Couchbase, a developer data platform for AI applications, has released its eighth survey of global IT leaders. The study found that businesses unable to use AI effectively could lose an average of 8.6% of their monthly revenue, resulting in an annual loss of almost $87 million per company. 21% of enterprises admit to having “zero” or “insufficient” control over AI use, allowing employees too much or too limited access to tools, increasing risk. 64% are concerned that they are not taking advantage of AI as quickly as possible due to “decision paralysis.” Investment in AI technologies, including GenAI and agentic AI, is expected to surge by 51% from 2025 to 2026, accounting for more than half of all digital modernization spend.

“The evolution from GenAI to agentic AI is creating vast opportunities for enterprises that can harness these technologies effectively,” said Julie Irish, Chief Information Officer at Couchbase. “Creating and operating innovative AI applications at scale is essential for successful enterprises. The right data strategy, including methods to ensure high data quality, scalability, and accessibility, is more important than ever to ensure companies unlock the value of AI.”

Key findings include:

  • Falling behind the AI wave has significant consequences: 99% of enterprises have encountered issues that disrupted AI projects or prevented them outright, including problems accessing or managing the required data; perception that the risk of failure had become too high; and an inability to stay on budget. These issues had real consequences, eating up 17% of AI investment and setting strategic goals back by six months on average.
  • Closing the data understanding gap is key to control: 70% of enterprises admit their understanding of the data (e.g., the quality and real-time accessibility of data) needed to power AI is “incomplete,” contributing to 62% not fully understanding where they are at risk from AI (e.g., through security or data management issues). Conversely, those with greater understanding are more confident and are 33% more likely to be prepared for agentic AI.
  • Data architecture is evolving and requires consolidation: The exemplary data architecture is crucial for AI. Yet enterprises say their current architecture has an average lifespan of 18 months before it can no longer support in-house AI applications. 75% of enterprises have a multi-database architecture, which makes it more challenging to ensure accurate, consistent AI output; 61% do not have the tools to prevent proprietary data from being shared externally, which increases security and compliance risks; and 84% cannot store, manage, and index high-dimensional vector data needed for efficient AI use. To address these challenges, all surveyed enterprises are consolidating and simplifying their AI technology stacks to make controlling AI easier and more efficient.
  • Encouraging experimentation contributes to AI success: Corporate attitudes about AI notably impact its success. Enterprises encouraging AI experimentation have 10% more AI projects enter production and incur 13% less wasted AI spend than enterprises with a more restrictive approach.
  • New developments in AI are rapidly reaching parity: The proportion of AI spend on agentic AI (30% of total), GenAI (35%), and other forms of AI (35%) is almost even, despite agentic AI and GenAI being much newer concepts. This suggests enterprises are investing heavily in keeping up with AI development, as 66% worry that AI and different approaches to AI are evolving faster than their organizations can keep pace.
  • Inability to keep up with AI increases risk of being replaced: Enterprises recognize AI’s potential for disruption, allowing smaller organizations with a better grasp of the technology to replace larger, less agile competitors. More than half (59%) of IT leaders are concerned that their organizations risk being replaced by smaller competitors, yet at the same time, 79% believe they can do the same and displace their larger competition.

“The data reveals both tremendous opportunities and significant risks presented by AI,” continued Irish. “While 73% of CIOs are excited about AI’s potential and feel compelled to use it more, the enterprises that master their data will be the ones that truly capitalize. The key is having robust controls and an architecture that suits enterprises’ purposes. When enterprises build the right foundation to support critical applications containing AI workflows, and target use cases with a clear ROI, CIOs will be best positioned to turn AI into a genuine competitive advantage.”

“A modern developer data platform is essential for enterprise AI success,” added Matt McDonough, SVP of product at Couchbase. “With capabilities like vector search, integrated AI Services, and support for agentic AI development, Couchbase empowers customers to develop agentic systems and applications at scale, while delivering compelling price-performance. By supporting the management of all data types involved in AI interactions, our platform helps enterprises unify AI, operational, analytical, vector, and mobile workloads into a single, multipurpose architecture. This holistic approach enhances data visibility, control, and protection and gives developers the tools they need to innovate with the next wave of AI technologies.”

Couchbase commissioned an online survey, conducted in April 2025 by Coleman Parkes (https://colemanparkes.com/), an independent market research organization. Eight hundred senior IT decision-makers, such as CIOs, CDOs, and CTOs, in organizations with 1,000 employees or more in the U.S., U.K., France, Germany, Turkey, Japan, India, Australia, and Singapore, were surveyed.