OpenAI released its first State of Enterprise AI report on December 8, 2025, drawing from usage data across more than one million business customers and surveys of 9,000 workers at nearly 100 enterprises. The headline numbers: weekly ChatGPT Enterprise messages grew 8x year-over-year, workers report saving 40 to 60 minutes daily, and a clear divide is emerging between organizations that have embedded AI into core workflows and those that haven't moved beyond basic use.
The report arrives at a moment when OpenAI faces intensified competition from Google's Gemini, Anthropic's Claude, and Meta's Llama models. It reads as part market research, part reassurance to enterprise customers that their investment is paying off. The data points, drawn from de-identified aggregated usage patterns and self-reported survey responses, offer the most detailed public picture yet of how large organizations actually use generative AI tools.
The Scale of Enterprise Adoption
OpenAI now serves more than 7 million ChatGPT workplace seats, up 40% in just two months. ChatGPT Enterprise seats specifically have increased approximately 9x year-over-year. The company announced in early November 2025 that it had crossed one million business customers globally, positioning itself as what it calls the fastest-growing business platform in history.
But raw seat counts tell only part of the story. The more telling metric is usage intensity. Average reasoning token consumption per organization increased by approximately 320x over the past 12 months. That figure suggests companies aren't just buying seats and letting them sit idle. They're feeding increasingly complex queries into models designed for multi-step reasoning.
Custom GPTs and Projects (configurable interfaces that let workers build repeatable workflows with specific instructions and knowledge bases) saw weekly users increase 19x year-to-date. About 20% of all Enterprise messages now route through these customized tools rather than the standard ChatGPT interface. BBVA, the Spanish banking group, regularly uses more than 4,000 custom GPTs across its operations.
On the API side, more than 9,000 organizations have now processed over 10 billion tokens, and nearly 200 have exceeded 1 trillion. Codex, OpenAI's code-generation model, saw weekly active users double and weekly messages increase 50% over the six weeks prior to the report's publication.
How Workers Say They're Using AI
The survey of 9,000 workers across nearly 100 enterprises produced specific productivity claims. Seventy-five percent of surveyed workers said AI improved either the speed or quality of their output. On average, ChatGPT Enterprise users attributed 40 to 60 minutes of time saved per active day to their use of AI.
Workers in data science, engineering, and communications reported the highest time savings, ranging from 60 to 80 minutes daily. Accounting and finance users reported the largest benefits per message sent, followed by analytics, communications, and engineering.
Broken down by function, 87% of IT workers reported faster issue resolution. Marketing and product users reported faster campaign execution at 85%. HR professionals reported improved employee engagement at 75%, and 73% of engineers reported faster code delivery.
The more granular finding concerns what workers say they can now do that they couldn't before. Seventy-five percent reported being able to complete tasks they previously could not perform, including programming support, code review, spreadsheet analysis and automation, technical tool development, and custom GPT design.
Coding-related messages increased across all functions, not just engineering. Outside of engineering, IT, and research, coding-related messages grew 36% over the past six months. Non-technical teams are increasingly doing work (basic scripting, data manipulation, automation) that was previously confined to specialized roles.
The Intensity Gap Between Leaders and Laggards
The report's most striking finding concerns the widening gap between frontier workers and median workers, and between frontier firms and median firms. OpenAI defines frontier users as those in the 95th percentile of adoption intensity.
Frontier workers send 6x more messages than the median worker. The gap widens for specific tasks: frontier workers send 17x as many coding messages as the median, 11x as many writing and communication messages, and 10x as many analysis and calculation messages.
Among workers who focus on data analytics, frontier users engage with the data analysis tool 16x more than the median. Users who engage across roughly seven distinct task types report five times more time saved than those using only four task types. Depth of use correlates directly with reported benefits.
At the firm level, the pattern repeats. Frontier firms (95th percentile) generate approximately 2x more messages per seat than the median enterprise and 7x more messages to custom GPTs. The companies seeing the largest returns have invested in workflow standardization and embedded AI into repeatable processes rather than treating it as an ad hoc productivity tool.
Among monthly active users, 19% have never used data analysis features, 14% have never used reasoning features, and 12% have never used search. Those numbers drop to 3%, 1%, and 1% respectively among daily active users. People who use AI more frequently use more of its capabilities.
