Using GPUs as leverage, MSFT beats the case, FTC fails under Khan | E1775

Episode Summary

Executive Summary: The episode centers on two big themes: how AI-era scarcity of GPUs is creating new leverage and conflicts in venture capital, and how Microsoft is winning major antitrust battles by framing product choices as consumer-friendly bundling rather than harmful monopolization. The conversation also critiques Lina Khan’s FTC strategy as overly broad and better suited to rulemaking than case-by-case enforcement.

Main Topics: GPU scarcity as venture leverage (Priority: 5/5): The hosts discuss how access to scarce NVIDIA GPUs is becoming a strategic asset that investors and infrastructure providers can use to attract startups, create preferred deal flow, and shape the AI ecosystem. Inflection AI and conflicted investment dynamics (Priority: 5/5): Inflection’s massive round and planned GPU cluster are used as the central example of how Microsoft, NVIDIA, and other investors may be using capital, hardware access, and partnerships in ways that create overlapping incentives. CoreWeave’s pivot from crypto to AI infrastructure (Priority: 4/5): CoreWeave is highlighted as a successful pivot from Ethereum mining to GPU cloud services, showing how crypto-built infrastructure can be repurposed for AI demand and become a strategic supplier in the market. Microsoft’s FTC and EU antitrust battles (Priority: 5/5): The transcript covers Microsoft’s legal victory over the FTC regarding Activision Blizzard and the separate EU probe into Teams bundling, contrasting a losing anti-big-tech posture with more traditional antitrust theories like bundling and exclusivity. Critique of Lina Khan’s antitrust approach (Priority: 5/5): Jason argues that Khan’s FTC is too focused on future hypothetical harm and politically motivated big-tech skepticism, rather than clear consumer harm or enforceable market conduct violations. Broader reflections on work, productivity, and remote culture (Priority: 3/5): A long tangent argues that AI is improving productivity by removing repetitive work, while remote work has mixed effects on collaboration, onboarding, and mental health, especially for younger workers.

Key Arguments: GPU access is becoming a form of leverage in startup investing, similar to how distribution or platform access has historically created power. In AI, conflicts of interest are common and often tolerated if they create business advantage; the Valley often operates on 'no conflict, no interest.' CoreWeave’s move from crypto mining to GPU cloud shows how infrastructure built for one wave can be retooled for the next. Microsoft’s Teams bundling is a cleaner antitrust target than speculative future-harm cases because consumers can point to the product being free and pre-installed. The FTC should focus on clear violations like dark patterns, bundling, and deceptive consumer practices rather than trying to preemptively stop big firms from growing. Satya Nadella’s commitment to bring Call of Duty to Nintendo weakens the FTC’s argument that Microsoft would restrict access after acquiring Activision. AI tools are already delivering measurable productivity gains by automating repetitive knowledge work and research. Remote work improves some people’s productivity but hurts onboarding, culture, and the ability to disengage from work for many others.

Data Points: Inflection AI valuation: $4 billion - Referenced in connection with its $1.3 billion funding round. Inflection AI funding round: $1.3 billion - Used as the anchor for discussing GPU access and investor leverage. Inflection GPU cluster size: 22,000 GPUs - Described as the world’s largest AI-specific GPU cluster. NVIDIA H100 GPUs to be installed: More than 18,000 - Part of Inflection’s planned expansion. Inflection GPUs already in use: 3,500 - Existing installed base before the new deployment. CoreWeave Series B valuation: $2.1 billion - NVIDIA participated in the round alongside Daniel Gross and Nat Friedman. CoreWeave Series B size: $220 million - Discussed as part of CoreWeave’s pivot to AI infrastructure. Microsoft investment in OpenAI: $10 billion - Cited to illustrate conflicts with its investment in Inflection. Slack user growth impact: Approximately 20x - Used rhetorically to show how Microsoft Teams bundling overwhelmed Slack growth. Teams bundled into Office 365: 2017 - Year Microsoft initially added Teams for free to Office 365. Slack complaint to EU: 2020 - The initial complaint about Microsoft bundling Teams. Activision acquisition value: $70 billion - Announced as Microsoft’s largest acquisition and the largest gaming acquisition ever. Warzone monthly active players: 52.5 million - Used to show the scale of Activision’s gaming franchise. Warzone daily active players: 1 to 3 million - Illustrates the ongoing engagement with the game. OpenPhone price: $13 per user per month - Ad read promoting business phone functionality. Lemon.io discount: 15% off the first four weeks - Podcast sponsorship offer. OpenPhone discount: 20% off the first six months - Podcast sponsorship offer. Veed avatar languages: 75+ languages - Promoted as part of AI avatar video tools. Veed avatars: 50+ avatars - Shown as part of the AI video offering.

Pivotal Quotes: "No conflict, no interest." — Jason Calacanis: Used repeatedly to describe how tech investors and companies operate despite overlapping incentives. "I love innovation and venture. I love when people innovate on the five partners with the $300, $400 million fund, making two investments a year, sitting on 12 boards." — Jason Calacanis: A critique of traditional venture capital, contrasted with faster, more leverage-based models. "The FTC executes laws, they don't create them. They're the umpire, not the league." — Jason Calacanis: Part of the argument that Lina Khan is overstepping her role by trying to shape markets instead of enforcing rules.

Implications: AI infrastructure, especially GPU access, may reshape startup funding and competition. Meanwhile, antitrust may increasingly hinge on concrete conduct like bundling and dark patterns, not speculative future harm.