Is the AI rally still in early innings?

Key Concepts

• AI Infrastructure: The foundational hardware (processors, memory, servers) required to support AI development.
• DRAM (Dynamic Random Access Memory): A type of semiconductor memory essential for high-speed data transfer in AI servers.
• AI Capex (Capital Expenditure): The massive investment spending by corporations to build out AI-capable data centers and hardware.
• Compute Scarcity: The current market imbalance where the demand for processing power and memory significantly outstrips supply.
• Token Pricing: Used as a proxy for the cost of computing power; rising prices indicate high demand and limited supply.
• Free Cash Flow (FCF): The cash generated by companies after accounting for capital expenditures, which allows for organic growth without debt.

---

1. The Evolution of AI Investment

Brian Mulberry, Chief Market Strategist at Zacks Investment Management, highlights a shift in AI investment focus. While initial interest centered on hyperscalers and advanced processors (like Nvidia), the current "rolling wave" of investment has moved toward memory chip manufacturers.

• The Disparity: Processing speeds have increased 3x over the last year, while memory capacity has only grown 1.5x. This creates a bottleneck that necessitates a massive increase in memory hardware.
• Server Architecture Shift: Modern AI servers have transitioned from using 8 memory chips to 96, representing a 12x increase in demand for memory components per unit.

2. Market Dynamics and Economic Context

• Market Resilience: Despite geopolitical tensions in the Middle East and rising oil prices, the market remains resilient. Mulberry notes that while fuel costs may impact industrial and material sectors, the market is not yet at "maximum stress" (e.g., oil is not exceeding $100/barrel).
• The "Tech Stack" Effect: The AI boom is creating a ripple effect throughout the tech stack. Software companies are reporting that their revenue growth is currently limited only by the availability of computing power, reinforcing the necessity of hardware infrastructure.

3. Key Players and Financial Health

• Memory Manufacturers: Companies like Micron, Western Digital, and SanDisk have seen significant year-to-date returns (200%–400%) due to the 12x surge in demand for their products.
• Financial Independence: A critical factor for these companies is their ability to fund expansion through Free Cash Flow. For example, SanDisk has moved from zero to approximately $40 billion in cash on hand. This allows them to expand manufacturing capacity organically without relying on debt financing or external partners.

4. Nvidia’s Strategic Expansion

Mulberry identifies Nvidia’s move to integrate its high-performance processors into PCs as a "game-changer."

• Impact: This transition from enterprise-scale "Blackwell" servers to retail-sized units is positively impacting partners like Dell and Microsoft.
• Leadership: Nvidia remains the "gold standard" for processing speed, and their ability to scale this technology down for consumer hardware is a significant market development.

5. Future Outlook and Strategic Insights

• Durable Growth: Mulberry argues that the AI investment cycle has "durable legs" because the industry is still in the early stages of development. The true capabilities of AI will remain unknown until the current computing infrastructure (data centers) is fully built out.
• Investment Thesis: He suggests that the "boat has not been missed." Given that the demand for these components is expected to persist for the next 3–5 years, he views current valuations in the memory sector as attractive.
• The Bottleneck: The primary challenge remains the ability of manufacturers to scale production capacity to meet the 12x increase in orders.

Synthesis/Conclusion

The AI market is currently transitioning from a focus on software and primary processors to a critical phase of infrastructure expansion, specifically in memory capacity. The "bottleneck" of compute scarcity is driving massive capital expenditure, which is being met by memory manufacturers with strong balance sheets. Because these companies can self-fund their growth and the demand for computing power is expected to remain high for several years, the AI infrastructure sector presents a long-term, durable investment opportunity.