Consider establishing direct equity positions in Micron (MU) and SK Hynix (000660.KS) to capitalize on their structural pricing power as dominant suppliers of High-Bandwidth Memory (HBM) essential for AI accelerators.

Diversify beyond software into the physical AI and AI infrastructure theme by targeting critical hardware bottlenecks, including power generation, electrical equipment, and advanced liquid cooling.

Track Rivian Automotive (RIVN) to capture upside from strategic manufacturing spinouts, such as industrial automation platform Mind Robotics, which unlocks non-dilutive balance sheet value.

Seek pre-IPO or secondary market access to high-demand private market leaders, specifically vertically integrated AI supercomputing provider Cerebras Systems and satellite connectivity giant SpaceX via Starlink.

Position in domestic energy storage leaders like Redwood Materials, which are primed to support power-hungry data centers as Western supply chain restrictions limit foreign battery competitors.

Cerebras Systems (Private / IPO)

• Cerebras has advanced from developing specialized AI chips to delivering full-stack supercomputing systems and operating specialized AI data centers.
  • The company operates with an estimated 40% to 50% gross margin, commanding strong valuation multiples by addressing complex computing bottlenecks in a massive total addressable market.
  • Represents a shift where vertically integrated hardware providers take over large portions of the compute supply chain rather than relying on legacy chip designs.

Takeaways

• Look for potential public market access or private secondary shares in Cerebras as high-performance compute demand accelerates beyond traditional GPU architectures.
• Evaluate hardware-centric AI companies based on their free cash flow generation and full-system integration rather than purely software-like gross margins.

SpaceX (Private)

• Highlighted as a premier example of full-stack physical industry dominance, successfully moving from launch systems to global satellite connectivity through Starlink.
  • Demonstrates the "multi-act" business model common in physical tech, where foundational infrastructure enables massive recurring-revenue downstream services.
  • Benefits from deep vertical integration and substantial government and commercial contracts.

Takeaways

• Monitor private tech funds or pre-IPO access vehicles that offer exposure to SpaceX, particularly as Starlink continues to capture commercial telecom market share.
• Use SpaceX as a benchmark for how capital-intensive physical tech companies can leverage core manufacturing infrastructure into high-margin service models.

Rivian Automotive (RIVN) / Mind Robotics

• Rivian collaborated with Eclipse to spin out Mind, a general-purpose robotics company designed specifically for manufacturing environments.
  • Rather than focusing on humanoid designs, Mind develops mobile, high-dexterity robotic platforms trained on proprietary industrial manufacturing data.
  • Carving out the technology enables external capital funding and allows the robotics entity to operate as a supplier across the automotive and industrial sectors without burdening Rivian's balance sheet.

Takeaways

• Track Rivian's strategic partnerships and ecosystem spinouts, which can unlock non-dilutive balance sheet value and accelerate operational automation.
• Focus on practical, non-humanoid industrial robotics companies addressing immediate warehouse and factory floor labor shortages over consumer-oriented humanoids.

High-Bandwidth Memory (HBM) Suppliers: Micron (MU), SK Hynix (000660.KS), Samsung (005930.KS)

• The global memory market remains an oligopoly controlled by Micron, SK Hynix, and Samsung, creating a structural bottleneck for AI infrastructure.
  • Massive capital expenditure requirements and advanced packaging technicalities create immense moats against new entrants.
  • High demand for High Bandwidth Memory (HBM) to support AI accelerators provides these producers with substantial commercial pricing power.

Takeaways

• Consider direct equity exposure to leading memory manufacturers (MU, SK Hynix) as long-term AI infrastructure buildouts require increasing ratios of high-performance memory per compute cluster.
• Watch for emerging hardware startups attempting to introduce alternative memory architectures or heterogeneous AI interconnects to lower data center costs.

Redwood Materials (Private)

• Led by former Tesla (TSLA) executives, Redwood Materials has expanded beyond lithium-ion battery recycling into manufacturing large-scale stationary energy storage systems.
  • Well-positioned to capture demand from the data center boom as regulatory policies restrict Chinese battery manufacturers (such as CATL) from supplying domestic utility grids.
  • Illustrates the value of multi-phase industrial expansion: transforming recycled feedstock into critical domestic energy grid infrastructure.

Takeaways

• Position in the domestic energy transition and battery supply chain, where deglobalization and supply chain security policies favor Western manufacturers.
• Monitor commercial partnerships between clean energy storage developers and hyperscale AI data centers needing independent power capacity.

Physical AI & AI Infrastructure (Investment Theme)

• Approximately 85% of global GDP (roughly $100 trillion) resides in physical industries like manufacturing, energy, defense, mining, and shipping, compared to relatively small software sub-sectors.
  • Data centers face critical physical constraints including power availability, thermal cooling capacity, and equipment lead times (e.g., gas turbines facing up to 5-year lead times).
  • Growth opportunities are shifting toward capital-intensive "picks and shovels" supporting computing demand, such as nuclear power, sodium-ion batteries, advanced HVAC/liquid cooling, and on-premises CPU appliances (such as Oxide).

Takeaways

• Diversify AI portfolios beyond foundational software models into core physical enablers: electrical equipment, industrial automation, and power generation.
• Seek out opportunities in on-premises infrastructure, as enterprise security, latency concerns, and data sovereignty reverse total reliance on public cloud environments.