Survey: When AI Factories Fail, More Than Half of UK Enterprises Can’t Tell You Why

• Executives rate their organisation's ability to diagnose AI failures at nearly twice what their engineers report, leaving those accountable for AI investments without an accurate view of failures as they occur

• UK enterprises are farther along with AI deployment than their US counterparts, but further behind on the systems to govern it

PALO ALTO, Calif. & LONDON, September 02, 2026--New research from Virtana found that as UK enterprises run AI in production inside a demanding regulatory environment, more than half (53%) are operating AI infrastructure they cannot fully observe. The AI Factory Reality Check study, United Kingdom edition, found that UK executives are far more confident than their own engineers that AI failures can be diagnosed. This study of 238 UK enterprise decision-makers examines how AI factories operate and continues a research series Virtana began in the United States in May 2026. The comparison between the two markets shows that scaling AI faster than the systems to govern it carries compounding consequences in the UK, where the regulatory stakes are higher. Despite 59% of UK executives saying their organisation can automatically identify the root cause of a failure across systems, only 34% of the infrastructure engineers who field those alerts agree. For the people authorising that investment, and answering for it to their boards and increasingly to regulators, that divergence means less control than they assume.

"In the UK, operational observability and regulatory accountability have become inseparable concerns," said Paul Appleby, CEO of Virtana. "Enterprises are deploying AI into production under UK GDPR, emerging AI Act obligations, and sector-specific oversight across financial services, healthcare, and national infrastructure. The systems required to prove an AI factory is performing properly are the same systems that satisfy a regulator. For the boards responsible for AI oversight, sovereignty ultimately comes down to whether an organisation can see, attribute, and prove what its AI systems are doing across every environment it runs in."

UK Enterprises Lead US in AI Deployment, Trail in AI Workload Performance

The comparison between the US and UK markets reveals UK enterprises are scaling AI further and faster than their US counterparts, while reporting wider leadership disconnects, less predictable workload performance, and a steeper regulatory cost for every governance decision that gets deferred.

• 59% of UK enterprises are already scaling AI across teams versus 54% in the US

• Only 26% of UK enterprises describe AI workload performance as highly predictable, compared to 34% in the US

• The divergence between what executives and infrastructure engineers report about diagnostic capability is 25 points in the UK, compared to 17 points in the US

• In both markets, security and compliance reviews are being deprioritised as AI factory demands grow (50% of US enterprises, 39% in the UK)