CoreWeave Crash, AI Trade Cracks, Anthropic IPO Red Flags, Bill Gates Doomer
Episode Details
In this episode of the All-In Podcast, hosts Jason Calacanis, Chamath Palihapitiya, and David Sacks are joined by guest Ben Horowitz from Andreessen Horowitz to discuss the state of enterprise AI, infrastructure investments, and whether the AI bubble is bursting. The episode opens with a deep dive into CoreWeave, a dominant Neocloud company founded by Michael Intrator. To fund massive acquisitions of GPUs—specifically the highly coveted H100 units produced by Nvidia (led by Jensen Huang)—CoreWeave utilizes a specialized debt-financing structure known as The Box. This isolates revenue streams from investment-grade clients like Microsoft. However, recent spikes in Credit Default Swaps (CDS) yields and a rising macroeconomic Risk-free rate signal potential financial strain, despite strong compute demand validated by Andy Jassy regarding Amazon and its AWS division. The panel contrasts CoreWeave’s rigid contracts with the flexible Spot pricing utilized by Elon Musk at SpaceX, noting that a rapid Payback period is essential for survival in capital-intensive markets. The conversation then shifts to the profound Deflationary effect of AI on the software industry. As Open Source AI and autonomous AI agents drive down coding costs, they democratize the creation of Bespoke software. While generic SaaS is threatened, specialized vertical integration startups like Harvey and feature-rich platforms like 11 Labs maintain strong moats via proprietary data. Next, the hosts evaluate the highly anticipated IPO of Anthropic, an AI lab that forms a foundation model Duopoly alongside OpenAI (steered by Sam Altman), amidst emerging competition in Frontier models from tech giants like Meta and Google. David Sacks severely criticizes Anthropic for its attempts at Regulatory Capture, pointing to its aggressive lobbying against Distillation (AI) and its recent fear-mongering campaign alleging that highly efficient SLMs (Small Language Models) could go rogue and infect vulnerable IoT Devices. Countering this narrative, Brian Armstrong successfully deployed an open-weight small model at Coinbase, demonstrating immense cost-efficiency without catastrophic risk. Sacks also notes that Anthropic's flagship model, Claude, exhibits concerning signs of moral misalignment engineered into its constitution—an epistemic flaw previously highlighted by Mustafa Suleyman. Finally, the hosts dissect the recent surge in AI Doomerism heavily promoted by Bill Gates. They draw parallels between his alarmist AI narratives and his historical involvement in shaping public policy during COVID-19, including his financial support for controversial Gain-Of-Function Research. The hosts suggest his current focus is an attempt to regain centralized global influence, much like how his previous philanthropic efforts served to repair a public image fractured by intense Antitrust scrutiny.
The episode discusses the potential financial strain on CoreWeave due to rising interest rates impacting its debt financing model, contrasting it with the deflationary impact of AI on software and the competitive landscape for AI models. It also touches on the motivations behind AI doomerism narratives.
Portfolio lens: This set of ideas represents a basket focused on the infrastructure and foundational models of the AI revolution, examining both the capital intensity risks and the disruptive potential of AI-driven software commoditization.
Generated with gemini-2.5-flash-lite on 10/18/2026, 3:06:27 AM. For research only. Not financial advice.CoreWeave's Debt Financing Vulnerability
CoreWeave's reliance on debt financing, particularly its 'box' structure, makes it vulnerable to rising risk-free rates, potentially impacting its ability to service its debt and operate profitably.
The discussion highlights that CoreWeave needs to raise significant additional debt to finance its contracted GPU purchases. As the risk-free rate increases, CoreWeave's cost of borrowing rises, eroding margins and pushing it closer to a break-even point, as evidenced by the spike in its Credit Default Swaps (CDS) yields.
- CoreWeave's debt yields have hit 15%, indicating bond traders' concerns about its ability to generate cash to support its debt.
- The risk-free rate is rising due to central bankers and bond vigilantes.
- CoreWeave has about $35 billion of debt and needs to raise another $35 billion.
- The average interest rate CoreWeave pays is about 8.5%, but some debt has spiked to 15%.
- The panel estimates CoreWeave is plus or minus $5 billion away from its break-even point.
- 75% of CoreWeave's revenue comes from three companies, creating customer concentration risk.
- Further increases in the risk-free rate.
- Failure to secure new debt financing at manageable rates.
- Downgrades from credit rating agencies.
- Significant revenue shortfalls from its top three customers.
- Continued high demand for GPUs and compute power may sustain CoreWeave's revenue.
- The 'box' financing structure isolates revenue streams, potentially mitigating contagion risk between different financing structures.
- CoreWeave's ability to secure asset-backed financing for its data centers and GPUs.
- The lifespan of GPUs may extend beyond initial estimates, improving asset value.
- Analyze CoreWeave's latest debt filings and maturity schedules.
- Track CDS spreads and bond yields for CoreWeave and similar 'Neocloud' companies.
- Monitor customer concentration and revenue trends for CoreWeave's top clients.
- Assess the current and projected risk-free rate environment.
Anthropic's IPO Valuation and Competitive Pressures
Anthropic's ambitious IPO valuation may be challenged by increasing competition from open-source models and the inherent deflationary pressures of AI on software, despite its strong revenue growth.
