The S&P 500 reached an all-time high, with stock valuations at levels not seen since 2000. This reflects a regime where equity prices are priced to yield only 1.8% annualized returns, not 5.2% as in 2019.
Alphabet’s Gemini app grew from 400 million to over 1 billion monthly users in under 12 months. The free consumer version drives adoption, with 73% of new users engaging with AI features in the app’s latest update. This expansion is not driven by marketing spend but by the integration of open-source models, reducing Google’s reliance on premium subscription revenue.
Amazon and Microsoft both reported cloud revenue growth of 24% year-over-year in Q2, with Alphabet’s cloud revenue rising 19% despite a 15% year-over-year decline in enterprise pricing. The gap between cloud growth and pricing power is narrowing, as Google’s AI infrastructure now serves 62% of the U.S. AI workloads in public sector applications.
Alphabet’s AI spending increased by $1.3 billion in Q2, with 85% of that allocated to model training infrastructure. This spending is not offset by cost savings in existing services. The cost of training new models has risen 40% since Q1, with new models requiring 30% more compute capacity than predecessors.
These developments create a structural pressure on Alphabet’s AI margins. The current pricing power in AI services is declining, as new entrants like Anthropic achieve full model deployment at 60% lower cost than Google’s internal models. The market now expects AI infrastructure costs to absorb 25% of total revenue by 2025.
Alphabet’s valuation is now 14% above its long-term earnings multiple. This premium is sustained by the expectation that AI adoption will drive revenue growth, not margin expansion. The company’s ability to pass on costs to customers is constrained by regulatory scrutiny of AI pricing, with three new filings in Q2 challenging AI service pricing in healthcare and finance.
The result is a pricing power erosion in core AI services. Alphabet’s net margin in AI operations is expected to fall from 28% to 21% in Q3, driven by higher infrastructure costs and competitive model deployment.
