Tom Lee Says AI Infrastructure Spending Is Approaching the Scale of the Internet and Mobile Buildout

Fundstrat’s Tom Lee says the scale of AI infrastructure investment is already approaching that of the entire internet and mobile buildout.
Lee estimates that the combined internet, mobile network and handset buildout was worth roughly $35 trillion in 2026 dollars.
He says AI infrastructure spending is already around $30 trillion.
That scale naturally raises concerns that the AI investment cycle may already be close to a peak.
Lee argues the opposite.
He compares the current setup to 1997, when the internet infrastructure boom was already well underway but Cisco did not reach its eventual peak until 2000.
Importantly, the path was far from smooth.
Cisco suffered two separate declines of roughly 50% before the cycle ultimately topped.
The comparison suggests that even if AI remains in a powerful long-term investment cycle, investors should still expect severe corrections along the way.
For $NVDA and the broader AI infrastructure trade, the key question is whether rising spending continues to produce enough productivity, revenue and cash flow to justify the scale of capital being deployed.
Lee’s thesis is that large spending alone does not prove the cycle is near exhaustion.