OpenAI has recently released new data highlighting a dramatic surge in enterprise usage of its artificial intelligence tools over the past year. Notably, ChatGPT message volume has grown eightfold since November 2023, with workers consistently reporting daily time savings of up to an hour. These findings emerge just a week after CEO Sam Altman reportedly circulated an internal “code red” memo concerning the competitive threat posed by Google, underscoring OpenAI's strategic imperative to solidify its position as the preeminent enterprise AI leader amidst intensifying market pressures.

While OpenAI maintains a significant lead in the U.S. enterprise market, with approximately 36% of businesses utilizing ChatGPT Enterprise compared to 14.3% for Anthropic, according to the Ramp AI Index, a substantial portion of its revenue still originates from consumer subscriptions. This consumer base, however, faces increasing challenges from Google’s Gemini. Furthermore, OpenAI must contend with rival AI firm Anthropic, which primarily generates revenue from B2B sales, and a growing number of open-weight model providers vying for enterprise customers. With infrastructure commitments totaling $1.4 trillion over the coming years, fostering robust enterprise growth is deemed essential to OpenAI's long-term business model. Ronnie Chatterji, OpenAI’s chief economist, emphasized this point during a briefing, stating that while consumers are important for economic growth, it is the adoption and scaling of transformative technologies by firms that historically yield the most significant economic benefits, akin to the impact of the steam engine.

OpenAI’s new data suggests that adoption within larger enterprises is not merely increasing but becoming more deeply integrated into daily workflows. Beyond a rise in message volume, organizations leveraging OpenAI’s API (developer interface) are consuming 320 times more “reasoning tokens” than they did a year ago. This metric implies that companies are increasingly employing AI for more complex problem-solving tasks, or alternatively, are extensively experimenting with the new technology, potentially leading to high token consumption that may be expensive and unsustainable in the long term. TechCrunch has engaged OpenAI regarding enterprise budget allocation for AI and the sustainability of this rapid growth rate.

The company is also observing notable shifts in how its tools are deployed. The use of custom GPTs, which enable companies to embed institutional knowledge into specialized assistants or automate specific workflows, has jumped nineteen-fold this year and now accounts for 20% of all enterprise messages. OpenAI cited digital bank BBVA, which reportedly utilizes over 4,000 custom GPTs, as a prime example. Brad Lightcap, OpenAI’s chief operating officer, highlighted that this demonstrates the extent to which powerful technology can be customized to serve specific utility for users. These integrations have translated into meaningful time savings, with participants reporting saving 40 to 60 minutes daily using OpenAI’s enterprise products, though this figure may not account for time spent learning systems, crafting prompts, or correcting AI output.

Additionally, enterprise workers are increasingly leveraging AI tools to broaden their own capabilities. Three-quarters of those surveyed indicated that AI enables them to perform tasks, including technical ones, that were previously beyond their scope. OpenAI reported a 36% increase in coding-related messages originating from teams outside traditional engineering, IT, and research departments. While OpenAI champions this as democratizing access to skills, it also acknowledges the potential for increased security vulnerabilities and other flaws due to a rise in less supervised “vibe coding.” When questioned on this, Lightcap referenced OpenAI’s recent release of Aardvark, an agentic security researcher currently in private beta, as a potential solution for detecting bugs, vulnerabilities, and exploits.

Despite the observed growth, OpenAI’s report also revealed that even the most active ChatGPT Enterprise users are not fully utilizing the most advanced tools available, such as data analysis, reasoning, or search functions. During the briefing, Lightcap mused that this underutilization stems from the fact that fully adopting AI systems necessitates a significant mindset shift and deeper integration with enterprise data and processes. He suggested that the adoption of these advanced features will naturally take time as companies retool their workflows to better comprehend the full spectrum of possibilities. Lightcap and Chatterji also emphasized a report finding that indicated a “growing divide in AI adoption,” wherein “frontier” workers are using more tools more frequently, leading to greater time savings compared to “laggards.” Lightcap differentiated between firms that view AI as a mere piece of software to be purchased and distributed, and those that are truly embracing it “almost more like an operating system,” fundamentally re-platforming many of their company’s operations. OpenAI’s leadership, keenly aware of the firm’s substantial infrastructure commitments, framed this divide as an opportunity for laggards to catch up. For workers concerned about AI systems replicating their work, however, “catching up” might evoke a sense of impending change.