‘These leaders recognize AI as the most consequential shift of their lifetime, and they’re asking us how to reinvent their companies around it’
OpenAI is refocusing its business on workplace professionals, betting that AI “co‑workers” embedded in everyday tools will drive its next phase of growth and help cover the massive costs of running its models.
Denise Dresser, OpenAI’s first chief revenue officer, says corporate leaders are no longer experimenting at the margins.
“What has struck me most is their immense sense of urgency and readiness,” she said about her first 90 days. “These leaders recognize AI as the most consequential shift of their lifetime, and they’re asking us how to reinvent their companies around it.”
Her comments align with OpenAI’s broader shift, reported by the Canadian Press (CP), away from splashy consumer projects and towards “high‑value professional work” as the company races rival Anthropic to secure enterprise customers and move towards profitability.
Earlier this year, Anthropic, the US company behind the Claude chatbot, launched an AI assistant for corporate legal departments that can help with contract review, non‑disclosure agreement screening and routine document workflows, according to media reports. The tool is being offered as part of the firm’s Cowork suite and is intended to sit alongside in‑house lawyers rather than replace them.
A unified AI ‘superapp’
To meet that demand, OpenAI is rolling out new models geared to professional work. CP reports the company plans to introduce a system for “high‑value professional work”, codenamed “Spud”, which it describes as its “smartest model yet”, with “stronger reasoning, better understanding of intent and dependencies, better follow‑through and more reliable output in production.” It has also launched GPT‑Rosalind, aimed at drug discovery and life sciences research.
Dresser outlined a broader enterprise platform built around OpenAI Frontier, described as “the underlying intelligence layer governing all of a company’s agents, and a unified AI superapp as the primary experience where employees get things done.” Rather than scatter separate tools across different teams, Frontier is meant to power AI “co‑workers” that move across systems and data, grounded in a company’s own context.
“One thing I hear over and over is that companies are tired of AI point solutions that don’t talk to each other and just create chaos,” she said. Frontier, she added, is helping customers such as Oracle, State Farm and Uber “build, deploy, and manage agents company‑wide” that operate across tools and retain context over time.
Dresser says the focus is now on integrating AI directly into daily workflows through a “unified AI superapp” combining ChatGPT, coding assistant Codex and “agentic” tools in one workplace interface. She said Codex has reached 3 million weekly active users, OpenAI’s APIs process more than 15 billion tokens per minute, and newer models such as GPT‑5.4 are driving “record engagement across agentic workflows.”
Recently, Ottawa launched a national initiative to build large‑scale AI supercomputing capacity that is expected to accelerate AI adoption across sectors. The federal government opened a competitive call for applications under the AI Sovereign Compute Infrastructure Program, inviting eligible organisations to propose large‑scale, AI‑optimised high‑performance computing systems based in Canada.
Focus on enterprise
Dresser said enterprise clients now generate more than 40% of OpenAI’s revenue and are on track to reach parity with consumer by the end of 2026. That builds on the company’s consumer strength, including ChatGPT’s 900 million weekly users, which help familiarise employees with the technology before it is formally rolled out inside organisations.
CFO Sarah Friar has given a similar picture. She told The Associated Press, in comments carried by CP, that business customers accounted for about 20% of OpenAI’s revenue when she joined in 2024, rising to roughly 40% and expected to reach about half by year‑end. At the same time, CP reports that about 95% of ChatGPT’s users “don’t pay anything”, highlighting why the company is leaning hard into paid enterprise adoption.
OpenAI and Anthropic, valued at US$852 billion and US$380 billion respectively, both lose more money than they make, according to CP. The privately held firms are under growing pressure to translate intense interest in AI into durable corporate revenue as they consider eventual stock market listings.
High stakes in an intensifying race
CP reports that OpenAI has stepped back from some consumer products, including its Sora video generator app, to free up computing power for its next‑generation models. Friar told AP, as cited by CP, that shutting Sora was “a little heartbreaking”, but “it’s not the main event right now. We need to make sure that our new model that’s coming has enough compute.”
CP notes that Anthropic has introduced advanced models including Claude Mythos and Opus 4.7 and has claimed annualised revenue of US$30 billion, intensifying the race for enterprise accounts. Researcher Luke Emberson of Epoch AI told CP that Anthropic appears to be “growing much faster than OpenAI” and could overtake it if current trends continue.
In 2025, OpenAI also said it will release a new AI-powered jobs platform by mid-2026.
Both companies are also contending with the cost of electricity‑hungry data centres. CP reports that Anthropic has imposed rate limits on heavy users and that both firms have introduced tiered service, prompting AI critic Ed Zitron to warn of a potential “subprime AI crisis” if businesses become dependent on services that later tighten access or raise prices.
