NVIDIA (NVDA) remains the primary core holding in the sector, backed by 97% year-over-year revenue growth and critical partnerships that secure the power and infrastructure needed to sustain computing dominance.

Upstream hardware suppliers like TSMC (TSM) for semiconductor manufacturing, alongside Corning (GLW) and Lumentum (LITE) for critical optical components, offer essential picks-and-shovels exposure to scaling compute demands.

To capitalize on the multi-decade energy and data center infrastructure boom, consider agile regional "NeoCloud" providers such as Nebius (NBIS) that are rapidly locking down constrained local land and power capacity.

At the application layer, allocate toward domain-specific AI software targeting high-impact commercial fields like autonomous driving and healthcare automation, where real-world deployment is accelerating fastest.

NVIDIA (NVDA)

• NVIDIA's business is expanding rapidly, with revenue growing 97% year over year driven by accelerating compute demand across frontier labs and enterprises.
• The company is positioning itself as a full-stack computing platform, managing the AI supply chain upstream and downstream—partnering with utilities, real estate developers, and infrastructure financiers like BlackRock and Goldman Sachs to secure land, power, and data center shells.
• NVIDIA runs models for all major AI players (OpenAI, Meta, Grok, Gemini, Anthropic) while also building domain-specific open-source models, including self-driving vehicle stacks (Alpamayo) and biological foundation models (ESM2, Protein Complexa).
• Jensen Huang addressed long-term competition and chip manufacturing:
  • Mentioned Elon Musk's planned TerraFab facility as a serious effort in domestic fabrication.
  • Predicted China will develop advanced native lithography systems by 2030, underscoring the urgency for the U.S. to maintain rapid execution and leadership.

Takeaways

• NVIDIA remains the dominant hardware and full-stack software foundation of the AI buildout, aggressively removing ecosystem bottlenecks (power, land, capital) to sustain long-term demand for its compute architecture.

AI Hardware Supply Chain & Manufacturing (TSM, GLW, LITE)

• Building global AI infrastructure requires a deep, pre-planned upstream hardware supply chain.
• Key suppliers explicitly noted as critical to scaling compute capacity include TSMC (TSM) for semiconductor manufacturing, Corning (GLW), and Lumentum (LITE) for optical and materials components, alongside advanced memory suppliers.

Takeaways

• Critical hardware enablers upstream from GPU designers represent essential infrastructure beneficiaries as compute deployment scales globally.

Data Center Infrastructure, Power, & NeoClouds (NBIS)

• Data centers and energy infrastructure were framed as the central economic engine of the next 20 to 25 years, likened to modern digital oil.
• Regional "NeoClouds" (such as Nebius (NBIS) and Firmus) are growing rapidly because they can move faster than traditional hyperscalers to secure local land, power, and facilities.
• Global demand is driving gigawatt-scale data center expansion across the U.S., Southeast Asia, Australia, and Europe.

Takeaways

• Power generation, utility capacity, and agile regional data center providers are primary beneficiaries of the physical constraints limiting AI expansion.

AI Software, Foundation Models, & Ecosystem

• $400 billion in venture funding has gone into AI-native companies over the past six months, with 80% building on open-source models.
• "Doomer" narratives predicting massive near-term job loss or existential civilizational risks were rejected as scientifically unfounded and counterproductive to technological progress.
• Narrow superintelligence is viewed as already here in specialized domains such as automated scan reading in radiology, autonomous driving, and computational protein synthesis.
• Both proprietary (closed) and open-source models will coexist, with open models serving as critical infrastructure for enterprise customization, data privacy, and global startups.

Takeaways

• Investment value in the application layer is expanding toward domain-specific use cases (such as healthcare, autonomous systems, and enterprise automation) powered by accessible open-source foundation models.