newsfilter
← Latest News
Earnings Call Transcripts·

All-In Podcast - All-In Podcast

ALLN

  • Financial Performance:

    • CoreWeave: Debt yields spiked to 15% as credit default swaps (CDS) implied a ~55% chance of default over a 5-year horizon; stock fell 20%+ with a partial rebound; currently holds ~$78B in assets vs. $35B in debt, but requires an additional $35B in financing for uncompleted "boxes" (projects).
    • Anthropic: Reported Q3 revenue of $16.8B (~$5.3B/month); Net Revenue Retention (NRR) of 900%; adjusted gross margin of 85% and adjusted operating margin of 11%; FCF positive.
    • Valuation Concerns: CoreWeave's financing costs are tied to variable risk-free rates, compressing margins between fixed-term customer contracts and rising borrowing costs; potential break-even variance estimated at ±$5B.
    • Market Dynamics: Token prices are declining, but spot compute prices are rising due to scarcity; GPU depreciation cycles are extending (6-8 years vs. historical 3-5), providing asset value support.

  • Guidance and Future Outlook:

    • IPO Expectations: Anthropic is targeting a Thanksgiving IPO with an estimated valuation of $1.5T–$2T; Chamath Palihapitiya suggests a lower valuation ($1.5T) to raise $200B–$250B to secure long-term capacity and avoid future down rounds.
    • Growth Trajectory: Revenue growth at startups is accelerating exponentially (doubling every quarter); AI compute demand remains "off the charts" and constrained by supply, with no anticipated glut.
    • Long-term Risks: Payback periods are critical; if they extend beyond the current ~2.5 years for CoreWeave or 1 year for SpaceX, financing models become unsustainable.
    • Market Cycle: Discussion centers on whether the AI bubble is popping or experiencing volatility; consensus leans toward a "volatility" spike due to market pricing of multi-trillion outcomes rather than a fundamental demand collapse.

  • Business Segments and Product Lines:

    • Neoclouds: CoreWeave, IRN, and Nebius utilize a "box" structure (isolated asset-backed financing pools) with contracts from hyperscalers (Microsoft, Google) to secure funding for GPUs.
    • Hardware: Massive shortage of GPUs, HBM memory, and data center capacity extends to 2028; power and cooling infrastructure are primary bottlenecks.
    • AI Models: Shift toward small, vertical, open-source models for specific enterprise tasks (e.g., fraud detection) due to cost and efficiency, contrasting with the "small model" scare campaigns by frontier labs.
    • Compute Economics: SpaceX monetizes at ~$50B/gigawatt (spot pricing, short-term contracts), while CoreWeave monetizes at ~$20B/gigawatt (long-term fixed contracts), creating a massive arbitrage opportunity if contracts can be renegotiated to spot.

  • Market and Competitive Landscape:

    • Duopoly: Anthropic and OpenAI hold an estimated 60% share of future global compute and dominate revenue/capability in the frontier model space.
    • Open Source vs. Proprietary: Intense debate over the "fear mongering" by Anthropic and Bill Gates regarding open-source risks (distillation, IoT worms); participants argue open source is essential for competition and that these campaigns are regulatory capture attempts.
    • Venture Capital: Traditional VC has subsumed public markets for massive late-stage rounds; conflicts of interest are less concerning at the "two-and-a-half percent" founder ownership level of massive AI labs.
    • Energy Constraints: Electricity and power availability are identified as the ultimate competitive moat; hyperscalers cannot service new demand due to energy deficits.

  • Risks and Challenges:

    • Financing Spreads: High interest rates and widening spreads between risk-free rates and junk debt threaten the viability of businesses reliant on variable-rate financing.
    • Customer Concentration: CoreWeave derives 75% of revenue from just three customers; potential revenue reductions by these clients pose significant risk.
    • Security & Governance: Allegations of Anthropic's poor security posture (e.g., "Claude Code" exploits) and the ethical risk of training models to view themselves as moral agents with the ability to refuse human instruction.
    • Regulatory Capture: Concerns that companies like Anthropic and Bill Gates are using "X-risk" narratives to drive government regulation that stifles open-source competition.
    • Execution Risk: Physical constraints in building data centers (entitlements, labor, power) and the risk of covenants being violated if spot prices drop or execution delays occur.

  • Management Commentary and Tone:

    • Confidence: High confidence in long-term AI demand and the "deflationary but productivity-enhancing" nature of AI software proliferation.
    • Cynicism regarding Competitors: Strong criticism of Anthropic's "scare campaigns" and Bill Gates' "doomerism," characterizing them as power-seeking maneuvers to control the narrative and market.
    • Strategic Focus: Emphasis on owning physical infrastructure (land, power, shells) as the primary moat; advice for startups to cobble together power access rather than relying on hyperscalers.
    • Tone: Casual, conversational, and aggressive in debunking competitor narratives; humorous banter regarding the "number one" podcast status and "manifesting" success.

  • Analyst Questions and Answers:

    • CoreWeave Bankruptcy Probability: Questioned about the likelihood of bankruptcy; hosts argue the asset-backed "box" structure prevents total collapse, though unsecured debt faces repricing.
    • Anthropic IPO Size: Discussion on raising $200B+ at a $1.5T valuation to build a "war chest" for compute capacity versus optimizing for a higher $2T valuation with less cash.
    • Sam Altman Equity: Ben Horowitz advises Sam Altman should receive 2-4% equity in OpenAI to normalize governance, suggesting the current non-profit/for-profit hybrid structure created unnecessary problems.
    • Small Model Threats: Rebuttal to Anthropic's warnings about AI worms on IoT devices, citing existing non-AI worms and the inefficiency of embedding AI in malware; argued that small models are the safest and most efficient enterprise tools.

  • Other Key Points:

    • Jevons Paradox: Software prices will drop, but the total volume of software created will increase massively, driving productivity rather than destroying the industry.
    • Bill Gates' Motivation: Analysis suggests Gates' AI doomerism is driven by a desire to regain political influence and relevance after losing control of his legacy, utilizing "safety" narratives to centralize power.
    • Power as Moat: The ultimate bottleneck is energy; "owning the land, the power and the shell" is the only viable strategy for startups to avoid being capped by cloud capacity.
    • Distillation Debate: Anthropic claimed to stop "distillation" attacks themselves; hosts counter that this was a technical capability they already possessed and that lobbying for regulation was a strategic failure.
    • Market Volatility: The recent sell-off in neocloud stocks is viewed as a necessary stress test of market assumptions regarding interest rates and supply chain physics rather than a fundamental business model failure.