OpenAI is making its most consequential strategic shift since launching ChatGPT. The company is discontinuing consumer products, redirecting computing resources toward a new professional-grade model, and reorienting the entire business around enterprise clients. As of this week, business clients now account for 40% of OpenAI's revenue — up from roughly 20% when CFO Sarah Friar joined in 2024 — and the company expects that figure to reach 50% by year-end.
The catalyst: a new model codenamed Spud, described internally as OpenAI's "smartest model to date," designed specifically for high-value professional tasks.
Sora Is Dead. Spud Is Coming.
The clearest signal of this pivot is what OpenAI has decided to stop doing. The company is shutting down Sora, its AI video generation product that Sam Altman promoted just months ago in a partnership with Disney. CFO Sarah Friar described the decision as "a bit heartbreaking" but necessary: "We realized it's not our main focus right now. We need to ensure our upcoming model has sufficient computing power."
That upcoming model is Spud. According to PBS reporting, it promises "enhanced reasoning, improved understanding of intent and dependencies, better follow-through, and more dependable output in production." Friar told the Associated Press: "A new model will be arriving from us shortly. We are very enthusiastic about it."
The timeline is compressed. OpenAI is discontinuing consumer experiments — including plans for advertising on ChatGPT and features allowing adult-themed conversations — to reallocate compute toward Spud and the enterprise stack.
The Numbers Behind the Pivot
OpenAI's user numbers are massive, but lopsided. The platform has over 900 million weekly users, but approximately 95% of them pay nothing. The economics are clear: free users consume compute, paying enterprise clients generate revenue.
The business shift is being driven by newly appointed Chief Revenue Officer Denise Dresser, the former Slack CEO who joined three months ago. Her mandate is positioning OpenAI as the premier platform for workplaces deploying AI agents. In an internal memo reported by The Verge, Dresser told employees: "It's evident to me that companies have moved beyond the experimental stage and are now utilizing AI for tangible work. Company leaders are acknowledging that AI represents perhaps the most significant shift of their lifetimes."
| Metric | Figure | Source |
|---|---|---|
| Weekly users | 900M+ | PBS / AP |
| Free users | ~95% | PBS / AP |
| Business revenue share (2024) | ~20% | PBS / AP |
| Business revenue share (now) | 40% | PBS / AP |
| Target by year-end | 50% | PBS / AP |
Sam Altman acknowledged the strategic refocus on the "Mostly Human" podcast earlier this month, stating that a more focused approach was necessary. Friar added: "In the tech sector, as companies grow, it's a natural occurrence to explore numerous exciting possibilities. Successful companies excel at narrowing their focus and refocusing, even if it is a painful process."
The Anthropic Pressure
The pivot does not exist in isolation. Anthropic claims annualized revenues of $30 billion — a figure OpenAI disputes, noting it may not account for revenue shared with cloud providers like Amazon and Google. Regardless of accounting methodology, the trend is undeniable: Anthropic's growth rate is outpacing OpenAI's.
Luke Emberson of Epoch AI told PBS: "The trends indicate that Anthropic is experiencing much faster growth than OpenAI. If this trend continues, they will likely surpass OpenAI soon."
The competitive dynamics are intensifying week by week. Just yesterday, Anthropic released Claude Opus 4.7, its most powerful generally available model, with significant improvements in software engineering, vision, and creative output. And Anthropic's restricted Mythos Preview model — too capable to release publicly — continues to generate industry buzz through its Project Glasswing cybersecurity initiative.
Dresser's internal memo took direct aim at Anthropic's positioning: "Their narrative is built on fear, restriction, and the notion that a small elite should control AI. Our optimistic message will prevail over time: create powerful systems, implement appropriate safeguards, broaden access, and support individuals in achieving more."
Half of America Is Already Using AI
The timing of OpenAI's business pivot aligns with a broader market reality. A new Epoch AI/Ipsos survey of over 2,000 U.S. adults, published April 13, found that 50% of Americans reported using an AI service in the past week. Among employed Americans who use AI, half reported using it at least as much for work as for personal tasks.
The workplace impact is already measurable:
- 27% of employed AI users say AI has replaced some of their existing work tasks
- 21% say they have started doing new tasks they would not have done without AI
- 76% of workers with employer-provided AI subscriptions use it primarily for work, compared to just 38% of free-tier users
That last data point is telling. The Ipsos data shows that when employers pay for AI access, work usage nearly doubles. This is precisely the market OpenAI is targeting with its enterprise pivot: corporations that will pay for premium AI tools and deploy them across entire workforces.
What OpenAI Is Betting On
The bet is straightforward: the consumer AI market is a race to the bottom where 900 million users generate minimal revenue, while the enterprise market rewards differentiated products with real pricing power. OpenAI is looking at a future where AI agents — autonomous systems that can complete multi-step professional tasks — are the core value proposition, not chatbot conversations.
The Spud model appears purpose-built for this vision. Unlike GPT-5.4, which excels broadly across coding, reasoning, and knowledge work, Spud is apparently optimized for dependability and follow-through in production environments — the traits enterprise customers care most about.
Both companies remain unprofitable. Skeptics point out that the AI industry is spending far more than it earns. But the competitive dynamics are forcing both labs to move fast: Anthropic's Claude is increasingly popular among software professionals, and ChatGPT's traffic share has dropped from 77% to 57% over the past twelve months while Gemini quadrupled its share to 25%.
The Bigger Picture
OpenAI's pivot reflects a pattern emerging across the AI industry in 2026: the era of AI as a novelty consumer product is ending. The companies that survive will be the ones that embed AI into professional workflows where it generates measurable ROI — not just engagement metrics.
For enterprise buyers, the competition between OpenAI and Anthropic is good news. Both companies are racing to build the most reliable, capable professional-grade models, and pricing pressure will likely benefit customers.
For the 855 million free ChatGPT users who never paid a cent, the message is less comforting: you were never the customer. You were the training signal.
---
Takeaway: OpenAI is discontinuing Sora and other consumer experiments to redirect compute toward Spud, a new professional model. Business revenue has doubled from 20% to 40% in under two years, with a 50% target by year-end. The shift is driven by intensifying competition with Anthropic, whose growth rate now outpaces OpenAI's, and a market reality confirmed by Epoch AI data showing that employer-paid AI subscriptions drive nearly twice the workplace adoption of free tools. The consumer-first era of ChatGPT is ending. The enterprise era is beginning.
---
AI-Generated Content
This article was researched, written, and verified by Sonarlink's AI. All claims are sourced from verified publications. No fake bylines.
More from Sonarlink
Anthropic Built Its Most Powerful AI Ever — Then Locked It Away
Claude Mythos Preview is the highest-performing AI model Anthropic has ever documented — and it's not for sale. Here's wh...
Claude Opus 4.7 Review: Sharper Engineering, Safer Cyber, Better Vision
Anthropic's latest generally available model excels at software engineering, instruction following, and creative output...
AI Agents in 2026: Best Agentic Workflow Tools for Enterprise
Enterprise AI agents moved from pilot to production in 2026. Here are the tools, frameworks, and real results driving th...