November reflected broad infrastructure acceleration and deepening capital concentration as Hyperscalers, model labs, and investors expanded long-term AI commitments. OpenAI secured a $38B, seven-year AWS cloud agreement to lock in Nvidia GPU supply, while Microsoft advanced $9.7B in partnerships with IREN and Lambda to extend multi-region training capacity. Meta and Anthropic added more than $650B in collective U.S. data-center expansion plans, underscoring how AI infrastructure investment is becoming a macroeconomic lever. The TPU-versus-GPU debate also intensified as Google’s Gemini 3 training on TPUs shifted perception of large-scale model economics, with Alphabet rising 4.4% and Nvidia slipping 3.9% on November 25th’s trading, amidst a 30–40% increase in TPU utilization across cloud providers.
Financing momentum stayed robust with $17.8B raised across 133 transactions, led by Project Prometheus ($6.2B) and other large rounds across coding, infrastructure, and physical AI. Anthropic reached a $350B valuation with up to $30B in Azure compute commitments, while frontier models like GPT-5.1, Gemini 3, Claude Opus 4.5, and Kimi K2 Thinking marked another leap in multimodal reasoning and performance. Overall, November captured a decisive shift toward scale, more capital, more compute, and clearer evidence that AI’s competitive edge is now being defined by infrastructure reach and ecosystem integration.
Stepmark Partners 