One in four enterprises still has not enabled connectors that give AI secure access to company data inside core tools. The report frames this as low-hanging fruit: organizations that turn on integrations with systems like Slack, SharePoint, Google Drive, and GitHub get context-aware responses and can automate actions across their existing workflows.
Industry and Geography Patterns
Growth was broad-based across industries. The median sector expanded more than 6x year-over-year, and even the slowest-growing sector exceeded 2x.
The fastest-growing sectors by customer growth were technology at 11x, healthcare at 8x, and manufacturing at 7x. Professional services, finance, and technology operate at the largest absolute scale in terms of message volume.
Technology companies use the API at a rate 5x higher year-over-year, primarily building customer-facing applications. Their top use cases include in-app assistants and search, agentic workflow automation, and coding and developer tools.
Professional services firms concentrate API spend on coding and developer tools to build custom tooling. Finance organizations often start with customer support, which represents a large, scalable cost center with proven ROI. Coding and developer tools rank second in finance as firms invest in system migration and custom applications for trading, risk, and compliance.
Customer service and content generation now represent approximately 20% of API activity. Non-technology firm API use grew 5x year-over-year, indicating adoption is expanding beyond technology-led product embedding into broader operational deployments across industries.
Geographically, international growth is accelerating. Among the largest markets, Australia, Brazil, the Netherlands, and France showed the fastest growth in business customers, each increasing more than 143% year-over-year. The United States, Germany, and Japan rank among the most active markets by message volume. Japan has the largest number of corporate API customers outside the U.S.
International API customer growth exceeded 70% over the past six months. The U.K. and Germany now rank among the largest ChatGPT Enterprise markets outside the U.S. by customer count.
Case Studies: What Measurable Impact Looks Like
The report includes six detailed case studies that illustrate specific implementations and claimed outcomes.
Intercom built its Fin Voice customer service agent on OpenAI's Realtime API. Latency decreased 48% since March. Fin Voice resolves 53% of phone calls end-to-end, and calls that require human agents are resolved 40% faster once Fin Voice completes initial steps. Intercom says Fin is saving customers hundreds of millions of dollars annually across chat, email, and phone channels.
Lowe's deployed Mylow on Lowes.com and Mylow Companion for store associates across more than 1,700 stores. The tools answer nearly one million questions per month. When customers engage with Mylow during online visits, conversion rates more than double. Customer satisfaction scores increase 200 basis points when associates use Mylow Companion to help customers in store aisles.
Indeed used GPT-powered matching to generate personalized job invitations. In experiments, invitations with LLM-generated explanations increased started applications by 20% and improved downstream success (interviews and hires) by 13%. Job seekers using Career Scout, an AI career coach, find and apply to jobs 7x faster and are 38% more likely to be hired.
BBVA built a legal chatbot to validate corporate signatory authority in Mexico. The solution automates more than 9,000 queries annually and enabled BBVA to redeploy the equivalent of three full-time employees toward producing over 11,000 legal validations per year. This delivered 26% of the Legal Services division's annual savings target.
Oscar Health developed member-facing chatbots to answer benefits, cost, and general health questions. The chatbots integrate with Oscar's systems and data to personalize responses using medical records, claims, and customer service history. The platform answers 58% of benefits questions instantly and handles 39% of benefits messages without human escalation.
Moderna used ChatGPT Enterprise to streamline Target Product Profile development, a process that typically requires multi-week cross-functional effort. Teams review evidence packs of up to 300 pages. The system extracts key facts, generates draft sections, and flags errors for human oversight. A core analytical step that took weeks now takes hours in some cases. Moderna believes each day gained in early planning helps deliver treatments to patients more quickly.
External Validation and Context
The report cites external research to support its findings. A 2025 Boston Consulting Group study found that over the past three years, AI leaders achieved 1.7x revenue growth, 3.6x greater total shareholder return, and 1.6x EBIT margin compared to laggards. AI leaders also outperformed on non-financial measures including patent output and employee satisfaction.
The BCG study surveyed 1,250 senior executives and AI decision-makers across nine industries. It found that only 5% of companies qualify as future-built for AI, generating substantial value through innovation. Thirty-five percent are scaling and beginning to generate value. The remaining 60% report minimal revenue and cost gains. BCG's report noted that agentic AI accounts for about 17% of total AI value in 2025 and is expected to reach 29% by 2028.
