Hot Take: The AI spending frenzy is not a growth story, it is a margin compression cycle disguised as innovation.
Big Tech is deploying tens of billions into AI infrastructure under the illusion of inevitable monetization, but the economics do not support the narrative. Compute costs are exploding faster than revenue models can stabilize, and usage based pricing is already facing resistance. The market assumes AI will behave like previous software cycles with high incremental margins, but the reality is closer to utilities with heavy fixed costs and commoditized output. This shifts the entire revenue profile from scalable software to capital intensive services with structurally lower returns on invested capital.
The margin pressure is immediate and visible in capex intensity. Data centers, GPUs, and energy contracts are locking companies into elevated cost bases that cannot flex downward if demand underwhelms. Labor costs are rising as well, with elite AI talent commanding premiums that erode operating leverage. Meanwhile competition is intensifying, not consolidating. Open models and aggressive pricing from multiple hyperscalers are eliminating any hope of durable pricing power. This is not a winner take all market, it is a race to the bottom on inference costs, which translates directly into EBITDA erosion across the sector.
Valuation remains anchored to outdated assumptions of software like margins and infinite scalability. That disconnect creates a valuation trap. Investors are underwriting multiples that assume margin expansion while the underlying cost structure is becoming heavier and more cyclical. The longer monetization timelines stretch, the more this turns into a cap table bloodbath where only the largest balance sheets can sustain the burn. Even then, returns degrade as capital intensity rises faster than revenue per unit of compute.
Investor Implication
Expect multiple compression as the market recalibrates AI from software narrative to infrastructure reality. Earnings revisions will trend downward as costs surface faster than revenues materialize. Capital discipline, not growth, becomes the defining factor.
Final Take: AI is not expanding margins, it is structurally resetting them lower.