cb.web.local

On-premises AI infrastructure surges amid sovereignty concerns

New global research released by Schneider Electric on 20 November 2026 reveals that 55% of enterprises are choosing company-owned, on-premises infrastructure for at least part of their AI deployments to address sovereignty, security, privacy and compliance concerns.

Artificial intelligence is rapidly outgrowing the experimental sandbox, and businesses are rethinking where their workloads should live. ]

New global research released by Schneider Electric on 20 November 2026 reveals that 55% of enterprises are choosing company-owned, on-premises infrastructure for at least part of their AI deployments to address sovereignty, security, privacy and compliance concerns. Conducted by Omdia on behalf of Schneider Electric, the survey polled enterprise leaders across Europe, North America and APAC, spanning healthcare, financial services, life sciences, government, education and manufacturing. Respondents included IT professionals, product engineering and research and development teams, alongside executive leadership and general management.

Surging budgets and measurable returns on AI investment

Across industries, AI pilot programmes are multiplying, backed by rising budgets and tangible returns. Rather than relying heavily on costly frontier AI models, the most technologically advanced systems available, many enterprises prefer to develop and manage their own IT and build optimised models, particularly for smaller tasks where costs can otherwise escalate unnecessarily.

Key findings on enterprise AI momentum include:

  • Primary adoption drivers: Increased productivity tops the list at 35%, followed by cost reduction (25%) and time savings (22%).

  • Active deployments and pilots: While 23% of respondents already have AI models delivering value, 63% plan to deploy generative AI applications and 73% are running five or more proofs of concept.

  • Proven returns: Among companies with implemented projects, 60% report ROI meeting or exceeding expectations, including 29% realising a 2-5x return and 17% seeing a 5-10x return.

  • Expanding budgets: Over half of respondents had 2026 AI budgets above $500K, with 31% exceeding $1M. Looking ahead, 78% expect budgets to rise next year (nearly 30% predict an 11-25% year-over-year increase), while 16% expect unchanged spending and only 6% foresee a decrease.

“As AI moves from experimentation to real-world application, enterprises are focused on making these technologies deliver measurable business value at the right speed, accuracy and cost,” said Paul Tyrer, Global Vice President, AI Ecosystem, at Schneider Electric. “This research from Schneider Electric and Omdia highlights both the opportunities and the challenges organizations face as they formulate and implement AI strategies and supportive technologies. Understanding where organizations are making progress and where they need support is critical to helping them move forward with confidence.”

Infrastructure hurdles and the push towards on-premises AI

Turning promising pilots into full-scale production remains a steep climb. More than half of respondents report that at most one in five proofs of concept reach production, largely due to human and infrastructure bottlenecks:

  • Skills and integration gaps: 33% point to a lack of in-house expertise blocking full-scale deployment, while 32% struggle with architectural complexity and integrating AI into existing infrastructure.

  • Reasons for keeping AI in-house: Among the 55% favouring company-owned, on-premises environments, data privacy and security (30%), reliability requirements (22%) and regulatory compliance (21%) are the primary drivers.

  • Deployment roadblocks: When rolling out new infrastructure, scaling AI IT clusters (27%), managing costs (25%) and sourcing power and cooling infrastructure (22%) emerge as the top three hurdles.

Relying on partner ecosystems to bridge the expertise gap

To clear these operational hurdles, most businesses are looking outside their own walls. Nearly six in ten (59%) rely on external partners to shape their AI strategy, and 67% depend on outside support for implementation, working across cloud providers, infrastructure vendors, software platforms, systems integrators and specialist AI partners.

“As enterprises work to develop and manage their own AI environments, the lack of internal expertise remains a significant barrier and a leading reason proofs of concept fail,” said Maxine Holt, Vice President, Enterprise & Channel Research, at Omdia. “By leveraging existing partners and resources such as reference designs and validated architectures, organizations can navigate the complexity of AI infrastructure while aligning investment with deployment needs and accelerating time to adoption and business value.”

Supporting this transition, Schneider Electric is advancing AI-ready data centre modernisation through new power and liquid cooling technologies, reference designs developed with NVIDIA and AMD, and RapidScale prefabricated modular data centre solutions built to simplify deployment and reduce infrastructure risk alongside chip manufacturers, system integrators and server providers.