Chamath Palihapitiya
· Chamath Palihapitiya
· June 14, 2026 at 16:00
· ⏱ 6 min read
| Read on Substack ↗
Summary
The article reports on major AI and space developments: OpenAI's pivot to a distribution company aiming for personal AGI, Bezos' new AI engineering venture, SpaceX's orbital data center and record IPO, and Ramp's focus on metered AI token spending. It signals accelerating AI infrastructure investment and inference demand, validating both public and private company narratives.
•OpenAI filed a confidential S-1 after Anthropic disclosed IPO preparations; OpenAI valued at $852B, Anthropic at $965B.
•OpenAI's 'third phase' goals include an automated AI researcher by March 2028, faster economic growth, and a personal AGI for everyone.
•A $200/month ChatGPT Pro subscription can consume up to $14,000 in tokens at API prices, costing OpenAI ~$3,500 in compute—implying ~75% inference margins.
•SpaceX revealed AI1, an orbital data center satellite (120kW compute, 70m wingspan), and went public as SPCX at a $1.77T valuation, ending its first day at ~$2T.
•Project Prometheus (Bezos co-CEO) raised $12B at $41B valuation to build an 'artificial general engineer' compressing design-to-manufacturing cycles.
•Ramp raised $750M at $44B valuation, betting on metered AI token spending; GPT-4-level token costs fell from $60 to $0.40 per million tokens since 2023.
The article details SpaceX's AI orbital data center prototype (120kW compute, interchangeable payload) and its record IPO, validating the orbital compute thesis and strong market demand for space-base
The article details SpaceX's AI orbital data center prototype (120kW compute, interchangeable payload) and its record IPO, validating the orbital compute thesis and strong market demand for space-based AI infrastructure.
Risk: High valuation multiples and execution risk on satellite constellation buildout; regulatory hurdles for orbital data centers.
The article reports ~75% inference margins at OpenAI/Anthropic and token cost structures that imply substantial GPU demand. As the dominant AI chip supplier, Nvidia directly benefits from continued sc
The article reports ~75% inference margins at OpenAI/Anthropic and token cost structures that imply substantial GPU demand. As the dominant AI chip supplier, Nvidia directly benefits from continued scaling of frontier AI models and inference workloads.
Risk: Competition from custom AI chips (e.g., from OpenAI, Anthropic, or Project Prometheus) could erode Nvidia's market share over time.
This newsletter, published June 14, 2026,
features Chamath Palihapitiya
discussing SPCX, NVDA.
2 trade ideas extracted by AI with direction and confidence scoring.