Why the Markets Are Pricing AI Wrong | Gavin Baker

AI infrastructure demand is accelerating despite recent market volatility, driven by a fundamental disconnect between spot and contract GPU pricing. While market sentiment suggests over-earning, hyperscalers and AI-native companies continue to scale compute capacity, as token demand remains robust across both frontier and open-source models. The competitive landscape functions as a high-stakes game where breaking long-term supply agreements risks terminal loss of market share, forcing sustained investment. Although credit market tightening presents a potential hurdle, the shift toward operating cash flow funding and the emergence of disaggregated inference architectures—such as SRAM-based accelerators—strengthen the ROI case for AI. Regulatory policy remains the primary existential risk, as the industry struggles to effectively communicate the economic and societal benefits of data centers to the public.