OpenAI Abandons Sora to Prioritize 'Spud' Model and Combat Anthropic's Enterprise Lead
OpenAI Abandons Sora to Prioritize 'Spud' Model and Combat Anthropic's Enterprise Lead
OpenAI is discontinuing its Sora video generator to reallocate computing power toward its upcoming 'Spud' AI model. The strategic pivot, which includes abandoning a $1 billion Disney partnership, aims to position OpenAI against Anthropic in the competitive enterprise sector ahead of a planned IPO.
In a dramatic recalibration of its artificial intelligence roadmap, OpenAI has officially discontinued its Sora video generation platform. The decision, which halts both the consumer application and developer API, marks a significant departure from the company's previous multi-modal "side quests" in favor of core enterprise capabilities.
By freeing up massive clusters of GPU compute, OpenAI is clearing the runway for its next major large language model, internally codenamed "Spud".
This pivot arrives during a critical juncture for the AI giant, which faces intensifying enterprise competition from Anthropic and is actively restructuring its leadership and product divisions ahead of a highly anticipated initial public offering (IPO).
The End of the Video Era for OpenAI
Launched to massive fanfare and briefly topping the iOS App Store, Sora captured public imagination with its hyper-realistic video generation. However, behind the scenes, the platform had become a logistical and financial burden. Employees reportedly viewed Sora as a "drag" on the company's computing resources during a period when raw processing power is the primary bottleneck for advancing foundational models.
The immediate casualty of this shutdown is OpenAI's landmark $1 billion partnership with Disney. Reached late last year, the deal would have licensed iconic Disney characters for use within the Sora ecosystem. In a brief statement following the shutdown, Disney noted it "respects OpenAI's decision to exit the video generation business".
Meanwhile, the remnants of the Sora team, led by Bill Peebles, will pivot toward "world simulation" research for robotics, focusing on the long-term goal of automating the physical economy.
Enter 'Spud': The Superapp Strategy
With the massive compute overhead of video generation removed, OpenAI is reportedly weeks away from unveiling "Spud". In internal memos, CEO Sam Altman described the upcoming model as "very strong," emphasizing its potential to "really accelerate the economy".
Rather than chasing flashy consumer demos, Spud appears squarely aimed at productivity. Industry leaks suggest the model will serve as the engine for an upcoming OpenAI desktop "superapp". This unified platform will reportedly weave together:
- ChatGPT: The core conversational interface.
- Codex: Advanced software engineering and coding agent capabilities.
- Atlas: A natively integrated browser for seamless web execution.
To support this shift, Fidji Simo's product organization has been aggressively rebranded as "AGI Deployment," signaling a hyper-focus on shipping capable, scalable enterprise tools rather than experimental media applications.
The Anthropic Threat and the Enterprise Pivot
OpenAI's retreat from video is not occurring in a vacuum. It is a direct response to the meteoric rise of Anthropic in the B2B sector. While OpenAI spent months navigating copyright headaches and compute shortages for Sora, Anthropic doubled down on core reasoning, coding, and agentic workflows.
Through tools like Claude Code and its newly launched Mac computer-use capabilities, Anthropic has captured significant mindshare among developers and Fortune 500 companies. Anthropic's approach—favoring deep, autonomous text and code execution over multimodal spectacle—has proven highly lucrative.
By shelving Sora, OpenAI is tacitly acknowledging that the immediate battleground for AI supremacy lies in enterprise productivity, not Hollywood disruption. Spud and the superapp strategy represent OpenAI's "code red" effort to reclaim the narrative and close the workflow gap with Claude.
IPO Ambitions and the Economics of Compute
Beneath the product announcements lies a stark financial reality. OpenAI is reportedly targeting a late-2026 IPO with an astronomical valuation approaching $1 trillion. To justify that multiple, the company must demonstrate robust, high-margin revenue streams and disciplined resource allocation.
Sora, with its massive inference costs and lingering copyright liabilities, did not fit that narrative. The company is currently raising an additional $10 billion to push its total funding past the $120 billion mark, while Altman has formally handed off safety and security oversight to Mark Chen. This allows Altman to focus entirely on capital raising, securing supply chains, and driving global data center expansion.
The message from San Francisco is clear: the era of AI experimentation is ending, and the era of ruthless enterprise execution has begun. Whether 'Spud' can effectively neutralize Anthropic's momentum remains to be seen, but OpenAI is finally bringing its full compute weight to the fight.