The State of Enterprise AI

The adoption of artificial intelligence within large organizations has shifted from experimental pilots to core infrastructure. According to the “State of Enterprise AI 2025 Report” by OpenAI, more than 1 million business customers now utilize their tools, with significant growth in workflow integration and API consumption. The data indicates that while adoption is global and accelerating, a distinct gap in maturity is emerging between leading “frontier” (the top 5% of users) organizations and the median enterprise.

This analysis breaks down the current landscape of enterprise AI, examining usage patterns, productivity impacts, and the specific strategies separating high-performing organizations from laggards.

What is The State of Enterprise AI?

The state of enterprise AI in 2025 was defined by a transition from individual task assistance to complex, multi-step workflow automation. Organizations are no longer simply asking models for outputs; they are delegating deeper operational processes to AI systems.

Key metrics defining this landscape include:

• Scale of adoption: There are now over 1 million business customers using OpenAI’s tools, with 7 million workplace seats.
• Intensity of use: API reasoning token consumption per organization increased by 320x year-over-year.
• Global reach: International growth is outpacing early adopters, with countries like Australia, Brazil, and the Netherlands showing customer growth rates exceeding 150%.

Accelerating usage and workflow integration

The aggregate volume of AI interaction within enterprises has grown significantly. Since November 2024, weekly enterprise messages have increased approximately 8x, with the average worker sending 30% more messages. This growth is driven not just by more users, but by a deepening intensity of usage per employee.

The rise of custom GPTs and projects

A primary driver of this deepened integration is the shift toward configurable interfaces. Custom GPTs and “Projects” allow organizations to tailor AI with specific instructions, knowledge bases, and custom actions.

• Adoption Rate: Weekly users of Custom GPTs and Projects have increased by 19x year-to-date.
• Workflow Share: Approximately 20% of all enterprise messages are now processed via a Custom GPT or Project.
• Function: These tools are primarily used to codify institutional knowledge or automate workflows through internal system integrations.

Leading organizations, such as BBVA, report deploying more than 4,000 GPTs, suggesting that AI agents are becoming persistent tools embedded in daily operations.

API consumption and developer workflows

Beyond chat interfaces, companies are integrating models directly into products and internal systems via APIs.

• Token volume: Over 9,000 organizations have processed more than 10 billion tokens.
• Reasoning models: The 320x increase in reasoning token consumption indicates that firms are systematically integrating more intelligent models into products and services.
• Software development: Usage of Codex for end-to-end software tasks (generation, refactoring, testing) is gaining traction, with weekly active users doubling in recent weeks.

The widening gap between leaders and laggards

A critical finding in the 2025 report is the divergence in adoption intensity. While tools are broadly available, the depth of utilization varies dramatically between “frontier” workers (the top 5th percentile) and median users.

Individual usage disparities

Frontier workers operate with a significantly higher intensity than their peers:

• Message volume: Frontier workers send 6x more messages than the median worker.
• Task-specific intensity: In data analytics roles, frontier workers use analysis tools 16x more than the median.
• Coding gap: The largest relative gap exists in coding, where frontier workers send 17x as many messages as the median.

Organizational maturity gaps

This divide is mirrored at the firm level. Frontier firms generate approximately 2x more messages per seat than the median enterprise. More notably, they generate 7x more messages to Custom GPTs. This suggests that leading firms are investing in AI strategy and infrastructure that promotes standardized, repeatable workflows rather than ad-hoc usage.

Conversely, a significant portion of active users have yet to engage with advanced capabilities. Among monthly active users, 19% have never used data analysis features, and 14% have never utilized reasoning models.

Impact on productivity and workforce capabilities

Data from surveyed enterprises indicates that AI adoption is yielding measurable time savings and expanding the technical capabilities of non-technical staff.

