How Does a Value Investor Think About AI?

AI users should ultimately create more durable investment value than the infrastructure companies benefiting from the current AI capital-spending boom.

• Rewey argues that AI data centers are not currently overbuilt because leading model developers still report a shortage of compute capacity, unlike the internet bubble’s excess fiber investment.
• In reviews of 2Q26 earnings and conference calls, companies emphasized using AI both to expand infrastructure and to reduce costs and increase revenue.
• AI infrastructure beneficiaries include data centers and suppliers of cooling, electrical systems, racks, cables, cement and contracting services, while the Atlanta Fed’s 3Q GDPNow estimate stood at 5.9% as of 8/4/26.
• Rewey expects infrastructure spending to face cyclical risks and difficult comparisons in late 2027 or 2028, even though current CAPX demand shows no sign of slowing.
• In 2Q26, several companies in financial services and consulting reported measurable cost savings from integrating AI into workflows, supporting a broader productivity opportunity.

Takeaway: Investors should look beyond AI infrastructure sellers toward companies using AI to improve operations, while recognizing execution risks, management differences and the possibility of both winners and losers.