SpaceX’s $2T Case, Nvidia’s Shock Selloff, America Turns on AI, Trump Pulls AI Order, Bond Crisis?
Episode Details
The All-In Podcast covers several massive shifts in the technology and macroeconomic landscape. It begins by analyzing Andrej Karpathy leaving OpenAI and joining anthropic, led by Dario Amodei, to focus on recursive self-improvement and advancements in Claude. Karpathy, known for coining Vibe Coding, adds massive talent firepower. The hosts address the growing Anti-AI Sentiment and Job Displacement fears, critiquing tone-deaf messaging from CEOs like Matthew Prince of Cloudflare and Mark Zuckerberg of Meta. Instead, they agree with Shyam Sankar that the industry should focus on end-user benefits. A pulled Biden Executive Order on AI sparks a debate on AI Regulation and the ongoing US-China Rivalry between the United States and China. The conversation then shifts to the highly anticipated SpaceX IPO. Beyond Starlink and Starship's promise of Full Reusability (Rockets), the hosts, including Chamath Palihapitiya, highlight a massive revenue unlock via Colossus data centers, making Elon Musk a major player in the AI Infrastructure Buildout. This includes a compute deal with Anthropic under the umbrella of EWS (Elon Web Services), and enabling Cursor to train its models, heavily boosting xAI and their model Grok. They also speculate on Amazon entering the fold and the future reality of data centers in space. Financially, Nvidia and its CEO Jensen Huang continue to crush earnings, driven by demand for the H100 and innovations in Disaggregated inference, which heavily benefits neo-clouds like CoreWeave. Despite competition from Google's TPU and Broadcom, Nvidia maintains a stronghold. Finally, Gavin Baker of Atreides Management joins David Friedberg to discuss macroeconomics, noting that despite rising Inflation and Interest Rates, American Energy Dominance—partially driven by geopolitical tensions in the Strait of Hormuz involving Xi Jinping and Donald Trump—keeps the US economy resilient.
The episode explores the rapid evolution of AI infrastructure, the shift toward recursive self-improvement in LLMs, and the geopolitical implications of US-China relations on energy and technology. Key themes include the emergence of SpaceX as a critical AI compute player through 'Elon Web Services' and the resilience of the US economy driven by energy dominance despite inflationary pressures.
Generated with gemini-3.1-flash-lite on 7/19/2026, 5:40:08 AM. For research only. Not financial advice.SpaceX as AI Infrastructure Play
SpaceX is transitioning from a launch provider to a critical AI infrastructure layer via its massive compute clusters and Starlink connectivity.
The S1 filing reveals SpaceX is building data centers (Colossus) at record speeds with massive offtake agreements from AI labs like Anthropic, positioning it as a foundational player in the AI buildout.
- SpaceX is building data centers in 66 days, significantly faster than industry peers.
- Anthropic has a $45 billion, 3-year compute deal with SpaceX.
- SpaceX is investing 60% of its $20B annual capex into AI compute.
- Successful IPO (ticker: SPCX) expected in mid-June.
- Deployment of Starship for rapid reusability.
- Potential future merger with Tesla.
- Compute contracts have 90-day cancellation clauses.
- High capital intensity and execution risk in building orbital data centers.
- Valuation premium at $2T+ market cap.
- Analyze the S1 filing for specific revenue breakdown and contract duration.
- Monitor the progress of Starship's rapid reusability milestones.
- Track the competitive landscape for gigawatt-scale data center construction.
AI Semiconductor Ecosystem Efficiency
The AI hardware market is shifting toward domain-specific architectures and disaggregated inference, favoring companies that can provide integrated power, cooling, and compute solutions.
Nvidia's earnings and the success of neo-clouds like CoreWeave demonstrate that the AI compute stack is becoming more efficient, allowing for longer useful lives for hardware and better financing terms.
- Nvidia's CPU business is projected to reach $20B.
- Disaggregated inference allows mixing and matching accelerators, extending the lifespan of existing GPUs.
