Focus on the "AI Bottleneck" by investing in the power and storage infrastructure required to run models rather than the models themselves. Nebius (NBIS) represents a high-conviction opportunity as a specialized AI cloud provider, with current entry prices sitting near institutional levels. For high-speed data transfer, Lumentum (LITE) is a critical play as data centers transition from traditional copper wiring to advanced photonics and lasers. Investors should look toward Iris Energy (IREN) and Core Scientific (CORZ), which are repurposing distressed crypto-mining infrastructure into high-value AI data centers. To address the massive memory requirements of AI inference, prioritize NAND flash memory providers like Western Digital as supply shortages loom. Finally, protect your portfolio by using put options on overextended names like Nvidia (NVDA) and Broadcom (AVGO) to hedge against potential market volatility.

This analysis explores the investment strategy of Leopold Aschenbrenner and his Situational Awareness Fund, as discussed by ElioTrades. The core thesis focuses on the "Bottleneck Trade": investing not in the AI models themselves, but in the critical infrastructure (power, compute, and hardware) that AI requires to function.

Bloom Energy (BE)

• Context: Aschenbrenner identified a "power crunch" where AI data centers require massive amounts of electricity (1-5 gigawatts) that local utilities cannot provide quickly. • The Solution: Bloom Energy produces solid oxide fuel cells that allow data centers to operate off-grid and on-site using natural gas. • Performance: The fund accumulated shares at an average price of $100; the stock recently traded around $310.

Takeaways

• Speed to Market: Bloom allows data centers to go live in 90 days rather than the 5-7 years required for traditional grid upgrades. • Leverage: The fund used call options to turn tens of millions into billions, demonstrating the power of high-conviction, levered bets on infrastructure. • Current Status: While the 20x move may have passed, the company remains a fundamental player in solving the AI power bottleneck.

CoreWeave (Private/Direct AI Play)

• Context: Described as an "AI Hyperscaler," CoreWeave is a former crypto miner that pivoted to providing specialized GPU compute for AI. • Nvidia Relationship: They are considered Nvidia’s "golden child," receiving priority chip allocations before major tech giants like Google or Microsoft. • Growth: Revenue grew from $16 million in 2022 to a projected $8 billion in 2025, with a $30 billion backlog.

Takeaways

• Pure Play: Unlike Big Tech, CoreWeave sells only one thing: compute. • Risk Factors: High dependency on a few core clients (like Microsoft) and a critical reliance on their relationship with Nvidia. • Investment Theme: Look for "distressed" or pivot-capable infrastructure companies that can repurpose hardware for AI.

SanDisk (Western Digital Spin-off)

• Context: Aschenbrenner front-ran the shift from AI "training" to AI "inference" (the actual use of the model by consumers). • The Bottleneck: As AI agents serve billions of tokens, they require a "memory hierarchy." While HBM (High Bandwidth Memory) is popular, the massive scale of inference requires a "motherload" of NAND flash memory. • Price Action: The stock rose from a low of $30 to over $1,500.

Takeaways

• The Inference Wall: Investors should look beyond the chips used to build AI and focus on the storage and memory needed to run AI at scale. • Supply Shortage: AI is currently consuming NAND faster than factories ("fabs") can be built.

Lumentum (LITE)

• Context: As data processing speeds increase, traditional copper wiring "melts" or fails to keep up. Lumentum produces lasers (EML and DML) used in photonics/fiber optics. • Market Position: They operate in a duopoly, giving them significant pricing power as data centers transition from copper to light-driven data transfer.

Takeaways

• The "Wires and Cables" Play: As data rates pass 200 gigabits per second, photonics becomes a requirement, not an option. • Performance: The stock is up 143% on the year, driven by margin expansion and the hardware transition.

Crypto Miner Pivot (IREN, CORZ)

• Context: The fund invested in former Bitcoin miners that transitioned their facilities into AI data centers. • Top Picks: IREN (Iris Energy), up 772% in 12 months, and Core Scientific (CORZ), which recovered from bankruptcy.

Takeaways

• Asset Repurposing: Bitcoin miners already possess the power permits and cooling infrastructure needed for AI, making them "undervalued gems" during the transition. • Contrarian Success: Using "distressed" crypto assets to fuel AI growth was one of the fund's most profitable moves.

Nebius (NBIS)

• Context: A recent disclosure revealed this is now the fund's largest position (approx. $3 billion, or 15% of the fund). • Business Model: Similar to CoreWeave, it is an AI-centric cloud platform providing full-stack infrastructure and GPU clusters.

Takeaways

• Fresh Opportunity: Unlike Bloom Energy, the speaker suggests investors are currently able to enter this position at a price similar to Aschenbrenner’s entry. • Infrastructure Focus: Reinforces the theme that "AI Clouds" are the primary bottleneck for AI builders.

Hedging & Risk Management

• Context: Recent 13F filings show Aschenbrenner holding $9 billion in "puts" (bets that prices will fall) against Nvidia (NVDA), Broadcom (AVGO), Oracle (ORCL), and the Semiconductor ETF.

Takeaways

• Market Caution: Even the most bullish AI investors are beginning to hedge. This suggests we may be entering a "precarious" phase of the AI stock bubble. • Strategic Hedging: These short positions might be temporary protections against geopolitical risks (e.g., conflict in the Middle East) rather than a long-term bearish view on AI. • Investor Warning: Avoid "blindly blasting" cash into top AI names without understanding the supply chain or how to protect your downside.