Rick Sherlund on Enterprise AI Adoption, LLM Stress, and Data Center Demand

Shift in AI Trade Dynamics

• 💡 The AI trade is expected to move from broad enthusiasm to a more selective environment in the new year.
• 🎯 In 2026, the trade is anticipated to broaden beyond hardware, with a significant comeback expected for SaaS companies that have adopted AI and adapted their business models.
• 🚀 A robust IPO market is predicted, driven by scaled private AI companies and SaaS firms that missed the 2021 peak.
• 🤝 Mergers and acquisitions will be a major theme as large enterprise companies seek to integrate AI through acquisitions.

Enterprise Adoption and LLM Strain

• 🧠 While AI has been a novelty for consumers and beneficial for knowledge workers in research and summarization, it's now becoming integrated into business processes and workflows.
• ⚙️ The use of agents and complex workflows will dramatically increase the amount of inference required, potentially stressing Large Language Models (LLMs).
• ⚠️ LLMs are being asked to perform reasoning tasks they are not inherently built for, often relying on brute force GPU power.

Data Center Demand and LLM Business Models

• 📈 The explosion in inference demand will create enormous demand for data centers, with supply constraints expected to persist.
• 📊 The LLM market is highly competitive, with open-source models from various players, suggesting leaders will move up the stack rather than solely monetizing the LLM itself, similar to Microsoft with Windows or Oracle with databases.