AI adoption is near-universal, but only 10% of tech firms achieve ROI at scale: KPMG

AI adoption is now near universal among technology companies but converting that adoption into repeatable returns remains elusive. The finding shifts the AI story from adoption to execution, scale and economics. Only around 10% of technology organisations surveyed by KPMG say they have achieved AI at scale, delivering consistent ROI across multiple use cases, even as all respondents report active AI initiatives.

The finding comes from KPMG’s 2026 Global Tech Report, based on insights from more than 155 senior leaders at technology companies with revenue of at least $1 billion. The report says the technology sector is moving into a phase where execution, rather than adoption, is becoming the key differentiator.

The gap between adoption and scale

While AI is now embedded across technology organisations, most respondents say it contributes 21–40% of their total digital value. Yet only 10% report deploying AI use cases at scale while delivering ROI across multiple use cases. KPMG says 74% of technology organisations are investing more than $50 million annually in digital technologies, while 51% are spending more than $100 million.

The spending is also generating measurable value. Eighty-three percent of organisations say they realised more than $50 million in digital value over the past 12 months, while 62% say they generated more than $100 million. But investment alone is not the constraint. Sixty percent of respondents say legacy systems and processes reduce the effectiveness of technology investment.

“Technology leadership is no longer defined by innovation alone, but by how effectively organizations operationalize it at scale,” Anna Scally, Global Head of Technology, Media and Telecommunications at KPMG International, said in the report. “Differentiation will likely come from execution.”

The challenge is pronounced as companies move AI projects from experimentation into enterprise-wide deployment. Nearly nine in 10 technology organisations describe themselves as innovators or fast followers, while more than 90% report active deployment across AI, cloud, data and cybersecurity. However, the largest share across several technology functions report that their strategies are funded but are hitting barriers to scaling.

KPMG points to integration complexity, operating-model limitations and organisational structure as key factors affecting outcomes. Sixty-one percent of technology respondents say their technology strategies require frequent revision, reflecting the speed at which technologies are evolving.

Workforce becomes another fault line

The report points to a tension within the workforce. While 90% of respondents say their talent strategy includes AI roles and 89% are investing in agentic AI for a hybrid workforce, 54% report that employees feel left behind by AI. Forty-six percent report disconnected AI projects, compared with 32% across sectors.

At the same time, technology companies show high confidence in AI. Eighty-three percent say employees trust AI outputs to inform strategic or operational decision-making, compared with 62% across sectors.

“Across ASPAC, the real story is not adoption — it is orchestration,” Simon Dubois, ASPAC Head of Technology & Media at KPMG in Australia, said. “AI is everywhere, but scaling it in a way that is consistent, governed and enterprise-wide remains a complex challenge.”

The report flags cybersecurity as a growing concern as AI becomes embedded in critical systems. Thirty-six percent of technology leaders identify cyberattacks among their top AI concerns today; that rises to 41% over the next 24 months. Regulatory and compliance risk rises from 30% today to 31%.

For technology companies, the next phase of AI investment will depend less on proving that the technology can be deployed and more on proving that it can generate repeatable business value. KPMG recommends modernising legacy technology stacks, strengthening data foundations, redesigning talent strategies, and building governance into AI deployment.