Organizations across the globe are transitioning from the initial pilot and experimentation phase of their artificial intelligence (AI) initiatives to scaling them on a larger, enterprise-level scale.

According to a survey by Deloitte, 25% of respondents indicated that their organization has successfully moved 40% or more of their AI experiments into production. Another 54% expect to achieve this milestone within the next three to six months. These findings suggest that organizations globally have progressed significantly beyond the experimentation phase and are now actively scaling AI across their entire organization.

The survey, which polled more than 3,200 leaders globally in Q3 2025 and interviewed 15 executives and AI and data science leaders, revealed a growing trend in the deployment of AI at the enterprise level with accelerating implementation.

These results align with findings from a recent study released last week by early-stage venture fund Plug and Play. The research, which gathered input from 41 of the world’s largest enterprises, found that only 7% of respondents remain in early exploration, while 74% reported AI running in selected-function or scaled product environments.

Moving beyond genAI

The Deloitte study also unveiled more advanced applications of AI, highlighting the rapid evolution of AI beyond generative AI (genAI). Notably, physical AI is already embedded in operations, with 58% of respondents utilizing it to some extent, with projections indicating a substantial adoption rate of 80% within the next two years.

The emergence of frontier‑scale foundation models, more embedded AI capabilities, and significant improvements in edge computing, are reducing the historical constraints associated with physical automation. These technologies now empower robots and intelligent devices to operate more reliably.

The research revealed that while, manufacturing, logistics, and defense lead the global adoption of physical AI, markets in Asia Pacific (APAC) are at the forefront. A remarkable 71% of APAC respondents reported at least minimal use of physical AI, compared with 56% in both the Americas and Europe, the Middle East, and Africa (EMEA) region.

PwC’s Strategy& predicts that the global physical AI market will reach approximately EUR 430 billion (US$496 million) by 2030. The automotive sector is set to lead adoption, accounting for roughly EUR 171 billion (US$197 billion) of that amount and driven by the imminent technical breakthrough in autonomous driving.

Industrial automation and warehousing are expected to follow suit, representing about EUR 69 billion (US$80 billion). These sectors are poised for growth due to their structured layouts, external sensor networks, and repetitive workflows.

In Europe, Strategy& estimates that early physical AI adoption will represent about EUR 80-110 billion by 2030, driven primarily by automotive and industrial automation.

Beyond physical AI, the Deloitte study also revealed that enterprises globally are actively advancing their deployment of AI agents, establishing governance frameworks to manage the associated risks. Notably, 21% of the companies surveyed have developed mature models for the governance of autonomous agents.

However, the pace of AI agent deployment is outpacing the development of corresponding guardrails. In contrast to the proportion of organizations that have established governance models for AI agents, a significantly larger percentage, 74%, are planning to deploy agentic AI within the next two years.

Enterprises’ approaches to AI

The Deloitte study found that organizations are adopting different approaches to their adoption of AI. Notably, 34% of organizations are using AI to deeply transform their businesses, introduce new products and services, and reinvent core processes. These organizations are leveraging AI to unlock new value for both themselves and their clients, creating innovative digital solutions and revenue streams.

After these, 30% of organizations are designing key processes around AI but are keeping their business models unchanged. These organizations are taking a more cautious approach, being perhaps more apprehensive about the return on investment (ROI) and opportunities brought about by AI, as well as the associated risks.

More concerning, 37% of respondents reported using AI at a superficial level, with minimal or no change to existing processes. Rather than reimagining their businesses, these organizations are optimizing what already exists without true transformation.

To implement these approaches, organizations employ various strategies. The Plug and Play study found that the build-versus-buy debate has evolved beyond a simple dichotomy, with a significant majority of enterprises, 66%, adopting a hybrid approach. In comparison, 29% prefer a buy-first strategy and only 5% primarily build in-house.

An hybrid model allows companies to strategically distinguish between sources of competitive advantage and perceived commodities. This enables them to quickly bring solutions to production and focus internal teams on the pieces that actually matter strategically. Additionally, organizations can achieve more customization compared to the “build-only” approach, while also reducing the risk of becoming completely dependent on a single vendor by keeping certain capabilities in-house, unlike the “buy-only” approach.

Impact of AI

Across organizations worldwide, AI is already delivering gains in efficiency and productivity, cited by 66% of the leaders polled by Deloitte. 53% reported enhanced decision-making and data-driven insights, and 25% indicated that AI is already having a transformative effect on their companies, more than double from 12% a year ago.

However, benefits in areas like revenue growth, reduced operational costs, and enhanced customer relationships are taking longer to achieve. In particular, 74% of organizations hope to grow revenue through their AI initiatives in the future, compared to just 20% that are already achieving this today. Similarly, 65% of respondents hope to reduce costs by adopting AI, compared to 40% who are already experiencing cost reductions. Furthermore, 60% of respondents are eager for improved customer relationships in the future, while only 38% are already witnessing such improvements.

Barriers to scaling AI

Despite the rapid deployment of AI and promising early results, several obstacles are hindering the scaling of AI initiatives. According to a Deloitte study, insufficient worker skills are a significant barrier to integrating AI into existing workflows, alongside cost.

The Plug and Play research identified data foundations as the biggest constraint, cited by 71% of respondents. Issues like data quality, limited access, and fragmented pipelines are making it difficult to move from pilot to production, the study found.

Governance and legal friction come next at 53%, followed by security at 39%. Further down, 29% of organizations struggle to demonstrate clear ROI, and 26% face challenges integrating with legacy systems.

Featured image: Edited by Fintech News Switzerland, based on image by thanyakij-12 via Magnific