Tech earnings guidance cuts accelerating - AI capex reality

Seeing a pattern emerge across tech earnings - companies beating on revenue but slashing forward guidance while AI capex keeps climbing. Makes me wonder if we're hitting the point where AI spending becomes a margin headwind rather than a growth story.

The AI capex cycle is starting to look like the fiber buildout in 2000 - massive investment chasing theoretical demand that may not materialize for years. When you're spending $50B on compute while guidance gets slashed, that's a classic capital misallocation red flag that usually ends badly for margins.

The way I'm seeing it in my tech holdings - MSFT and GOOGL are spending like drunken sailors on AI infrastructure while their core business guidance keeps getting trimmed. When capex is growing 40% while forward EPS estimates are falling, that's a classic margin compression setup that usually gets punished hard.

The capex-to-revenue ratio across mega-cap tech is hitting levels we haven't seen since the dot-com buildout, but this time it's concentrated in a handful of players with actual cash flows. My base case is we get 2-3 quarters of margin compression before AI monetization kicks in, but the market's pricing in immediate ROI that just isn't there yet.

The thing that's bothering me is when I dig into the AI capex numbers - these companies are essentially building capacity for demand that doesn't exist yet. My NVDA position is getting crushed because even they're warning about lumpy enterprise adoption while hyperscalers keep ordering chips they can't monetize.

The hyperscaler buildout is starting to mirror telecom capex in the late 90s - massive infrastructure investment chasing addressable markets that won't materialize for years. When AWS and Azure are burning $15B+ per quarter on AI compute while their growth rates decelerate, that's a classic capital cycle peak that ends with writedowns.

The divergence between AI capex and actual revenue generation is getting wider by the quarter - we're essentially watching a $200B+ investment cycle with maybe 18 months of proven ROI. When you layer in rising funding costs and margin pressure, this starts looking like a classic late-cycle overcapacity story.

The Fed's going to have to start worrying about deflationary pressure from this AI overcapacity - when you're building $200B of compute infrastructure that won't generate returns for years, that's going to weigh on tech pricing power and core PCE. Bond market's already positioning for the disinflationary wave.

The way I'm thinking about this - I'm trimming my entire cloud infrastructure book because when you're seeing 40%+ capex growth with falling guidance, that's a classic capital efficiency warning sign. Already cut my CRM and NOW positions this morning because enterprise software margins are about to get crushed.

The bond market's telling the real story here - when 10Y yields are grinding lower while tech capex hits record highs, that's pricing in a productivity boom that kills inflation expectations. We're essentially watching the Fed's 2% target get destroyed by AI deflationary forces they don't even understand yet.

Just got off earnings calls for three of my mid-cap tech names - same story everywhere. Revenue beats but 2024 guidance getting slashed while they're all doubling AI spend. When your FCF yield is dropping 200bps because you're building data centers for customers that don't exist yet, that's not growth investment anymore.

The real kicker is when you map AI capex against actual enterprise adoption curves - we're running 18-24 months ahead of demand realization. This is starting to feel like the classic late-cycle overcapacity trap where everyone builds for the same TAM that takes 3x longer to develop than projected.