# Legendary Investor SHORTS the AI Bubble (Nov 25th is the day) Source: https://www.youtube.com/watch?v=qrGpdYtwuo0 Recap page: https://rapidrecap.app/video/qrGpdYtwuo0 Generated: 2025-11-13T03:02:39.398+00:00 --- ## Quick Overview The video discusses current AI developments, including the rapid construction of gigawatt-scale data centers by hyperscalers, OpenAI's legal battle against The New York Times over user data privacy, the introduction of GPT-5.1 by Sam Altman, and a study showing generative AI boosts online retail productivity by $5 per consumer annually, all while contrasting these advancements with Michael Burry's public short position on AI infrastructure. **Key Points:** - Epoch AI data suggests hyperscalers plan to build gigawatt-scale data centers rapidly, with some coming online as early as 2026, marking one of the fastest infrastructure build-outs in history. - OpenAI publicly refutes The New York Times' demand to hand over 20 million private user conversations, stating it disregards privacy protections and they are fighting the baseless lawsuit. - Sam Altman announced GPT-5.1, noting improvements in instruction following, adaptive thinking, intelligence, and style. - A study on Generative AI in online retail showed a 16.3% sales lift and an implied annual incremental value of approximately $5 per consumer. - Investor Michael Burry is publicly shorting the AI bubble, claiming hyperscalers are artificially inflating earnings by extending the useful life of AI compute equipment depreciation. - The video also highlights the launch of a new mobile app called 'Vibe Code' which uses AI to generate images and audio based on user input. - The video concludes by emphasizing the importance of building an AI resilience ecosystem and achieving public oversight commensurate with AI capabilities. ![Screenshot at 00:00: A Gantt chart from Epoch AI illustrating the aggressive timeline for building gigawatt-scale data centers by major AI players like Anthropic, xAI, Microsoft, Meta, and Amazon, projecting some to be operational by 2026.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-00-00.png) **Context:** The video presents a rapid-fire overview of several major, concurrent events in the AI landscape around November 2025, covering massive infrastructure investment, legal disputes regarding user data privacy (specifically OpenAI vs. The New York Times), new model releases (GPT-5.1), and empirical evidence of AI's productivity benefits in e-commerce, all framed against the backdrop of skepticism from figures like Michael Burry regarding the sustainability of current AI spending. ## Detailed Analysis The video starts by showcasing Epoch AI's projection that several gigawatt-scale data centers, including projects by Anthropic, xAI, Microsoft, Meta, and Amazon, are expected to become operational between 2025 and 2027, illustrating the massive capital expenditure (capex) race in AI infrastructure. Concurrently, the video highlights Michael Burry's public short position on AI infrastructure, asserting that hyperscalers are committing accounting fraud by artificially extending the useful life of expensive Nvidia chips, which he estimates will understate depreciation by $176 billion between 2026 and 2028. In response to a major legal challenge, OpenAI posted a statement "Fighting the New York Times' invasion of user privacy," demanding the NYT cease its request to turn over 20 million private user conversations, which OpenAI argues is an overreach that risks user privacy without resolving the core lawsuit. Shifting focus to product development, the video notes Sam Altman announced the release of GPT-5.1, which he views as a nice upgrade, particularly praising improvements in instruction following and adaptive thinking. Furthermore, the presenter demonstrates a new AI-powered mobile app called 'Vibe Code' that allows users to generate images and track moods via text prompts. The video then pivots to positive productivity evidence, referencing a study showing generative AI boosts online retail productivity, resulting in a 16.3% sales lift and an added value of about $5 per consumer annually, with the biggest win being a pre-sale chatbot that answered product questions in many languages. Finally, the speaker discusses OpenAI's recommendations for achieving a positive AI future, emphasizing the need for public oversight, accountability, and building an AI resilience ecosystem analogous to cybersecurity infrastructure development. ### AI Infrastructure Race - Epoch AI chart details gigawatt-scale data center buildout timelines for Anthropic, xAI, Microsoft, Meta, and Amazon, with first operational dates projected between 2025 and 2027 - Some projects aim for completion in just 1-2 years from construction start. ### Michael Burry's AI Short Thesis - Cassandra Unchained highlights Burry's claim that hyperscalers are committing fraud by extending compute equipment useful life (e.g., Google extending to 6 years by 2025) to artificially boost earnings - Burry estimates $176 billion in understated depreciation from 2026-2028. ### OpenAI Privacy Stance - OpenAI publicly fights The New York Times' demand to hand over 20 million private user conversations, calling the demand an overreach that risks privacy and shouldn't become collateral in a dispute. ### New AI Model Release - Sam Altman announced GPT-5.1 is out, highlighting improvements in instruction following, adaptive thinking, intelligence, and style. ### Generative AI Productivity Study - A study on online retail showed generative AI boosts productivity with a 16.3% sales lift and $5 per consumer annually, with pre-sale chatbots being the biggest win. ### AI Governance & Resilience - OpenAI recommends establishing shared safety standards among frontier labs, promoting public oversight, and building an AI resilience ecosystem similar to cybersecurity infrastructure. ![Screenshot at 00:00: A Gantt chart from Epoch AI illustrating the aggressive timeline for building gigawatt-scale data centers by major AI players like Anthropic, xAI, Microsoft, Meta, and Amazon, projecting some to be operational by 2026.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-00-00.png) ![Screenshot at 00:15: OpenAI blog post titled "Fighting the New York Times' invasion of user privacy," showing the company's strong defense against the data demands.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-00-15.png) ![Screenshot at 00:29: Cassandra Unchained tweet detailing Michael Burry's accusation that hyperscalers are artificially inflating earnings by extending asset useful life, citing an estimated $176 billion understatement in depreciation.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-00-29.png) ![Screenshot at 00:41: Google search results for 'project suncatcher,' showing news headlines about Google's 'moonshot' project to put AI data centers in space using solar power.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-00-41.png) ![Screenshot at 02:52: Demonstration of the 'Vibe Code' mobile app, which uses AI to generate images based on text prompts like creating a gradient background.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-02-52.png) ![Screenshot at 03:32: Prototype screen of the 'Vibe Code' app showing a 'Mood Tracker' feature allowing users to log feelings and activities.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-03-32.png) ![Screenshot at 06:18: Table from Michael Burry's claims showing the increasing useful life \(years\) for Network/Compute Depreciation across Meta, Google, Oracle, Microsoft, and Amazon from 2020 to 2025.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-06-18.png) ![Screenshot at 10:12: OpenAI blog post section titled "AI progress and recommendations," outlining their responsibility to guide AI power toward broad, lasting benefit.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-10-12.png) ![Screenshot at 12:58: OpenAI's direct answers section, where they state they strongly believe the NYT demand is an overreach that risks user privacy.](https://ss.rapidrecap.app/screens/qrGpdYtwuo0/00-12-58.png)