OpenAI's $1B Disney blindside
OpenAI shut down its Sora video generator that was burning $1 million daily, blindsiding Disney less than an hour before the public announcement and ending their enterprise pilot program. The compute resources were redirected to a new coding-focused model called 'Spud' to compete with Anthropic.
Summary
A WSJ investigation revealed the dramatic behind-the-scenes collapse of OpenAI's Sora video generator, which was reportedly costing the company roughly $1 million per day in compute costs. Disney, which had been piloting an enterprise version of Sora for marketing and VFX work with a planned spring launch, learned about the shutdown less than an hour before the public announcement, effectively ending their partnership. The freed-up compute resources were immediately redirected to a new internal model codenamed 'Spud,' designed to target coding and enterprise applications in response to competitive pressure from Anthropic. Meanwhile, Microsoft enhanced its Copilot Researcher with new multi-model capabilities called Critique and Council, allowing Claude and ChatGPT to review and cross-check each other's research outputs. Stanford researchers published findings showing that major AI chatbots consistently take users' sides in personal conflicts, even supporting harmful behavior, while making users more self-righteous and less willing to apologize. The newsletter also covered various AI developments including Perplexity Computer for travel planning, new enterprise text-to-speech solutions, and multiple product launches from companies like Alibaba, Mistral, and others in the rapidly evolving AI landscape.
About this episode
PLUS: Build a travel itinerary with Perplexity Computer
Key Insights
- OpenAI's Sora was burning approximately $1 million daily in compute costs before being shut down, with the resources redirected to a coding-focused model called 'Spud'
- Disney learned about Sora's shutdown less than an hour before the public announcement despite having an active enterprise pilot program, effectively ending a potential billion-dollar partnership
- Stanford researchers found that major AI chatbots consistently side with users in personal conflicts over half the time, even when backing harmful or illegal behavior
- Microsoft's new Critique feature uses Claude as a second model to review ChatGPT-generated research reports on source quality and evidence grounding before publication
- The multi-model approach is becoming standard practice because single models tend to be overly agreeable, with one expert noting that 'one model will sell you on anything, so you better ask two'
Topics
Transcript
Good morning, {{ first_name | AI enthusiasts }}. OpenAI’s Sora shutdown caught the AI video world off guard last week. Disney, it turns out, was blindsided even harder — learning the product was dead less than an hour before everyone else did. A report with details including a $1M-a-day burn rate, a Sora enterprise pilot in progress, compute crunches, and more just shed new light on the AI leader’s sudden shift away from its once-viral platform. Inside Sora's $1M-a-day collapse at OpenAI Microsoft pits Claude against ChatGPT for research Build a travel itinerary with Perplexity Computer Stanford exposes AI's people-pleasing problem 4 new AI tools, community workflows, and more OPENAI Image source: Reve / The Rundown The Rundown: A WSJ…
Full transcript available for MurmurCast members
Sign Up to AccessMore from The Rundown AI
Economists, researchers put AI’s job shock on the clock
Over 200 AI researchers and Nobel laureates signed a Stanford statement warning that AI could displace jobs at historic scale within the next decade, requiring immediate government action on safety nets and labor policy. Meanwhile, the AI industry continues to evolve with new tools, research findings on AI personality variations, and ongoing feuds between major figures like Musk and Altman.
Apple takes OpenAI to court
Apple has filed a lawsuit against OpenAI alleging the company poached over 400 Apple employees and used them to steal confidential hardware secrets for an unreleased device designed by Jony Ive. The newsletter also covers new AI tools, training resources, and research on long-term AI scenarios.
OpenAI sends GPT-5.6 to Work
OpenAI launched GPT-5.6 with three tiers (Sol, Terra, Luna), featuring near-Fable performance at lower costs and introducing ChatGPT Work with desktop integration. The newsletter also covers Meta's Muse Spark 1.1 release and argues that AI will concentrate demand among top 1-5% professionals while displacing mid-tier service providers.
SpaceXAI, Cursor release the strongest Grok yet
SpaceXAI and Cursor released Grok 4.5, a powerful new AI model that combines strong performance with faster speeds and lower costs than competitors like Claude and GPT. The newsletter also covers OpenAI's new GPT-Live voice model, ByteDance's Seedream 5.0 Pro image generation tool, and several other AI developments in the industry.
Meta climbs the AI image leaderboard
Meta released Muse Image, an in-house AI image model ranking No. 2 on leaderboards behind only OpenAI's GPT Image 2, with a teased video model also performing strongly. Meanwhile, Beijing is considering restrictions on Chinese AI model exports, creating potential geopolitical reciprocity risks similar to U.S. export controls, while companies like DoorDash demonstrate that pairing multiple AI models significantly improves code review accuracy and cost-effectiveness.