Quantifiable time savings

Enterprise users report saving between 40 and 60 minutes per day on average. These savings are not uniform across functions:

• High savers: Data science, engineering, and communications workers report saving 60–80 minutes per day.
• Impact correlation: Users who engage across seven or distinct task types report saving five times more time than those using AI for only four task types.

The expansion of technical work

AI is blurring traditional role boundaries. Non-technical teams are increasingly performing tasks previously restricted to specialists.

• Coding across functions: Outside of engineering and IT, coding-related messages have grown by 36% over the past six months.
• Skill acquisition: 75% of workers report being able to complete tasks they previously could not perform, such as data analysis and technical troubleshooting.

To support this shift, organizations are increasingly investing in Vibe Coding workshops and Cursor training to upskill employees in AI-assisted development.

Industry-specific adoption patterns

While growth is occurring across all sectors, the rate of acceleration varies. The technology sector leads in pure growth, but traditional industries are rapidly scaling their deployments.

Sector growth rates (Year-over-Year)

1. Technology: 11x growth.
2. Healthcare: 8x growth.
3. Manufacturing: 7x growth.

Use case diversification

The application of AI varies by industry needs:

• Technology: Primarily focuses on building and scaling customer-facing applications and coding workflows.
• Manufacturing: Organizations in this sector are leveraging AI to streamline operations. Specialized AI consultancy for manufacturing often focuses on supply chain and process optimization.
• Retail: Retailers use AI for customer experience and associate support. For example, Lowe’s deployed AI to answer product questions for associates, resulting in a 200 basis point increase in customer satisfaction. Retailers looking to replicate this success often engage in AI consultancy for retail.
• Finance: Financial institutions prioritize customer support and compliance. BBVA automated legal queries to unblock commercial operations, delivering 26% of their legal division’s annual savings KPI.

Real-world case studies in enterprise AI

The report highlights several organizations that have moved beyond pilots to achieve material business impact.

Intercom: Customer service automation

Intercom utilized the Realtime API to build “Fin Voice,” a voice AI agent.

• Challenge: Reducing latency in phone support to prevent customer abandonment.
• Solution: An AI agent capable of handling complex, multi-step requests with low latency.
• Outcome: Latency decreased by 48%, and the AI resolves 53% of calls end-to-end.
• Relevance: This demonstrates the efficacy of modern customer service automation in high-stakes environments.

Moderna: Accelerating R&D

Moderna applied AI to the development of Target Product Profiles (TPPs), a critical step in drug development.

• Challenge: Reviewing evidence packs of up to 300 pages to create product blueprints.
• Solution: Using ChatGPT Enterprise to extract facts, generate draft sections, and flag details for human review.
• Outcome: Core analytical steps were reduced from weeks to hours.

Steps for organizational readiness

The primary constraints for AI adoption in 2025 are frequently organizational rather than technical. Leading firms consistently exhibit specific operational behaviors that facilitate scale.

1. Deep system integration: Leaders enable connectors to give AI secure access to company data, providing context-aware responses.
2. Workflow standardization: Successful firms actively promote the sharing of repeatable solutions (Custom GPTs) rather than relying on individual experimentation.
3. Executive sponsorship: Clear mandates and resource allocation from leadership are required to transition from pilot to production. AI for executives programs are essential for leaders to understand how to steer this transition.
4. Change management: Systematic investment in governance, training, and internal “AI champions” is necessary to bridge the gap between frontier and median workers. Organizations often require dedicated AI change management strategies to execute this effectively.

Conclusion

The 2025 landscape of enterprise AI is characterized by rapid scaling and a clear distinction between organizations that treat AI as a peripheral tool and those that integrate it as core infrastructure. While productivity gains are evident, with 40 to 60 minutes saved daily per user, the deeper value lies in the expansion of workforce capabilities and the automation of complex workflows. As the gap between frontier and median adopters widens, the focus for enterprises must shift from access to integration, standardization, and comprehensive workforce upskilling.

Frequently asked questions (FAQ)