Chamath Palihapitiya

Emil Michael confirms that Google (Gemini) and OpenAI (Microsoft) have agreed to "all lawful use" terms, unlike Anthropic. Chamath explicitly states, "Google is the bet... the market value creator." Anthropic's "moral" refusal to support kinetic military operations creates a massive vacuum in the defense sector. Google and OpenAI are explicitly named as the compliant alternatives that will absorb this government revenue. Furthermore, Chamath notes Google will likely launch a "virtual coworker" feature within 90 days to crush Anthropic's enterprise moat. Long GOOGL and MSFT as they capture the defense market share abandoned by Anthropic and integrate agentic AI into enterprise suites. Regulatory backlash from progressive constituents regarding military contracts.

Emil Michael confirms that Google (Gemini) and OpenAI (Microsoft) have agreed to "all lawful use" terms, unlike Anthropic. Chamath explicitly states, "Google is the bet... the market value creator." Anthropic's "moral" refusal to support kinetic military operations creates a massive vacuum in the defense sector. Google and OpenAI are explicitly named as the compliant alternatives that will absorb this government revenue. Furthermore, Chamath notes Google will likely launch a "virtual coworker" feature within 90 days to crush Anthropic's enterprise moat. Long GOOGL and MSFT as they capture the defense market share abandoned by Anthropic and integrate agentic AI into enterprise suites. Regulatory backlash from progressive constituents regarding military contracts.

Chamath states, "Yesterday, Elon shut down the Model S and Model X production lines in Fremont... to make Optimus." This confirms Tesla is transitioning from a pure EV car company to a robotics/labor substitution company. If Optimus scales ("put two in your garage"), the TAM expands from transportation to the entire labor market. LONG. The market likely still prices TSLA as an auto manufacturer, not a labor supplier. Execution risk on Optimus; loss of revenue from high-margin S/X models in the interim.

SpaceX is a one-of-a-kind company that recently had its highest price-to-sales ratio, and now it will grow into its valuation and continue to expand its valuation over time, making it a compelling long-term holding.

The AI hardware, memory, and chip complex faces enormous supply constraints and scarcity, which will drive continued strong performance for at least another couple of years.

Power constraints are the key bottleneck in AI, limiting compute for model companies like OpenAI and Anthropic. Hyperscalers (Oracle, Amazon, Meta, Microsoft, Google) have excess capacity and will benefit from the power crunch, gaining leverage over AI model companies.

Power constraints are the key bottleneck in AI, limiting compute for model companies like OpenAI and Anthropic. Hyperscalers (Oracle, Amazon, Meta, Microsoft, Google) have excess capacity and will benefit from the power crunch, gaining leverage over AI model companies.

"Is on-prem the new cloud?... Once you use these tools, it is very difficult for a company to be able to control how their data is used... The only solution is to have the pendulum swing all the way back and have private provisioned networks." Enterprises will panic about leaking IP to public models (like OpenAI/Anthropic). This forces a massive capex cycle into local hardware and private clouds. Companies that sell the hardware for local inference (Apple Mac Studios were explicitly mentioned as the current solution) and the chips to run private clusters (Nvidia, AMD) will capture this spend. Long hardware providers enabling "Sovereign AI" and local inference. Public cloud providers (AWS/Azure) solve the privacy layer faster than hardware can be deployed on-prem.

"Is on-prem the new cloud?... Once you use these tools, it is very difficult for a company to be able to control how their data is used... The only solution is to have the pendulum swing all the way back and have private provisioned networks." Enterprises will panic about leaking IP to public models (like OpenAI/Anthropic). This forces a massive capex cycle into local hardware and private clouds. Companies that sell the hardware for local inference (Apple Mac Studios were explicitly mentioned as the current solution) and the chips to run private clusters (Nvidia, AMD) will capture this spend. Long hardware providers enabling "Sovereign AI" and local inference. Public cloud providers (AWS/Azure) solve the privacy layer faster than hardware can be deployed on-prem.

1. THE FACT: Chamath reiterates a prediction that "copper is the only game in town" for AI demand, unless superconductivity or carbon nanotubes emerge. He notes AI is a huge demand driver for a "very under-resourced material." 2. THE BRIDGE: The massive demand from the AI industry, coupled with copper being an "under-resourced material," points to a supply-demand imbalance that will drive copper prices higher. 3. THE VERDICT: AI's significant demand for copper, an under-resourced material, creates a strong long opportunity for copper.

PayPal is being targeted for acquisition at a low market cap; the current offer price is unlikely to be final and the clearing price will be 10-15% higher, possibly attracting a competitive bid from a well-capitalized buyer like Elon Musk.

Palantir is a necessary middleware layer that protects enterprise data and usage when interacting with LLMs, filling a gap that the LLMs themselves do not address.

Power constraints are the key bottleneck in AI, limiting compute for model companies like OpenAI and Anthropic. Hyperscalers (Oracle, Amazon, Meta, Microsoft, Google) have excess capacity and will benefit from the power crunch, gaining leverage over AI model companies.

Data centers are massively compute constrained due to power and land approval issues. Bloom Energy provides an on-site natural gas power solution that allows data centers to get clean air permits quickly, avoiding grid dependency and enabling construction, leading to strong stock performance.

"Is on-prem the new cloud?... Once you use these tools, it is very difficult for a company to be able to control how their data is used... The only solution is to have the pendulum swing all the way back and have private provisioned networks." Enterprises will panic about leaking IP to public models (like OpenAI/Anthropic). This forces a massive capex cycle into local hardware and private clouds. Companies that sell the hardware for local inference (Apple Mac Studios were explicitly mentioned as the current solution) and the chips to run private clusters (Nvidia, AMD) will capture this spend. Long hardware providers enabling "Sovereign AI" and local inference. Public cloud providers (AWS/Azure) solve the privacy layer faster than hardware can be deployed on-prem.