Post

Recent liquidity shortages, combined with panic over uncertainty around capital expenditures, are creating a huge “reverse tech bubble”—even Walmart’s forward P/E has reached 45x.



This explains why the AI bubble “bursts” almost every year. U.S. investors have long enjoyed asset-light business models and simply can’t stomach businesses that need massive capital expenditures to make money. 99% of people who trade stocks don’t use AI at all, so they have no firsthand experience of its value. Everyone is used to seeing the steady monthly growth of SaaS and doesn’t understand that AI revenue comes in steps and bursts: if it jumps 60% in a month, they imagine that pace can continue for a long time; if it grows just 10% in a month, they say demand has hit a bottleneck.

But we need to keep a clear head. We shouldn’t assume that AI demand equals coding equals 300b just because the market is falling; nor should we ignore the fact that cloud providers are accountable to their customers and responsible for their own spending just because the market is falling. It’s like when people in 2024 used the price of a GPT subscription multiplied by the potential number of users to estimate OpenAI’s commercial potential. When the technology and product are still evolving rapidly, we shouldn’t mistake the initial product form for the end state.

The truly useful question now is: Could a market downturn, as in 2000, end up driving a contraction in the industry?

On the demand side, the view is that it won’t have much of an impact. The key difference isn’t that the big companies investing today are more responsible or smarter, but that AI already has real paying customers and verifiable productivity use cases. In 2000, if internet companies couldn’t get VC funding, they had no money to keep buying servers or leasing bandwidth. When telecom companies saw excess capacity, they cut orders, hurting upstream suppliers in turn. Today, a downturn in capital markets has only a limited effect on end-consumer spending on AI.

On the supply side, that’s very likely. Semiconductor veterans who have been through multiple order reversals will be more cautious about bringing on capacity and capital expenditures when markets fall, even when they see mind-blowing demand. If capacity expansion slows, the result will be a renewed worsening of supply bottlenecks down the road.
View Original
This page contains third-party content and does not constitute any advice, nor does it represent Gate's endorsement of such views. For details, please see disclaimer.
WMTWMT+1.54%


Add a comment
Add a comment

Comment 3
GateUser-508e3b65
Can BTC find support at the bottom?
RainbowFartTrader
Walmart at 45x, the endgame for tech stocks is consumer stocks (doge)
YieldFarmer
The idea of AI revenue coming in stair-step pulses is interesting, but cloud providers’ capex discipline hasn’t been that reliable historically either. Let’s wait for Q2 guidance