Sunday, October 4, 2026
AI Infrastructure · News & Analysis
Home › Chips & Hardware › Report
Chips & Hardware · Report

Cathie Wood claims AI inference costs are dropping 99.99% year-over-year, signaling rapid pricing deflation.

If accurate, extreme cost declines accelerate AI adoption but compress margins for inference-accelerator vendors and raise commoditization risks.
Trade pressSlicast · October 3, 2026 at 02:01 UTC · US · Source: tokenpost.com
importance 55

ARK Invest founder Cathie Wood says sharply declining AI inference costs could lift productivity and corporate earnings while pushing inflation below most investors' expectations. She describes this combination as "benign deflation."

Wood said the cost of running AI inference at the same performance level is falling 99.99% annually. Inference uses a trained AI model to produce responses or perform tasks.

She cited OpenAI's annualized revenue run rate rising from $20 billion to $70 billion as evidence that falling costs are accompanying rapid demand growth. An annualized run rate projects a current revenue pace across a full year and differs from reported annual revenue.

ARK Invest projects real GDP growth could reach the high single digits—a pace Wood acknowledges many would consider "crazy." She also said interest rates could rise in an environment of stronger real growth.

Read the original
Cathie Wood claims AI inference costs are… · Slicast