AI Is Ready for More Than Chatbots, but Companies Are Falling Behind, Says OpenAI CFO

Artificial intelligence is no longer a futuristic experiment or a side project for tech teams. According to OpenAI’s Chief Financial Officer Sarah Friar, AI has now reached the level of core economic infrastructure. Yet despite its growing power and reach, most organizations are still failing to unlock its full value.

Speaking after attending the World Economic Forum’s annual meeting in Davos, Friar shared a clear message: there is a growing gap between what AI is capable of doing today and how companies are actually using it. This gap, which she refers to as a capability overhang, could determine which businesses and countries pull ahead in the coming years and which fall behind.

AI Has Entered a New Era

At Davos this year, AI dominated conversations in a way it never had before. Friar described the shift as striking. In previous years, artificial intelligence was often discussed as a future trend or an experimental tool. This time, it was treated as essential infrastructure, on par with energy systems, geopolitics, and national security.

That change reflects how deeply AI has already embedded itself into the global economy. Governments, corporations, and financial leaders are no longer debating whether AI matters. Instead, they are trying to figure out how to deploy it effectively and responsibly.

Yet despite this widespread recognition, Friar believes most organizations are only scratching the surface.

The Capability Overhang Problem Explained

The central idea Friar highlighted is what OpenAI calls capability overhang. Simply put, AI systems today can do far more than what most people and companies are asking of them.

Advanced AI models can assist with complex coding, deep research, strategic planning, data analysis, and even act as thought partners for decision-making. But in reality, many businesses still use AI in limited ways, such as basic content generation or simple customer support tasks.

According to Friar, this mismatch is not due to a lack of belief in AI. It is largely a matter of experience and execution. Companies often struggle to integrate AI deeply into workflows, redesign processes around it, or train teams to use it at a more advanced level.

Power Users Are Playing a Different Game

At OpenAI, Friar sees a clear difference between average users and frontier users. These advanced users are extracting far more value from AI by pushing it to its limits.

She notes that frontier users consume roughly seven times more AI intelligence than the average user. They rely on AI for heavy coding work, in-depth research, and complex problem-solving. Instead of treating AI as a tool, they treat it as a collaborator.

This level of engagement allows them to move faster, experiment more, and make better-informed decisions. It also shows what is possible when organizations commit to using AI as a core part of their operations rather than an add-on.

Global Differences in AI Adoption

OpenAI’s recent research, Ending the Capability Overhang, reveals that this gap is visible not just at the company level but also at the country level.

Across more than 70 countries where ChatGPT is widely used, there are major differences in how advanced AI features are adopted. Some countries use these tools up to three times more per person than others, and these differences are not explained by income alone.

Large economies like the United States and India lead in total number of users. Smaller but wealthier nations such as Singapore and the Netherlands rank highest in per-capita usage. However, what stands out most is that advanced AI adoption is spreading in unexpected places.

Countries like Pakistan and Vietnam are among the heaviest users of agentic AI tools, using them at more than twice the global average. This suggests that innovation is not limited to rich economies. In many cases, necessity and ambition are driving faster adoption.

Productivity Gains Are Already Showing

The impact of advanced AI use is not theoretical. OpenAI’s findings suggest that early adopters are already seeing real productivity benefits.

Workers who use AI deeply are able to shift their focus away from repetitive tasks and toward more complex and creative work. This leads to faster innovation, the development of new products and services, and overall economic growth.

In the long run, these productivity gains can translate into higher living standards and more competitive economies. The implication is clear: those who close the capability gap sooner will likely enjoy lasting advantages.

CFOs Are Focused on Results, Not Hype

Another insight that stood out to Friar in Davos came from discussions with other finance leaders. At a gathering of CFOs, she observed a shared mindset grounded in pragmatism.

There is broad agreement that AI is inevitable. The real debate is not about whether to adopt it, but how to make it pay off. CFOs are focused on return on investment, data quality, and system simplicity. For them, AI adoption is a change-management challenge rather than a question of belief.

This perspective aligns with Friar’s broader message. The biggest obstacles to AI adoption are often organizational, not technological.

OpenAI’s Rapid Growth Mirrors AI Demand

OpenAI’s own growth reflects the rising demand for AI capabilities. Since Friar joined the company as CFO in June 2024, the company’s financial performance has accelerated dramatically.

In 2023, OpenAI generated $2 billion in annual recurring revenue. That figure rose to $6 billion in 2024 and surged past $20 billion in 2025. At the same time, the company’s computing capacity expanded nearly tenfold over two years.

This growth highlights the tight link between AI capabilities and infrastructure. As models become more powerful and more widely used, demand for computing resources continues to skyrocket.

Friar has also indicated that an initial public offering remains a possibility in the future, though timing will depend on broader strategic considerations.

Expanding Beyond Core AI Tools

In addition to scaling its infrastructure, OpenAI is expanding into new consumer-facing areas. One recent example is the launch of ChatGPT Health, a dedicated experience that allows users to connect medical records and wellness apps to personalize interactions.

The goal is to make AI more useful in everyday life while maintaining strict privacy standards. OpenAI has emphasized that personal medical data will not be used to train its models.

This move reflects the same philosophy Friar described in Davos: pairing powerful technology with practical applications that deliver real value.

Why the Capability Gap Matters

The mismatch between AI’s abilities and its current use is not just a missed opportunity. Over time, it could widen economic and competitive gaps between companies and countries.

Organizations that fail to move beyond shallow AI use may find themselves outpaced by competitors who integrate AI deeply into their operations. Similarly, countries that encourage advanced AI adoption could gain long-term advantages in productivity and innovation.

Friar’s message is not one of alarm but of urgency. The tools are already here. The challenge now is learning how to use them well.

The Road Ahead for AI Adoption

Closing the capability overhang will require more than enthusiasm. It will take training, cultural change, process redesign, and a willingness to experiment. It will also require leaders to think beyond quick wins and invest in long-term transformation.

As AI continues to evolve, the gap between what it can do and how it is used will either narrow or grow wider. The choices organizations make today will shape their competitiveness for years to come.