Denny Fish on AI Race: Models Leapfrogging & Infrastructure Demand

The AI Race and Model Advancement

• 💡 The pursuit of artificial general intelligence (AGI) is driving a race among companies like Google, Meta, Anthropic, and OpenAI.
• 🚀 We can expect continuous advancements where models leapfrog each other with each new iteration, as seen with GPT, Gemini, Claude, and XAI.
• 📈 Recent data from Nvidia's earnings and Gemini 3 indicate that scaling laws are intact, fueling aggressive demand for GPUs and TPUs during the training phase.

Investor Considerations in the AI Infrastructure Boom

• ⚠️ Investors should be aware that a cessation of model advancements would signal diminishing scaling laws and could slow infrastructure demand.
• 📊 Despite market fluctuations, the underlying expectation for AI infrastructure build-out remains strong, supported by hyperscalers raising CAPEX expectations.
• ⏳ Unlike the dot-com bubble, the current AI infrastructure build is expected to be prolonged due to natural governors like data center construction speed, permitting, and power availability.
• 🔌 Short-lived assets like GPUs and TPUs are only deployed once facilities are ready with power, leading to significant carrying costs if not utilized promptly.
• 🎯 The demand for GPUs remains exceptionally high, with ten entities in line for each one produced by TSMC, indicating we are just beginning the deployment phase into data centers.