BlackRock's Jay Jacobs: Why the AI Boom Is Still Just 0.8% of GDP

Key Concepts

• Thematic Investing: A strategy focusing on long-term structural shifts (AI, demographics, energy, geopolitics) rather than traditional sector-based allocations.
• AI Compute Intensity: The measurement of processing power required for AI tasks, with "token consumption" serving as a key metric for growth.
• Physical AI: The transition of AI from digital-only (chatbots) to physical applications like robotics and autonomous vehicles.
• Agentic AI: AI systems capable of performing multi-step tasks autonomously, significantly increasing compute demand.
• Infrastructure Bottlenecks: The physical constraints (power grids, real estate, raw materials) that limit the speed of AI data center deployment.
• Real-World Asset (RWA) Tokenization: The process of placing traditional financial assets on a blockchain to improve liquidity and settlement speed.

---

1. The Evolution of Portfolio Management

Jay Jacobs highlights that portfolio management has evolved from 1970s factor research to 1990s style boxes and GICS sector classifications. The current shift is toward a thematic lens.

• Thematic vs. Sector: Investors often mistakenly use the "Technology" sector to gain AI exposure. However, the tech sector includes software companies that may be negatively disrupted by AI. Thematic ETFs provide more granular, precise exposure to specific disruptive forces.
• Institutional Adoption: BlackRock’s 60/40 model increased thematic allocation to 7.5% by the end of 2025, while the average advisor model remains at 3.6%. This suggests an "under-allocation" to structural growth among retail investors.

2. The AI Buildout: Infrastructure and Economics

Contrary to market skeptics who fear a bubble, Jacobs argues the AI buildout is in its infancy and is fundamentally different from past speculative bubbles (e.g., the 90s telecom boom).

• Capex vs. GDP: US GenAI infrastructure spending is currently ~0.8% of GDP, significantly lower than the 4.5% spent on UK railroads in the 1860s or 2% on US electricity in the 1920s.
• Monetization: Unlike speculative telecom builds, AI compute is being monetized in real-time. Token consumption grew 17x in the last year, indicating that demand is currently outstripping supply.
• Earnings-Driven Returns: AI stock returns in 2025 were driven by earnings growth rather than multiple expansion, signaling that these companies are using free cash flow to build profitable infrastructure.

3. Infrastructure and Geopolitical Shifts

The report projects $100 trillion in global infrastructure investment by 2040, driven by national security and supply chain resilience.

• Defense Spending: Global military spending is growing, with a shift toward space-based systems (US procurement grew from $12B in 2020 to $34B). Defense is no longer just a sub-sector of industrials; it now encompasses software, raw materials, and rare earth metals.
• Grid Reliability: Data center developers are bypassing strained grids, favoring regions with high reliability (e.g., Arizona). There is a "structural mismatch" where data centers take 1.5 years to build, but transmission lines take 8–10 years, creating a bottleneck for cloud providers.

4. Digital Assets and Tokenization

• Bitcoin ETPs: The iShares Bitcoin Trust (IBIT) reached $70B in AUM in 341 days, acting as a bridge between traditional finance (TrifI) and decentralized finance (DeFi).
• Stablecoins: Transaction volume for stablecoins is outpacing spot crypto trading, suggesting a shift toward real-world economic utility rather than pure speculation.
• Tokenization: While currently focused on cash/stablecoins, the long-term promise of tokenization includes 24/7 trading, instantaneous settlement, and improved liquidity for private markets.

5. Frameworks for Long-Term Investing

Jacobs suggests three filters for evaluating generational mega-forces:

1. Technology Maturity: Is the technology proven?
2. Use Case: Is there a clear, scalable application?
3. Probability of Fulfillment: Is the market adoption inevitable?

Notable Quote:

"The building itself is the easy part. It’s everything else that goes into that building or plugs into that building that can be very hard." — Jay Jacobs, on the physical constraints of the AI data center buildout.

Synthesis/Conclusion

The global economy is undergoing a structural transformation where digital AI is colliding with physical constraints. Investors should move beyond broad sector indices and adopt thematic tools to capture growth in AI, defense, and infrastructure. The current AI rally is supported by real earnings and massive demand for compute, distinguishing it from historical bubbles. Looking ahead, the intersection of AI and healthcare, along with the rise of "Physical AI," represents the next frontier for long-term capital allocation.