EY’s latest Entrepreneur Ecosystem Barometer finds that 76% of established U.S. entrepreneurs have at least partially integrated AI, but the spending phase is giving way to an ROI test: revenue growth, customer experience and operating efficiency now matter more than adoption headlines.
As reported by Forbes and detailed in EY’s July 28 survey, the study covered 500 U.S.-based entrepreneurs whose companies generate at least $5 million annually. Ten percent said AI is fully integrated across their processes, yet more than one-third have cut spending on technology tools, including AI, during the past year when the value was not clear.
The top expected returns are direct and measurable. Among businesses already investing in AI, 57% expect the greatest value in sales and marketing-led revenue growth, followed by operational cost reduction at 55%. Customer experience, product and service innovation, and back-office efficiency were close behind.
That is a meaningful shift for Windows-centric IT teams and managed service providers. A Copilot, CRM assistant, meeting transcription tool or custom workflow no longer wins budget approval merely because it demonstrates generative AI capability. It needs a baseline: quotes produced per salesperson, time spent on administration, error rates, conversion rates, support resolution time, or another metric that can be compared before and after deployment.
The practical lesson is to avoid treating AI as a standalone software category. The strongest candidates are narrow workflows with existing systems of record, clear users and an obvious handoff to a person. Microsoft 365 Copilot, Power Automate, Dynamics 365, Teams transcription and third-party line-of-business tools can fit that model—but only if they are attached to a process the business already measures.
The survey reports that 99% of respondents have some approach to AI security and data-risk management. That figure signals broad awareness, not necessarily mature governance. For smaller organizations especially, a policy document is not equivalent to enforcing Microsoft Entra ID conditional access, limiting which SharePoint and OneDrive data an assistant can retrieve, reviewing third-party app permissions, and keeping confidential content out of unapproved consumer AI services.
That does not make skills shortages less important. Thirty-seven percent cited talent as a barrier to growth, while 46% said talent and expertise stand in the way of scaling AI. The immediate need is not only for data scientists. It is for employees who can validate outputs, map business processes, manage data safely and identify where automation creates a real operational gain.
For IT leaders, the next AI project should therefore begin with a business owner and a measurable workflow—not a model comparison. The entrepreneurs EY surveyed are willing to use AI widely, but their message to vendors and internal technology teams is now blunt: show the money, or expect the budget to move elsewhere.
As reported by Forbes and detailed in EY’s July 28 survey, the study covered 500 U.S.-based entrepreneurs whose companies generate at least $5 million annually. Ten percent said AI is fully integrated across their processes, yet more than one-third have cut spending on technology tools, including AI, during the past year when the value was not clear.
AI budgets are being held to business metrics
The top expected returns are direct and measurable. Among businesses already investing in AI, 57% expect the greatest value in sales and marketing-led revenue growth, followed by operational cost reduction at 55%. Customer experience, product and service innovation, and back-office efficiency were close behind.That is a meaningful shift for Windows-centric IT teams and managed service providers. A Copilot, CRM assistant, meeting transcription tool or custom workflow no longer wins budget approval merely because it demonstrates generative AI capability. It needs a baseline: quotes produced per salesperson, time spent on administration, error rates, conversion rates, support resolution time, or another metric that can be compared before and after deployment.
The practical lesson is to avoid treating AI as a standalone software category. The strongest candidates are narrow workflows with existing systems of record, clear users and an obvious handoff to a person. Microsoft 365 Copilot, Power Automate, Dynamics 365, Teams transcription and third-party line-of-business tools can fit that model—but only if they are attached to a process the business already measures.
The bottleneck is integration, not access to models
EY says data readiness, integration complexity, talent and security are among the leading barriers to scaling AI. That makes the familiar work of IT administration newly central: identity controls, file permissions, data classification, application integration, endpoint management and audit trails determine whether an AI pilot can become a production service.The survey reports that 99% of respondents have some approach to AI security and data-risk management. That figure signals broad awareness, not necessarily mature governance. For smaller organizations especially, a policy document is not equivalent to enforcing Microsoft Entra ID conditional access, limiting which SharePoint and OneDrive data an assistant can retrieve, reviewing third-party app permissions, and keeping confidential content out of unapproved consumer AI services.
Hiring is not disappearing; the job mix is changing
EY found that 88% of respondents expect their workforce to grow in the next 12 months. Rather than positioning AI chiefly as a headcount-reduction tool, surveyed entrepreneurs described redesigning roles around a mix of human judgment and automated routine work.That does not make skills shortages less important. Thirty-seven percent cited talent as a barrier to growth, while 46% said talent and expertise stand in the way of scaling AI. The immediate need is not only for data scientists. It is for employees who can validate outputs, map business processes, manage data safely and identify where automation creates a real operational gain.
For IT leaders, the next AI project should therefore begin with a business owner and a measurable workflow—not a model comparison. The entrepreneurs EY surveyed are willing to use AI widely, but their message to vendors and internal technology teams is now blunt: show the money, or expect the budget to move elsewhere.
References
- Primary source: Forbes
Published: 2026-07-29T03:22:49+00:00
Loading…
www.forbes.com