Here's a summary of the podcast discussion, including every news item mentioned:
The podcast chapter discusses the recent earnings reports of Google and Tesla, focusing heavily on their capital expenditures (CapEx) and the broader implications for the AI industry and other tech giants.
**Key News Items & Discussion Points:**
**Google:**
* **Q-Results & CapEx:** Reported Q-results on the same day as Tesla. Google's CapEx was described as having "blown the doors off," "insane," and "outrageous."
* **Google Cloud Growth:** Google Cloud is growing 82% year-over-year and is now on a $100 billion revenue runway.
* **CapEx Forecast:** Google's CapEx is forecast to be $195 billion to $205 billion this year, with expectations for it to be even higher next year.
* **Stock Performance:** Google's stock was down 7-10% at the time of recording.
* **Negative Free Cash Flow (FCF):** Reported negative FCF for the first time *ever* since its IPO (approximately 22 years ago), attributed to aggressive investment in future opportunities, primarily AI.
* **AI Investment Scale:** This FCF dip signifies the "scale of the opportunity" Google sees in AI, indicating they're "spending it all on AI."
* **Return on Invested Capital (ROIC):** Google's 25-year average ROIC since going public has been 32%.
* **AI Strategy:** Google is seen as successfully navigating the AI transition with strong search, cloud, and silicon businesses. Fragmentation in AI models is viewed as beneficial for Google, allowing them to profit as a silicon and cloud provider.
* **CapEx Comparison:** CapEx spending is compared to "20% of this year's military budget."
* **Portfolio Companies:** Google owns 10% of SpaceX and a "good chunk" of Anthropic. Waymo is valued at $120 billion. A "hundred billion dollar mark to market" (implied gain) on Anthropic was noted in the quarter.
* **Google Cloud Platform (GCP):** Considered the best-suited enterprise layer for AI due to existing enterprise data (email, Drive), its model-agnostic nature, and allowing users to run any model or workflow.
* **Grok on GCP:** Grok (SpaceX AI) is available within Google Cloud equivalents like Sheets and Docs.
* **Worst-Case Scenario:** Even if Google's application layer and models don't succeed, they'll still have the "world's best infrastructure" that can "print cash."
* **SpaceX AI Payment:** Google is currently paying SpaceX AI $920 million per month (nearly $1 billion) to access a portion of SpaceX AI's AI compute.
**Tesla:**
* **Q-Results:** Reported Q-results on the same day as Google.
* **CapEx Surge:** Tesla's CapEx surged 142% year-over-year.
* **CapEx Expectation:** They expect $25 billion in CapEx.
* **Stock Performance:** Tesla's stock was down 14% at the time of recording.
* **Negative Free Cash Flow:** Also reported negative FCF.
**SpaceX / SpaceX AI:**
* **Stock Performance:** SpaceX stock is down 30% from its day-one closing price, now trading at $1.5 trillion (down from a $2 trillion IPO with an "Elon pop").
* **IPO Scale:** Described as an "unprecedentedly big" IPO.
* **Speaker's Position:** The speaker (Solving the Money Problem / SMR) is "still building a position" in SpaceX, finding lower prices beneficial for accumulation.
* **SpaceX AI Compute Advantage:** SpaceX AI is identified as being able to bring AI compute online "much faster and also more affordably" than companies like Google.
* **SpaceX AI Revenue:** SpaceX AI currently has an annualized revenue rate on its AI compute business of "not far off $30 billion a year."
* **StarMind:** Mentioned as "orbital AI compute" with the potential for "up to 1 million data centers in space."
* **Starship Flight 13:** An "incredibly successful test flight" occurred today, featuring a "soft splashdown" in the Indian Ocean. The heat shield was "pretty intact" despite a "significantly higher dynamic pressure" during ascent. Cameras onboard continued sending video. Starship is presented as the enabler for StarMind.
* **Valuation Thesis:** The speaker's 20-year SpaceX valuation model's biggest single profit center is **AI compute**, not Starlink, launch, or AI products.
* **Colossus:** SpaceX AI's "Colossus" AI compute infrastructure is noted for making "obscene" returns.
* **Grok Availability:** Grok is now available in the Microsoft Office suite as an embedded AI agent.
**Apple:**
* **Capital Allocation Criticism:** Apple was heavily criticized for its capital allocation strategy over the last decade.
* **Buybacks & Dividends:** Apple has bought back "half their stock," giving "hundreds of billions of dollars back in profits." Specifically, $900 billion in buybacks and $140 billion in dividends over the last decade (totaling $1.04 trillion).
* **"Skill Issue":** This strategy is deemed a "massive skill issue" and "extremely embarrassing" for a company of Apple's scale, arguing they failed to identify major future investment opportunities.
* **Apple Web Services:** The idea of "Apple Web Services" was discussed as a missed opportunity, acknowledging the difficulty of building a "five nines" reliable cloud service provider. Tim Cook is not seen as the right leader for such a venture.
**General Market & AI Observations:**
* **Market Reaction:** The market was "disappointed" or "retarded" in its reaction to the CapEx spending, leading to stock dips despite perceived strong future investments.
* **Demand for Intelligence:** A strong belief that "demand for intelligence will be near infinite for decades to come."
* **Democratization of AI:** Superintelligence is becoming democratized, with models simultaneously declining in cost and becoming more intelligent.
* **Demand for AI Compute:** Directly tied to demand for intelligence, "demand for AI compute will also be near infinite for decades to come." Increased efficiency in AI compute doesn't reduce infinite demand, it just means more "bang for buck."
* **Cloud Provider Reliability:** Achieving "five nines of reliability and uptime" for cloud services is extremely expensive, costing "hundreds of billions," and only a few "games in town" can manage it (implied Amazon, Microsoft, Google).
The podcast concludes with a promotion for the speaker's Patreon, where detailed 20-year valuation models for Tesla and SpaceX will soon be released, and a sponsor ad for AG1.