OpenAI released data on Monday detailing a surge in enterprise usage of its AI tools over the past year, with ChatGPT message volume increasing eightfold since November 2024 and workers saving up to one hour daily, amid competitive pressures from Google highlighted in a recent internal memo from CEO Sam Altman.
The data release comes one week after Altman distributed an internal “code red” memo addressing the threat posed by Google to OpenAI’s market position. This timing reflects OpenAI’s efforts to position itself as the leading provider of AI solutions for businesses. Enterprise adoption has expanded significantly, as evidenced by customer statistics. According to the Ramp AI Index, 36 percent of U.S. businesses use ChatGPT Enterprise, compared to 14.3 percent for Anthropic’s offerings. Despite this progress in the enterprise segment, OpenAI’s primary revenue source remains consumer subscriptions, which face direct competition from Google’s Gemini model.
OpenAI encounters competition not only from Google but also from Anthropic, whose revenue derives predominantly from business-to-business sales. Additionally, open-weight model providers are gaining traction among enterprise clients, intensifying the rivalry in this space. To sustain its operations, OpenAI has allocated substantial resources toward future development, committing $1.4 trillion to infrastructure over the coming years. This investment underscores the critical role of enterprise growth in supporting the company’s overall business model, as expanding corporate usage provides a stable foundation for long-term financial viability.
Ronnie Chatterji, OpenAI’s chief economist, addressed the economic implications during a briefing. He stated, “If you think about it from an economic growth perspective, consumers really matter. But when you look at historically transformative technologies like the steam engine, it’s when firms adopt and scale these technologies that you really see the biggest economic benefits.” Chatterji’s comments highlight the distinction between individual consumer engagement and the broader impact of organizational implementation, drawing parallels to past technological advancements that drove industrial productivity.
The latest findings indicate that enterprise adoption extends beyond initial implementation, with tools becoming embedded in daily operations. Organizations utilizing OpenAI’s API, which serves as the developer interface for integrating AI capabilities, now consume 320 times more “reasoning tokens” compared to one year ago. Reasoning tokens represent computational units used for complex tasks, such as advanced problem-solving. This escalation suggests businesses are applying AI to more intricate challenges or conducting extensive experimentation with the technology.
The rise in reasoning token consumption aligns with higher energy demands, as these processes require significant computational power. Such increases could elevate operational costs for enterprises, prompting questions about the long-term feasibility of current usage patterns. TechCrunch has sought clarification from OpenAI regarding how businesses allocate budgets for AI initiatives and the sustainability of this accelerated growth trajectory in token usage and associated expenses.
Shifts in deployment strategies further illustrate evolving enterprise practices. Usage of custom GPTs, which enable companies to embed proprietary knowledge into tailored AI assistants or streamline repetitive processes, has increased 19-fold this year. These custom GPTs now comprise 20 percent of all enterprise messages processed through ChatGPT. OpenAI cited BBVA, a digital banking institution, as an example, noting that the company employs more than 4,000 custom GPTs in its regular operations.
Brad Lightcap, OpenAI’s chief operating officer, discussed this trend during the briefing. He remarked, “It shows you how much people are really able to take this powerful technology and start to customize it to the things that are useful to them.” Lightcap’s observation points to the adaptability of OpenAI’s platform, allowing organizations to align AI functionalities with specific internal needs, from knowledge management to process automation.
Survey responses from enterprise users reveal tangible efficiency gains. Participants indicated average daily time savings of 40 to 60 minutes through the use of OpenAI’s enterprise products. These estimates focus on direct productivity improvements but do not account for initial setup efforts, such as system familiarization, prompt refinement, or output verification, which may offset some benefits during early adoption phases.
Enterprise workers also report expanded skill sets facilitated by AI integration. Three-quarters of surveyed individuals stated that AI tools allow them to perform tasks, particularly technical ones, that were previously beyond their capabilities. This includes activities requiring specialized expertise, broadening access to advanced functions across various roles within organizations.
A specific area of growth involves coding activities. OpenAI documented a 36 percent rise in coding-related messages from teams outside traditional engineering, IT, and research departments. This democratization of coding access enables non-specialists to engage in programming tasks, potentially accelerating development cycles. However, broader participation in code generation raises risks of introducing security flaws, bugs, or exploitable vulnerabilities into systems.
To mitigate these concerns, OpenAI has introduced Aardvark, an agentic security researcher tool currently available in private beta. Aardvark functions to identify bugs, vulnerabilities, and potential exploits in code and applications. Lightcap referenced this release when questioned about security implications, positioning it as a supportive measure for safer AI-assisted development practices.
Despite these advancements, utilization of premium features remains limited even among high-volume ChatGPT Enterprise users. Tools for data analysis, advanced reasoning, and integrated search see lower engagement rates. Lightcap explained this pattern by noting the necessity for organizational adjustments. He indicated that full adoption demands a fundamental change in approach, including deeper connections to internal data sources and process redesigns to explore AI’s full potential.
Lightcap emphasized that integrating advanced features requires time, as companies adapt workflows and gain familiarity with expanded possibilities. This gradual progression allows for measured implementation, ensuring alignment with existing infrastructure and operational norms.
The report also delineates disparities in adoption levels across enterprises. A growing divide separates “frontier” workers, who frequently employ multiple AI tools and achieve substantial time reductions, from “laggards” with minimal usage. Lightcap described varying organizational perspectives during the briefing. He said, “There are firms that still very much see these systems as a piece of software, something I can buy and give to my teams and that’s kind of the end of it. And then there are companies that are really starting to embrace it, almost more like an operating system. It’s basically a re-platforming of a lot of the company’s operations.”
Lightcap’s analogy contrasts superficial deployment—treating AI as a standalone application—with comprehensive integration, where AI underpins core business functions, akin to foundational software layers. This re-platforming involves restructuring operations to leverage AI across multiple facets of company activities.
OpenAI’s leadership views the current adoption landscape as a chance for trailing organizations to advance their AI strategies. Chatterji and Lightcap presented this divide as navigable, with opportunities for laggards to align with leading practices. For employees involved in training AI models to automate their roles, however, accelerating adoption may evoke a sense of impending change rather than opportunity.