Research from the St. Louis Federal Reserve found that U.S. generative AI adoption among workers ages 18 to 64 reached 54.6% by August 2025, up 10 percentage points from August 2024. Work-related adoption increased from 33.3% to 37.4% over 12 months. The share of work hours spent using generative AI increased from 4.1% in November 2024 to 5.7% in August 2025.
Academic studies cited in the OpenAI report include Noy & Zhang (2023), Dell'Acqua et al. (2023), Schwarcz et al. (2025), and Brynjolfsson et al. (2025), which found that AI has an equalizing effect, disproportionately aiding lower-performing workers.
What Leading Organizations Do Differently
The report identifies five patterns among organizations that capture the most value from AI:
Deep system integration: Leaders enable connectors to give AI secure access to company data inside core tools, enabling context-aware responses and automated actions.
Workflow standardization and reuse: They promote the creation, sharing, and discovery of repeatable solutions for common tasks. GPTs often power this work, while the most sophisticated organizations embed API-powered assistants directly into core internal systems.
Executive leadership and sponsorship: Leaders set clear mandates, secure resources, and align teams. They create space for experimentation while enabling deployment at scale.
Data readiness and evaluations: They codify institutional knowledge into machine-readable routines, build APIs for key data pipelines, and run continuous evaluations to track model performance on real-world outcomes.
Deliberate change management: They build structures that speed organizational learning, combining centralized governance and training with distributed enablement through embedded AI champions.
The report notes that OpenAI releases a new feature or capability roughly every three days. The primary constraints for organizations, it argues, are no longer model performance or tooling, but rather organizational readiness.
The Business Context
OpenAI's enterprise push comes against intensifying competition. Google reported that AI Overviews now has about two billion monthly users across more than 200 countries. Gemini App has more than 450 million monthly active users. ChatGPT reached 700 million weekly active users by August 2025 and now exceeds 800 million.
OpenAI's annual recurring revenue hit $13 billion by August 2025, up from $10 billion in June, with the company on track to surpass $20 billion by year-end. The company has five million paying business users, up from three million in June.
The infrastructure demands are substantial. OpenAI signed a seven-year, $38 billion agreement with Amazon Web Services. Its Stargate joint venture with SoftBank, Oracle, and MGX plans up to $500 billion in AI infrastructure over four years. OpenAI has a $30 billion-a-year lease with Oracle for 4.5 gigawatts of U.S. data center capacity and a five-year, $11.9 billion deal with CoreWeave.
What the Report Doesn't Address
The report is based entirely on OpenAI's own customer base and self-reported survey data. It does not compare outcomes across competing platforms or provide independent verification of productivity claims.
The survey methodology raises questions. Workers who respond to surveys about AI productivity may be more likely to be enthusiastic users. The 40 to 60 minute daily savings figure relies on self-reporting rather than measured time tracking.
The report does not address implementation costs, change management failures, or organizations that tried and abandoned enterprise AI rollouts. It presents successful case studies without discussing how representative they are of typical deployments.
Cost per user, total cost of ownership, and ROI calculations are absent. The report does not break down how the claimed productivity gains translate into actual business outcomes like revenue or profit impact for the average enterprise customer.
Security, compliance, and data governance challenges (significant concerns for regulated industries) receive no detailed treatment. The report mentions that one in four enterprises has not enabled data connectors but does not explore why.
What Comes Next
The report forecasts that the next phase of enterprise AI will be shaped by stronger performance on economically valuable tasks, better understanding of organizational context, and a shift from asking models for outputs to delegating complex multi-step workflows.
OpenAI expects organizations to discover new ways to serve customers and deliver value, not just improve efficiency. The company frames this as a transition from AI as productivity tool to AI as core organizational capability.
The gap between leaders and laggards presents both a warning and an opportunity. Organizations that have not yet moved beyond basic adoption can study the patterns of frontier workers and firms. The data suggests that depth of use matters: workers who engage across multiple task types and use advanced features report substantially larger productivity gains.
For enterprises still in pilot mode, the report offers a template: enable integrations, standardize workflows into reusable tools, secure executive sponsorship, invest in data readiness, and build change management structures that accelerate learning.
OpenAI says it will continue to share real-world evidence on how AI is influencing firms, workers, and the broader economy. The company's next enterprise milestone to watch is ChatGPT for Work seat growth and API token consumption through Q1 2026.