- Asset-backed financing for GPUs is becoming more available at lower rates (6-7%).
- Release of new benchmarks (MLPerf) for non-Nvidia accelerators.
- Advancements in DC-to-DC power architecture for data centers.
- Continued growth in hyperscaler and AI cloud capex.
- Potential for oversupply if AI demand growth slows.
- Geopolitical restrictions on chip exports to China.
- High valuation multiples for AI-adjacent infrastructure companies.
- Compare growth rates of AI semiconductor revenue across Nvidia, Broadcom, and other ASIC providers.
- Monitor the adoption of domain-specific architectures in enterprise vs. hyperscaler environments.
- Evaluate the impact of energy costs and power availability on data center expansion.
Watchlist
- SpaceX (SPCX)
- Nvidia (NVDA)
- Broadcom (AVGO)
- Anthropic
- xAI
- Cursor
- CoreWeave
- 10-year Treasury Yield
- Natural Gas (NG1)
Open Questions
- When will orbital data centers become a significant contributor to SpaceX's revenue?
- How will the US-China geopolitical tension impact the long-term supply chain for AI chips?
- Will the 'seasonal' nature of AI usage persist as agentic AI becomes more integrated into enterprise workflows?
- What is the true useful life of an H100 GPU in a disaggregated inference environment?
Key Topics & People
.jpg?width=300){kind=avatar}
.jpg?width=300){kind=avatar}
Host on the All-In Podcast who was absent during this episode.
Host on the All-In Podcast and entrepreneur.
The country facing economic shifts, political turmoil, and wealth disparities heavily discussed on the pod.
Former US President facing a fourth indictment.
Borrowing costs set by the Federal Reserve, currently pressuring real estate.
The massive capital expenditure cycle globally to build out hardware capacity and data centers for running AI models.
.jpg?width=300){kind=avatar}
CEO of Anthropic, criticized for allegedly seeking an FDA-like regulatory moat for AI.
The transition process where specific jobs are automated by AI, even if overall employment grows.
AI systems continuously learning and iterating to produce exponentially better results
.jpg?width=300){kind=avatar}
CEO of Nvidia, known for guiding the company's AI hardware dominance
_(cropped).jpg?width=300){kind=avatar}
CEO of Meta, who is initiating a price war by releasing strong agentic coding models at aggressively low prices.
A modern software development approach where non-developers generate production code using AI.
The emerging political and industry frameworks intended to control AI development and deployment.
The concept of hosting AI data centers in orbit to circumvent terrestrial energy and zoning constraints.
Investor and podcast guest discussing space, IPOs, and technology markets.
Critical maritime chokepoint designated to be reopened as part of the peace agreement.
An investment firm founded by Gavin Baker that is actively investing in secondary markets and late-stage privates.
.jpg?width=300){kind=avatar}
.jpg?width=300){kind=avatar}
CEO of Cloudflare who discussed cutting jobs to replace 'measurers' with AI.
Tech company led by Matthew Prince that also cited AI efficiencies while making layoffs.
The leader of China, recently engaged in high-level geopolitical meetings with US officials.
The strategic advantage the US holds as the world's leading producer and exporter of oil and natural gas.
The architectural shift in AI where specific processing tasks are separated and routed to optimized hardware, extending the lifecycle of existing GPUs.
The engineering capability to land, lightly refurbish, and relaunch a rocket on the same day, essential for economical space exploration.
A nickname for the emerging business of renting out compute power from Elon Musk's massive data centers.
The ongoing geopolitical, technological, and economic competition between the United States and China, prominently featuring an AI arms race.
A planned federal mandate to regulate frontier AI models that was unexpectedly pulled.
CTO of Palantir who argues the industry should listen to end-users of AI rather than just model makers.
The growing negative public perception and fear of artificial intelligence, driven by job loss concerns and dystopian corporate messaging.
A highly respected AI researcher and founding member of OpenAI, known for leading Tesla's self-driving team and recently joining Anthropic.