Interview, Fireside Chat, Conference Presentation
BlackRock's Tony Kim on AI's Next Winners? Chips, Memory, Robotics & Quantum
- Market Cap Shift: The global technology sector has undergone a fundamental value transformation from a software-centric model to a compute-hardware-centric model, with roughly $30 trillion in chips and hardware, $22–23 trillion in the "Mag 7," and $10+ trillion in software/services.
- Compute Density Explosion: Since the "AI era" began in 2023, the base layer of compute capacity has increased by 10,000x, elevating a $10,000 server from 2023 standards to a million-dollar server.
- Data Center Re-architecture: AI requirements are driving a redesign of data centers where data movement is shifting from kilometers (cloud) to millimeters (on-chip), causing exponential increases in bandwidth, power consumption, and heat density.
- Co-Design Paradigm: Leading labs are moving toward "co-design," tightly integrating silicon architecture with specific model parameters, exemplified by the new "Jalapeno" chip developed with Broadcom.
- The "Rampocalypse": A severe shortage in memory (RAM) is emerging as compute-intensive models increasingly require memory capabilities to mimic the human brain, creating a demand-supply mismatch with a 3–4 year construction lag for new fabs.
- Energy Infrastructure: Data center power demands are shifting toward gigawatt-scale facilities, driving investment in 800-volt architectures, solid-state transformers, and behind-the-grid nuclear solutions (SMRs) to manage power density.
- Quantum and Long-Term Bets: BlackRock maintains a decade-long horizon for frontier technologies, with consensus timelines for utility-scale error-corrected quantum computers, orbital data centers, and fusion energy converging on 2030.
- Robotics Landscape: China is accelerating in robotics with 130–140 companies and 30–40 potential IPOs expected this year, compared to only 0–3 in the US, leveraging low-cost manufacturing to potentially combine Western AI brains with Eastern physical bodies.
- Investment Thesis: The investment strategy prioritizes the "now" (3-year window) where 90% of capital is allocated to compute and memory shortages, while reserving asymmetric bets for 5–10 year horizons like space-based computing and quantum.
- Enterprise Transformation: Traditional companies are being restructured by Private Equity roll-ups to replace legacy layers with "token flow" architectures, embedding proprietary knowledge into context layers (ontologies) for AI agents.
- Future Market Drivers: Key focus areas for the next 12 months include continued proof of data center redesign, potential IPOs of major AI foundation labs, and tangible progress toward orbital data center deployment.
- Personal Leadership Philosophy: Kim cites "dead people" (historical figures like Churchill and Napoleon) and "kind people" (investors who took chances on his early career) as primary influences rather than traditional mentors.