OpenAI's Major "Slimming Down" on the Eve of Its IPO: Cutting Sora, Abandoning Disney, and Moving Away from Microsoft
- Core Viewpoint: In its IPO sprint phase, OpenAI is systematically cleaning up non-core businesses with high costs, low returns, or legal risks through a series of actions such as shutting down Sora, terminating the Disney agreement, and scaling back its e-commerce business. The aim is to focus on its core AI models and present a clearer, healthier growth story to the capital markets.
- Key Elements:
- Sora's Business Model Not Viable: Monthly revenue of only $367,000, daily operating costs as high as $15 million, annualized burn rate of approximately $5.4 billion, with revenue covering an extremely low proportion of costs, making it a financial burden.
- Termination of Disney Agreement: The $1 billion content licensing agreement was voided, eliminating potential legal risks and contractual uncertainties.
- Systematic "De-Microsoft-ization": Through equity dilution, canceling cloud priority rights, and signing a major agreement with Amazon, OpenAI is reducing its dependence on Microsoft while deploying its core enterprise platform on AWS.
- E-commerce Business Failure: The ChatGPT built-in e-commerce feature Instant Checkout was shut down due to a conversion rate close to zero, indicating its exploration failed and its exit from this field.
- Focus on Core Data: Preparing the IPO narrative by highlighting core metrics such as $20 billion in annual revenue and 800 million weekly active users, while cleaning up financial and operational "noise."
- Reference to Uber Precedent: Following Uber's example of divesting its high-cost autonomous driving division (annual burn of $457 million) before its IPO to achieve profitability. However, the scale of the Sora project OpenAI is cleaning up (annual burn of $5.4 billion) is an order of magnitude larger.
On March 24, OpenAI announced the shutdown of its video generation tool Sora, which had been online for less than six months, while simultaneously terminating a $1 billion content licensing agreement with Disney (as reported by Variety and Bloomberg). According to statistics, Sora's monthly revenue was $367,000, with a daily operating cost of $15 million. The revenue-to-cost coverage ratio was 0.08%.
Just the day before, according to documents obtained by CNBC, Microsoft was listed as the top risk factor in the IPO preparation documents sent by OpenAI to investors. Three weeks prior to that, according to Awesome Agents, ChatGPT's built-in e-commerce feature, Instant Checkout, was quietly shut down due to a conversion rate close to zero.
A company valued at $730 billion, during its IPO sprint, is not showcasing a growth story but is instead cutting product lines, ditching partners, and drawing lines with its largest shareholder. This doesn't look like contraction; it looks more like a planned narrative cleanup.
Sora: Technically Impressive, Commercially Disastrous
When Sora launched last October, it was a phenomenal product on the App Store, surpassing 1 million downloads within 5 days and reaching over 3 million total downloads. However, according to data from third-party tracking platforms Appfigures and Similarweb, its 30-day retention rate was only about 1%, compared to TikTok's 32% 30-day retention rate during the same period.
According to Appfigures data, January downloads fell 45% month-over-month to approximately 1.2 million. January revenue was $367,000, down 32% from the December peak of $540,000. According to estimates by Cantor Fitzgerald analyst Deepak Mathivanan, Sora's peak daily video generation was 11.3 million videos, with a generation cost of about $1.30 per video, resulting in a daily operating cost of about $15 million and an annualized burn rate of approximately $5.4 billion.
For OpenAI, Sora's problem isn't its technology; it's that the business model doesn't work. Burning $5.4 billion annually for less than $5 million in revenue is a toxic number in any IPO prospectus.
The collateral cost of shutting down Sora is the invalidation of Disney's $1 billion investment agreement. According to Variety, the agreement was originally set for three years, authorizing OpenAI to use over 200 characters from Marvel, Pixar, and Star Wars for training and content generation, but excluded characters linked to real actors' likenesses and voices. According to Bloomberg, the money hasn't actually been transferred yet.
De-Microsoft-ization: A 16-Month Systematic Project
Viewing Sora and the Disney deal as isolated incidents misses a more important thread. Starting from Altman's firing and reinstatement in November 2023, OpenAI has completed a six-step operation over 16 months, systematically demoting Microsoft from controller to minority shareholder.
The PBC restructuring in October 2025 was a key turning point. According to Fortune, Microsoft's stake was diluted from 32.5% to 27%, and its exclusive cloud priority rights were simultaneously revoked. Six days after the restructuring was completed, OpenAI signed a $38 billion cloud computing agreement with Amazon, with all deployment targets expected to be completed by the end of 2026, according to ESM China. Amazon's total commitment amounts to $50 billion, with $15 billion as an initial allocation and the remaining $35 billion tied to milestones like the IPO process.
More critical is the split in cloud architecture. Microsoft's Azure retained OpenAI's stateless API calls (i.e., the foundational inference services for ChatGPT and the API), but Frontier, OpenAI's stateful enterprise-grade Agent platform, is exclusively deployed on AWS. According to Windows Central, Microsoft believes this violates the original exclusivity clauses and is considering legal action.
According to Next Platform analysis, of Microsoft's $625 billion revenue backlog, approximately $281.3 billion comes from OpenAI's Azure commitment purchases, accounting for 45%. Microsoft's FY2026 capital expenditure is estimated at $100-$125 billion, while its AI revenue is only about $13 billion annualized. This creates a counterintuitive situation: while OpenAI is busy "de-Microsoft-izing," Microsoft's financial dependence on OpenAI may be deepening.
E-commerce's Zero Conversion and Narrative Focus
The shutdown of ChatGPT Instant Checkout didn't attract much attention, but it points to the same logic as Sora's shutdown. This feature, launched last September, originally promised integration with over 1 million Shopify merchants but only integrated about 12. According to Awesome Agents, by the time of shutdown, the purchase conversion rate was close to zero, and a state sales tax collection system was never established.
After the shutdown announcement, according to public market data, Expedia's stock rose 13.69%, Booking's rose 8.46%, and Shopify's rose 3.96%. The market's interpretation was clear: OpenAI's exit from e-commerce is a positive for existing players.
The simultaneous contraction of three product lines points to a common goal: focusing the IPO valuation narrative on the core AI model. According to disclosures by OpenAI CFO Sarah Friar and official OpenAI data, $20 billion in annual revenue, 800 million weekly active users, and 1 million enterprise customers are numbers sufficient to support a clean growth story. Sora's $5.4 billion annualized cost, e-commerce's zero conversion, the legal risks of the Disney deal, and the uncertainty of the Microsoft relationship are all noise on the IPO prospectus.
Pre-IPO "Slimming Down" Is Not Unique to OpenAI
In 2020, Uber sold its self-driving unit ATG to Aurora for $4 billion, receiving a 26% equity stake. ATG had an annual R&D expenditure of $457 million. After cutting it, Uber achieved its first post-IPO profit. WeWork also cut non-core businesses before its 2019 IPO, but acted too late, seeing its valuation plummet from $47 billion to $8 billion, leading to IPO failure and bankruptcy in 2023.
OpenAI's move is closer to Uber's: proactively cutting high-cost, low-return business lines to clean up the financial structure before going public. The difference lies in scale. Uber cut ATG, which burned $457 million annually; OpenAI cut Sora, which burns $5.4 billion annually—a difference of an order of magnitude. Coupled with the $1 billion Disney deal being scrapped, the e-commerce team being disbanded, and the public downgrading of the Microsoft relationship, OpenAI is conducting the largest business line cleanup in tech history during its IPO sprint.
Altman is telling a growth story through subtraction.
