Investors should rotate capital out of "chip giants" like NVIDIA (NVDA) and Broadcom (AVGO), as the market has largely priced in GPU growth, and instead focus on the physical infrastructure powering AI. Bloom Energy (BE) is a high-conviction play for on-site power generation, offering a "plug-and-play" solution for data centers to bypass the overstressed electrical grid. Core Scientific (CORZ) provides a vertically integrated opportunity to own the energy and infrastructure required by major AI cloud providers. You can also target Bitcoin mining stocks that are repurposing their land and power permits for AI data centers, as these physical assets are being rapidly repriced. Finally, consider a bearish position on IT outsourcing firms like Infosys (INFY), which face significant displacement risks as AI automation replaces traditional offshore labor models.

Bloom Energy (BE)

Leopold Aschenbrenner has made this his largest conviction play, representing approximately 20% of his entire portfolio. The company specializes in solid oxide fuel cells that generate electricity on-site from natural gas.

• On-site Power Generation: These "Bloom Boxes" allow data centers to bypass the traditional energy grid, which is currently overstressed by human and AI demand.
• Efficiency: The technology converts natural gas directly into electricity via fuel cells, avoiding the clunky heating/cooling processes of traditional turbines.
• Financials: The company reportedly has a $20 billion demand backlog. Revenue grew ~34% in 2024, with a projected 40% increase in 2026.
• Scalability: They are modular and can be deployed quickly; the company is set to produce two gigawatts of capacity this year.

Takeaways

• The "NVIDIA of Energy" Potential: As AI labs (OpenAI, Anthropic) and hyperscalers (Microsoft, Meta) face energy shortages, Bloom Energy provides a "plug-and-play" solution to power massive GPU clusters.
• Grid Independence: Investing in BE is a bet on the thesis that the current electrical grid cannot scale fast enough for AI, making off-grid, on-site generation a necessity.

CoreWeave (Private) / Core Scientific (CORZ)

Aschenbrenner has shifted heavily into "NeoClouds"—specialized infrastructure providers that handle the physical complexities of AI.

• CoreWeave: He added $300 million this quarter, bringing his total estimated position to $800 million. CoreWeave provides the rack servers, cooling, and engineering required to run high-end GPUs.
• Core Scientific (CORZ): He owns roughly 10% of this company. It serves as a primary supplier for energy grids and infrastructure to CoreWeave.
• The "Levered Bet": By owning both the cloud provider (CoreWeave) and its infrastructure supplier (Core Scientific), he has created a vertically integrated bet on AI hosting.

Takeaways

• Infrastructure over Chips: The focus has moved from buying the chips (NVIDIA) to the companies that know how to house, cool, and power them.
• Activist Potential: With a 10% stake in Core Scientific, Aschenbrenner is moving into "activist investor" territory, potentially influencing company decisions to favor AI transitions.

Bitcoin Mining Sector (Various Tickers)

The portfolio shows a massive pivot into Bitcoin mining companies, but notably not for the purpose of mining cryptocurrency.

• The "Liquor License" Strategy: Mining companies already possess two things AI labs desperately need: Real Estate and Power Permits.
• Speed to Market: Obtaining new power permits can take years. Buying a mining company allows an AI firm to "gut" the crypto operations and immediately repurpose the power capacity for AI data centers.
• Asset Play: This is a "real world" bet on physical land and energy contracts rather than digital currency.

Takeaways

• Pivot Opportunity: Investors should look for Bitcoin miners that are actively transitioning their facilities into AI data centers, as these assets are being repriced for the AI boom.

NVIDIA (NVDA), Broadcom (AVGO), TSMC (TSM), Micron (MU)

In a surprising move, Aschenbrenner has exited or significantly reduced his positions in the major "Chip Giants."

• Priced In: The thesis is that by late 2025/early 2026, the market will have fully priced in the value of GPUs.
• Profit Taking: He reportedly made significant gains by selling $300 million in NVIDIA put options, profiting from recent price volatility.
• Intel (INTC) Exception: He sold his Intel equity but kept a "massive long position" open (likely via options) to free up liquidity for energy plays.

Takeaways

• The "GPU Peak": This represents a bearish sentiment on the short-term growth rate of chip stocks compared to the undervalued energy sector.
• Rotation: The "smart money" is rotating out of the "shovels" (chips) and into the "ground" (infrastructure/energy).

Infosys (INFY)

Aschenbrenner has taken out a massive short position against this IT outsourcing giant.

• AI Displacement: The bet is that new AI coding and automation tools (like Claude Code or GPT-5) will render cheap offshore IT labor obsolete.
• Business Model Risk: Companies like Infosys rely on human labor arbitrage; if AI can do the same work for near-zero cost, their business model collapses.

Takeaways

• Bearish on Outsourcing: This is a high-conviction bet against traditional "human-intensive" service industries that are vulnerable to AI automation.

Investment Themes & Risks

The "Situational Awareness" Thesis

• AGI by 2027: The core of these investments is based on the prediction that Artificial General Intelligence will be reached by 2027, requiring a massive, immediate build-out of physical infrastructure.
• From "Bits" to "Atoms": The next phase of AI wealth will be created in the physical world—factories, power plants, and real estate—rather than just software.

Risk Factors

• Single Thesis Risk: The entire portfolio is a "single bet" on AI. If AI spending plateaus or a macro recession hits, all these positions (Energy, Real Estate, Infrastructure) are likely to crash together.
• Execution Risk: Companies like Bloom Energy face massive manufacturing challenges to meet their $20B backlog.
• Timeline Risk: If AGI does not arrive by 2027, the massive capital expenditure (CapEx) currently flowing into these companies may dry up.