This is Bigger Than Nvidia. These 5 Stocks Win the Next AI Boom.

Key Concepts

• AI Infrastructure Buildout: The physical, industrial-scale expansion required to support AI, including power, cooling, and connectivity.
• Choke Points: Critical bottlenecks in the AI supply chain where demand significantly outstrips the current capacity to manufacture, build, or approve infrastructure.
• Hyperscalers: Large-scale cloud providers (Microsoft, Google, Amazon, Meta) driving the massive capital expenditure in AI.
• Picks and Shovels Strategy: Investing in the essential underlying infrastructure providers rather than the end-user software or primary chip designers.
• Base Load Power: The minimum level of electricity demand required to be met 24/7, essential for data center operations.

---

1. The Shift to Physical Infrastructure

Keith Kaplan, CEO of Tradesmith, argues that the AI growth narrative has shifted from the "GPU wave" (Nvidia) and the "Software wave" (Salesforce, Adobe) to the "Physical Buildout wave."

• Investment Scale: Hyperscalers are projected to spend over $700 billion on AI infrastructure in 2025, with projections exceeding $1 trillion by 2027.
• Industrial Significance: This represents the largest sustained surge in industrial infrastructure spending since the post-WWII era.
• The Problem: Demand has outrun the world’s ability to manufacture and approve projects, creating physical "choke points" that limit the flow of AI development.

2. The Five AI Choke Points

Kaplan identifies five critical areas where infrastructure is "bending" under demand:

1. High Bandwidth Memory (HBM): Essential memory stacks required to feed GPUs.
2. Photonics/Interconnects: Fiber-optic technology required to connect thousands of chips, as copper cabling cannot handle the data transfer speeds.
3. Thermal Management: Advanced liquid cooling systems required to manage the extreme heat generated by high-wattage AI chips.
4. Power Generation: The need for massive, 24/7 base load power (natural gas and nuclear).
5. Electrical Grid/Transmission: The physical hardware (transformers, switchgear) required to deliver power to data centers.

3. Featured Companies and Market Analysis

| Company | Ticker | Choke Point | Key Detail | | :--- | :--- | :--- | :--- | | Micron | MU | HBM | Only American producer of HBM; output committed through 2026. | | Coherent | COHR | Photonics | Leading supplier of optical transceivers; received a $2B investment from Nvidia. | | Vertiv | VRT | Thermal | Dominant supplier of coolant distribution units; essential for liquid cooling. | | Constellation Energy | CEG | Power | Largest US nuclear operator; reviving Three-Mile Island for Microsoft. | | Eaton | ETN | Grid | Power management giant; provides transformers and switchgear; multi-year backlogs. |

4. Key Arguments and Perspectives

• The "Choke Point" Analogy: Kaplan compares the supply chain to a pipe; regardless of how much "water" (demand) is upstream, the flow is restricted by the narrowest part of the pipe.
• Long-Term Horizon: Unlike the GPU wave, which Kaplan considers largely priced in, the infrastructure buildout is a 5-year mega-trend.
• Volatility as Opportunity: Kaplan dismisses recent market pullbacks (e.g., the 10% drop in Vertiv or the volatility in Micron) as profit-taking by algorithmic traders, suggesting these are entry points for long-term investors.
• The "Physics" Argument: Air cooling is described as "dead" for modern AI racks, which can draw 1,200 watts per chip. Liquid cooling is 3,500 times more efficient, making companies like Vertiv essential.

5. Notable Quotes

• "The money isn't going into software. It's not going into marketing. It's actually pouring into football field-sized buildings, gigawatt-scale power systems, and transmission lines that do not exist yet." — Keith Kaplan
• "The largest fortunes are going to go to the companies that are supplying what the visible players cannot function without." — Keith Kaplan

6. Synthesis and Conclusion

The AI growth story is transitioning from speculative software and chip design to a massive, multi-year industrial construction project. The primary investment opportunity lies in the "picks and shovels"—the companies providing the memory, optical connectivity, liquid cooling, nuclear energy, and electrical grid hardware. Despite significant stock price appreciation, Kaplan maintains that the demand for these physical components is so high that the market has not yet fully "baked in" the long-term growth potential, making current pullbacks attractive for long-term investors.