Interview, Fireside Chat, Podcast
E167: Google's Woke AI disaster, Nvidia smashes earnings (again), Groq's LPU breakthrough & more
- NVIDIA projects Q1 2024 revenue of approximately $24 billion, a 3x year-over-year increase, with a potential market capitalization reaching $10 trillion within two to three years.
- The infrastructure "build out" for AI is expected to continue for the next two to three years before reaching a terminal value, a one-time capital-intensive phase that may take a decade to fully mature and monetize.
- Historical patterns similar to the post-dot-com era are anticipated, where initial infrastructure over-investment will be followed by widespread application adoption across B2B and B2C sectors over the next decade.
- Terminal value is projected to shift from hardware enablers to the application layer, mirroring the internet transition where value moved from Cisco to companies like Netflix and Google.
- NVIDIA's market share is forecasted to drop to roughly 60% over five years as excess profits are competed away by new entrants, though analysts expect retention of a substantial portion of the total addressable market.
- The inference market is expected to become highly competitive, with new entrants like Grok potentially disrupting NVIDIA if they offer significantly faster and cheaper inference capabilities.
- Big Tech accounting incentives, specifically capitalizing infrastructure spend on balance sheets rather than expensing on the P&L, are driving near-term acceleration in AI spending, alongside a lack of M&A opportunities due to antitrust concerns.
- Google faces a risk of migrating users to open-source alternatives if it fails to provide unbiased and truthful answers, with projections suggesting a potential annual expenditure of $60 billion to $100 billion on training data licensing.
- Regulatory capture and potential ideological interpretations by federal agencies pose risks of a disinformation age if a single entity monopolizes training data and critical information interpretation.
- The "deep tech" business model is predicted to gain favor due to the extraordinary moats created by successfully navigating multiple technical hurdles, though these investments require high persistence and specific founder profiles over many years.
- Open-source models are expected to potentially dominate the market if commercial models continue to suffer from bias and lack of accuracy, serving as the primary source for unfiltered truth.
- Only the most accurate and unbiased AI models are expected to survive in the long term, with the market eventually forcing competitors to prioritize truth over ideology to retain users.