Investors should rotate away from AI software "spenders" like Microsoft and Google to focus on the "receivers" of infrastructure spending, specifically in energy, data centers, and physical hardware. High-conviction plays in the "defensive side of AI" include independent power producers like Vistra (VST) and electrical infrastructure leaders like Eaton (ETN) to hedge against semiconductor volatility. Consider taking profits on parabolic memory stocks like Micron (MU) and reallocating into "platform" or interconnect plays such as NVIDIA (NVDA) and Marvell (MRVL). With global oil inventories drawing down, Oil represents a mispriced macro risk that could provide significant upside as inflationary pressures return. Finally, accumulate Bitcoin (BTC), Ethereum (ETH), and Coinbase (COIN) now, as they provide the essential 24/7 payment rails required for the upcoming shift toward autonomous AI agents.

This analysis explores the key investment themes from Jordi Visser’s latest update on the AI secular trend, emerging bottlenecks, and the shifting global macro regime.

Artificial Intelligence Infrastructure (The "Physical World" CapEx)

Visser argues that we are only 12% to 18% of the way through a multi-year, once-in-a-lifetime secular trend. He compares the current AI build-out to the China commodity bull market of 2002–2007.

• Shift from Software to Physical: Unlike previous software cycles, this requires massive physical infrastructure: HBM (High Bandwidth Memory), chips, liquid cooling, copper, fiber, and power substations.
• The Bottleneck Risk: The primary risk is not a "bubble" popping, but physical bottlenecks. Supply chain disruptions, labor shortages, and high material costs are creating "lumpy" revenue recognition.
• The "Spenders" vs. "Receivers": Visser suggests rotating away from the "spenders" (Hyperscalers like Microsoft, Google, Amazon, and Meta) and focusing on the "receivers" of that CapEx.

Takeaways

• Focus on the "Five-Layer Cake": Investment should target energy/power, data centers, specialized chips, and physical infrastructure.
• Monitor the "Dodge Momentum Index": This tracks non-residential building projects; it shows that data centers are currently the only major growth engine in commercial construction.
• Anticipate Volatility: Expect "lumpy" earnings. Just because a company has massive orders doesn't mean they can deliver them on a linear schedule due to grid and supply constraints.

Semiconductors & Memory (NVDA, MU, MRVL)

A significant portion of the discussion focused on the divergence between different types of semiconductors and the potential "exhaustion" of the memory trade.

• Micron (MU): Visser has reduced his position in MU. He notes that while earnings are great, the chart looks "parabolic" and retail sentiment is reaching a fever pitch (highlighting the Roundhill Daily Inverse Generative AI Strategy ETF and other retail-heavy vehicles).
• NVIDIA (NVDA): Remains the "platform trade." While its multiple has actually compressed over two years due to massive earnings growth, Visser warns of the "Vera Rubin" build-out complexity.
• Marvell (MRVL) & Intel (INTC): Visser expresses a preference for MRVL and INTC (specifically when INTC was under $110) over memory names, citing better moats in optical interconnects and CPUs.

Takeaways

• Profit Taking: Consider reducing exposure to names that have had "eight-bagger" runs in a year, as the risk-reward profile shifts.
• Look for Moats: Memory is a "bottleneck trade" (commodity-like), whereas NVIDIA is a "platform trade." Platforms generally offer more long-term stability.
• Efficiency Risk: A major risk to memory is "model efficiency." If AI models figure out how to run on 1/10th of the memory, the current demand forecasts will collapse.

The Power & Industrials Sector (VST, ETN, CAT)

Visser identifies Industrials as the "defensive side of AI" and a critical component of his thematic portfolio.

• Independent Power Producers (IPPs): Names like Vistra (VST) are seen as essential. Visser views these as a way to reduce "beta" (volatility) while staying exposed to the AI theme.
• Global Correlation Break: He notes a "historic breakdown" between software and semiconductors, and between U.S. and Asian industrials.
• The Grid Problem: The U.S. power grid is a major bottleneck. Companies involved in electrical infrastructure (like Eaton) and heavy machinery (like Caterpillar) are beneficiaries of the data center build-out.

Takeaways

• Defensive Rotation: If the high-flying semiconductor names correct, money is likely to rotate into the "defensive AI" names like Vistra.
• Watch Korea and Japan: These markets are leading indicators for the industrial side of AI. Visser notes that Japanese construction and machinery sectors are currently breaking below their 200-day moving averages—a warning sign for global momentum.

Energy & Macro Risks (Oil & Inflation)

The "regime shift" Visser describes involves a return to a 1970s-style environment characterized by rising oil prices and persistent interest rates.

• Oil Inventories: Visser highlights a massive draw in global and U.S. oil inventories. He believes the market is ignoring a potential supply shock.
• Interest Rates: The market has shifted from expecting three rate cuts to potentially one rate hike. This "tightening liquidity" environment usually pressures high-valuation stocks.
• Sovereign Debt: High interest rates combined with massive national debt create a "fragile" macro backdrop that could lead to a sudden momentum unwind.

Takeaways

• Bullish Oil Sentiment: Visser recommends listening to experts like Art Berman; he suggests that oil is currently a "mispriced" risk that could spike if inventory draws continue.
• Inflationary Inputs: PPI (Producer Price Index) for inputs like silver, copper, and energy are rising, which will eventually squeeze profit margins for S&P 500 companies outside of the "Mag 7."

Crypto & The "Agentic" Future (BTC, ETH, COIN)

Visser links the next phase of AI (Agents) to the necessity of a redesigned financial system.

• AI Agents: Google I.O. signaled a shift from "chatbots" to "agents" that do work. Agents will require 24/7, automated payment rails.
• The Crypto Connection: Visser believes the "handoff" from AI infrastructure to the "Global Financial System redesign" (Crypto/Blockchain) is coming.
• Coinbase (COIN) & Stablecoins: He points to Coinbase and the growth of stablecoins (like Circle/USDC) as the primary beneficiaries of an "on-chain" economy driven by AI agents.

Takeaways

• Accumulate on Weakness: Visser views the current lull in Crypto as similar to the "pre-discovery" phase of the AI memory trade.
• The "May 2025" Outlook: He predicts that by next year, investors will look back and wonder why they didn't buy Ethereum (ETH) and Bitcoin (BTC) as the "agentic world" took off.