AI companies say businesses are capturing only a fraction of the technology’s potential

Wednesday 16 September 2026 - 23:00

By: Azzat Manal

$AI companies say businesses are capturing only a fraction of the technology’s potential$

Despite the rapid expansion of artificial intelligence across industries, technology executives say businesses are still at an early stage of turning the technology’s capabilities into measurable economic value.

Anthropic CEO Dario Amodei estimated that the global economy is currently capturing only around 5% to 10% of the potential value that artificial intelligence could generate. His remarks highlight a persistent gap between the capabilities of increasingly sophisticated AI systems and the extent to which companies have integrated them into everyday operations.

Amodei made the comments during Salesforce’s annual Dreamforce conference, where technology leaders discussed the development of AI and the challenges businesses face in moving from experimentation to broader deployment.

The adoption of AI remTITLE

ains uneven across companies and industries. While some organizations have incorporated generative AI into customer service, software development, marketing, data analysis and other business functions, many others are still testing the technology or limiting its use to specific tasks.

Salesforce CEO Marc Benioff said companies differ considerably in how quickly they are adopting AI. He acknowledged that technology providers have a role to play in helping customers overcome implementation challenges and make better use of new AI tools.

The comments come as businesses increasingly seek evidence that large investments in artificial intelligence can translate into higher productivity and revenue. Companies have spent heavily on computing infrastructure, software and AI-related services, but economists and business leaders continue to debate how quickly these investments will produce economy-wide gains.

One factor behind the gap is the difficulty of integrating AI into existing business processes. Deploying an AI model is often only the first step. Companies may also need to reorganize workflows, train employees, establish data-management systems and introduce safeguards before the technology can deliver significant improvements at scale.

Amodei’s assessment also comes amid a wider debate over the speed of AI development and the risks associated with increasingly capable systems. The Anthropic chief has recently warned that rapid advances could create significant challenges and has called for greater caution in the development of advanced AI.

A different position was expressed by Nvidia CEO Jensen Huang during the same conference. Huang argued that market incentives can play an important role in encouraging companies to develop and deploy safer AI systems, questioning whether additional legislation is necessarily required.

The contrasting views from Amodei, Benioff and Huang illustrate the range of opinions within the technology industry. While there is broad interest in expanding the economic use of AI, executives continue to differ over how quickly development should proceed and what safeguards are needed as the technology becomes more powerful.

The economic impact of AI will ultimately depend not only on advances in computing and model capabilities but also on how effectively businesses adapt their operations. Greater adoption could affect productivity, employment patterns and the way companies deliver products and services, although the scale and timing of those effects remain uncertain.

As organizations move beyond initial experimentation, the next phase of the AI boom is increasingly focused on practical implementation. Companies are expected to place greater emphasis on demonstrating tangible business outcomes, while policymakers and technology leaders continue to debate how innovation can be combined with appropriate safety measures.