AI trade cools on capex concerns; Alphabet's TPUs gain traction against NVIDIA GPUs
Wall Street endured a volatile week as the AI trade cooled amid concerns over sustainable returns on massive AI infrastructure spending. The Nasdaq Composite fell 4.6% while the S&P 500 slipped 1.95%. Despite Micron's blockbuster earnings—revenue more than quadrupled year-over-year—investor sentiment wavered Friday after reports that OpenAI may delay its IPO until next year, raising fresh questions about whether hyperscaler funding for the AI buildout can sustain current momentum.
Within the sell-off, Alphabet emerged as a relative bright spot, with Google Cloud revenue projected to surge 64% this year to $96 billion, and analysts modeling above 50% growth for 2027. Google's custom TPU chips are increasingly seen as a compelling alternative to NVIDIA's dominant GPUs, offering 20-30% cost savings through superior efficiency (consuming 20-40% less energy than NVIDIA processors). Google Cloud's pricing advantage and Anthropic's early adoption as a customer are positioning Alphabet as a credible third pillar in AI infrastructure alongside cloud incumbents AWS and Azure.
The market's rotation—away from AI hyperscalers (Apple, Microsoft, Meta, Amazon down; NVIDIA down 8.6%) and toward suppliers (Micron +16% Thursday, Corning +160% since Oct 2025)—underscores a critical architectural insight: companies building the picks-and-shovels of AI (memory, fiber, power) are capturing more margin than the companies writing the checks for infrastructure capex.