI trade.

The bottleneck isn't necessarily another GPU order. It could be transformers, grid connections, cooling, power generation and the companies supplying them.

There’s a fair counterpoint: efficiency is improving quickly, and the IEA says power use per AI task is falling.

But if AI demand keeps scaling faster than infrastructure can be built, are investors underestimating the value sitting one layer below the AI headline?

#AIStocksWhatNext

$TAKE

$MET

$NVDA

#stocks #NVIDIA #Investing
#AIStocksWhatNext