While Anthropic shows impressive revenue growth and is free cash flow positive, its potential $2 trillion valuation is questioned given the rapid commoditization of AI capabilities through open-source alternatives and the broader trend of AI driving down software costs. Concerns also exist about the company's lobbying efforts against open-source models.
- Anthropic reported Q3 free revenue of $16.8 billion, with net revenue retention at 900%.
- Adjusted gross margin is 85% and adjusted operating margin is 11%.
- ARR estimates range from $70 billion to over $100 billion.
- Anthropic is targeting an unprecedented valuation of $2 trillion.
- Open-source AI and AI agents are driving down coding costs and democratizing bespoke software creation.
- There's a trend of companies using open-weight small models (e.g., Coinbase's fraud detection) which are faster and cheaper than frontier models.
- Anthropic has been lobbying against 'distillation' and promoting fear-mongering campaigns about small models.
- The discussion highlights the potential for AI to create 'bespoke software' for specific use cases, potentially disrupting existing SaaS models.
- Successful execution of a large IPO at a high valuation.
- Continued strong revenue growth and profitability.
- Development of unique competitive moats beyond model performance (e.g., proprietary data, specialized enterprise solutions).
- Increased adoption of Anthropic's models by enterprise clients.
- The rapid advancement and adoption of open-source AI models that offer comparable or superior capabilities at lower costs.
- The deflationary impact of AI on software pricing, potentially limiting future revenue growth and margins.
- Increased competition from other frontier model developers (OpenAI, Google, Meta).
- Regulatory scrutiny or backlash against Anthropic's lobbying efforts.
- The potential for Anthropic's own models to exhibit 'moral misalignment' or unintended behaviors, as suggested by research on their constitution.
- The need for significant capital expenditure to maintain compute advantage, potentially requiring further down rounds if market conditions are unfavorable.
- Analyze Anthropic's S-1 filing for detailed financial metrics and risk disclosures.
- Track the performance and adoption rates of open-source AI models.
- Monitor pricing trends for AI compute and API access.
- Assess the competitive positioning of Anthropic's Claude models against OpenAI's GPT and other leading models.
- Evaluate the long-term impact of AI-driven software commoditization on the SaaS market.
Watchlist
- CoreWeave's CDS spreads
- Interest rates on high-yield corporate debt
- Anthropic's IPO pricing and post-IPO trading performance
- Adoption rates of open-source AI models
- Compute pricing trends for H100 GPUs
- Regulatory developments concerning AI models and open-source software
Open Questions
- What is the true cost of capital for CoreWeave if it needs to raise an additional $35 billion in debt?
- How will the increasing efficiency and decreasing cost of smaller AI models impact the demand for frontier models like Anthropic's Claude?
- What is the long-term sustainability of high valuations for AI companies given the rapid pace of innovation and potential for commoditization?
- Can Anthropic effectively defend its market position against open-source alternatives without resorting to regulatory capture tactics?
- What are the real-world implications of Anthropic's 'AI worm' research, and how does it compare to existing IoT security vulnerabilities?
- What is the actual risk of 'moral misalignment' in Anthropic's models, and how does it compare to other leading AI systems?
Key Topics & People
.jpg?width=300){kind=avatar}
CEO of Nvidia, known for his pragmatic approach to computer science and opposition to AI fear-mongering.
.jpg?width=300){kind=avatar}
.jpg?width=300){kind=avatar}
The CEO of OpenAI, leading aggressive expansion in hardware and model capabilities.
The belief and associated fear-mongering that artificial intelligence will cause catastrophic harm to humanity.
.jpg?width=300){kind=avatar}
Co-founder of Microsoft and early idol of Martin Shkreli.
The baseline central bank interest rate that influences overall corporate borrowing costs.
Flexible, high-margin short-term pricing for compute power as opposed to long-term locked contracts.
Consumer hardware connected to the internet, falsely framed as a primary vector for rogue AI.
The crucial metric defining how quickly a company recovers massive capital infrastructure expenses.
Controversial lab practices cited to demonstrate the dangers of unchecked foundational research.
AI leader who correctly warned about the dangerous epistemic distortions inside AI constitutions.
CEO of Coinbase who publicly proved the enterprise viability of fine-tuned small AI models.
Cost-efficient, compact AI models deployed for specialized, secure enterprise tasks.
Using massive AI outputs to train smaller, more efficient competitor models.
The strategic effort by incumbents to shape government legislation to stifle new competitors.
The most advanced, capital-intensive AI models produced by top-tier laboratories.
Customized software applications generated easily by AI specifically for specialized enterprise tasks.
Publicly accessible AI models driving down commercial margins and democratizing development.
The structural reduction in software development costs due to artificial intelligence tools.
.jpg?width=300){kind=avatar}
CEO of Amazon, pointing out a severe, multi-year supply shortage of data centers.
Insurance derivatives used by investors to hedge against corporate bond defaults.
Founder and CEO of CoreWeave who designed their unique financial framework.
Top-tier venture capital firm heavily invested in technology and AI.
_(cropped).jpg?width=300){kind=avatar}
Prominent venture capitalist providing insights on tech infrastructure and market trends.
Podcast host critical of regulatory manipulation and enterprise SaaS economics.
.jpg?width=300){kind=avatar}
Investor and podcaster analyzing complex debt structures and macroeconomic risk.
.jpg?width=300){kind=avatar}
Podcast host directing the dialogue on technology market shifts and venture capital.