Deepwater's Doug Clinton: AI trade still intact, 'has a few more years to go'

Key Concepts

• NVIDIA (NVDA): A leading semiconductor company heavily involved in AI infrastructure, particularly GPUs.
• HBM (High Bandwidth Memory): A type of memory crucial for AI workloads, currently in high demand (specifically HBM Blackwell).
• Large Language Models (LLMs): AI models used for stock portfolio analysis and stock picking at Intelligent Alpha.
• AI Trade: The investment trend focused on companies benefiting from the growth of Artificial Intelligence.
• Frontier Models: The most advanced and capable AI models, currently dominated by Google’s Gemini and OpenAI’s models.
• Distribution: The method by which AI models are integrated into user workflows and daily life.
• Mag-7: Refers to the seven largest US technology companies (often including NVIDIA).
• Forward Earnings: A company’s expected future earnings.
• Free Cash Flow: The cash a company generates after accounting for capital expenditures.
• Idiosyncratic Buyer: An investor making decisions based on specific company analysis rather than broad market trends.

NVIDIA and the AI Landscape: An Analysis of Market Dynamics

NVIDIA’s Position and China Revenue:

Doug Clinton, Founder and CEO of Intelligent Alpha, stated that his firm continues to hold NVIDIA stock, utilizing LLMs for stock analysis. He highlighted the ongoing uncertainty surrounding NVIDIA’s potential revenue from China, specifically regarding the H200 and Blackwell chips. Clinton believes that once actual revenue from China is reflected in NVIDIA’s Profit and Loss (P&L) statement, the market may finally acknowledge this potential. He noted that Intelligent Alpha anticipated this possibility six months prior. He maintains a positive outlook on the AI trade, predicting several more years of growth.

Valuation and Incremental Demand:

NVIDIA is currently trading at 26 times forward earnings, with projected free cash flow of $160 billion in the coming fiscal year. Clinton acknowledges the market is anticipating a slowdown in growth as NVIDIA operates from a large base. A key challenge identified is finding the “big incremental buyer” for NVIDIA stock. While indexes are contributing through overweighting, identifying a significant, independent investor is crucial. He suggests that clarity on demand, particularly heading into 2027, is needed to attract this incremental investment.

Supply Chain Constraints & Demand:

The demand for AI infrastructure is exceptionally high. Micron is reportedly sold out of HBM Blackwell memory for all of 2026. Similarly, power segment companies like GE Nov are experiencing turbine shortages for the next two years. This indicates a significant bottleneck in the supply chain. Clinton emphasizes the need to understand demand projections for 2027 to instill investor confidence.

The Evolving AI Trade & Model Distribution:

The AI trade has experienced shifts throughout 2025. Initially, OpenAI was considered the leader, but Google’s release of Gemini 2.5 and subsequently Gemini 3 shifted the landscape, placing the two companies in a competitive position. Clinton predicts that 2026 will be defined by the distribution of AI models, rather than solely model performance.

Google vs. OpenAI: The Distribution Battle:

Google possesses a significant advantage in distribution due to its vast user base across platforms like YouTube, Maps, Gmail, and Search (billions of daily users). OpenAI, conversely, needs to forge strategic distribution partnerships. Potential partners mentioned include Apple and Amazon, with rumors of a possible Amazon investment circulating. Clinton stated, “I think those two companies [Google and OpenAI] are really the two that are vying for frontier model sort of winner in 26.”

Historical Context & Market Reactions:

The discussion referenced earlier market reactions, including the “DeepSea scare” and a shift in preference towards the Alphabet/Broadcom ecosystem over the Microsoft/OpenAI alliance. This illustrates the dynamic and often unpredictable nature of investor sentiment within the AI sector.

Technical Terms & Concepts

• PNL (Profit and Loss Statement): A financial statement summarizing a company’s revenues, costs, and expenses over a specific period.
• HBM Blackwell: The next generation of High Bandwidth Memory, designed to accelerate AI workloads.
• Turbines: Used in power generation, highlighting the energy demands of AI infrastructure.
• Mag-7: A commonly used term referring to the seven largest publicly traded technology companies in the US stock market.

Logical Connections & Synthesis

The conversation flows logically from an assessment of NVIDIA’s current position to a broader analysis of the AI market. It begins with NVIDIA’s valuation and potential revenue streams (specifically China), then expands to the overall demand for AI infrastructure and the challenges of sustaining growth. The discussion then pivots to the competitive landscape between Google and OpenAI, emphasizing the critical role of distribution in the future of the AI trade. The historical context provided demonstrates the volatility of the market and the need for careful analysis.

Main Takeaway:

The AI trade remains robust, but future growth hinges on factors beyond model performance. NVIDIA’s stock performance is tied to realizing revenue from China and attracting incremental buyers. The battle for AI dominance is shifting from model development to effective distribution, with Google holding a significant advantage due to its existing user base. Investors need to closely monitor demand projections and the evolving partnerships between AI developers and platform providers to navigate this dynamic landscape.