The gap between AI programmes that pay off and AI programmes that don’t isn’t the model.
According to new Gartner research, organisations that report successful AI initiatives invest up to four times more, as a percentage of revenue, in foundational areas such as data quality, governance, AI-ready people and change management, compared with organisations reporting poor AI outcomes.
That finding comes from a global survey of 353 data and analytics (D&A) and AI leaders that Gartner ran between November and December 2025. It lands alongside a second, more sobering number from the same research: only 39% of technology leaders are confident their organisation’s current AI investments will have a positive impact on financial performance.
Read together, these two figures describe most enterprise AI programmes today. Investment is happening.
Confidence in the return is not. Gartner’s research points to a specific, correctable reason why.
The mandate is data foundations, not more models
Rita Sallam, Distinguished VP Analyst, Gartner Fellow and Chief of Research at Gartner, frames the period through 2030 as a shift in what the D&A function is actually responsible for delivering. As she put it, “D&A leaders play a central role in achieving their organization’s AI value ambition.”
Gartner’s research sets out that mandate specifically: new trusted data, context foundations and what the firm calls perceptive intelligence, delivered through six structural shifts in how the D&A function organises, builds and creates value.
Two of those six shifts speak directly to the kind of engineering and delivery work that closes the gap Gartner has identified.
Context is now critical infrastructure, not a nice-to-have
Gartner’s research found that organisations with the highest maturity of AI-ready D&A capabilities are achieving up to 65% greater business outcomes, including revenue growth and cost optimisation, than lower-maturity peers.
The reason, according to Gartner, is that AI success by 2030 will not be decided by better models. It will be decided by whether agents have governed, contextual access to the right data.
That reframes what “context” means at an architectural level. Semantics and metadata, historically treated as data-management housekeeping, are now, in Gartner’s words, mission-critical infrastructure.
An AI agent without high-quality context and a trustworthy data layer beneath it cannot function autonomously, regardless of which model sits on top of it. Gartner’s recommendation to D&A leaders is direct: redesign the D&A architecture so the context layer becomes the central brain that AI agents draw on for trusted intelligence.
This is the argument for treating data and context engineering as a first-class discipline rather than a preparatory step that happens once and gets left alone.
It is also precisely the layer of work that determines whether the other five shifts in Gartner’s framework are achievable at all.
Siloed engineering practices will not scale AI ambition
Gartner is equally direct about what fails. Organisations trying to realise AI ambition at scale through siloed practices for data, AI, context and software engineering will not get there. The research calls for D&A organisations to move out of what Gartner describes as an endless loop of proof-of-concept cycles, and into interconnected data, AI, software and context engineering practices and skills operating at enterprise scale.
That distinction, between disconnected pilots and a genuinely integrated engineering practice, is where most enterprise AI programmes stall.
A proof of concept can succeed on a single dataset with a single team’s attention.
Enterprise scale requires the data pipeline, the context layer, the model integration and the software engineering around it to work as one connected system, governed consistently, rather than as four separate projects that happen to sit near each other on a roadmap.
Trust has become the precondition for value, not a compliance checkbox
Gartner’s fifth shift is about governance, and the supporting numbers explain why it now sits at the centre of the framework rather than at the edge of it.
A separate Gartner survey of 360 IT leaders, fielded in the second quarter of 2025, found that only 23% said they were very confident in their organisation’s ability to manage security and governance when deploying generative AI tools.
Sallam’s view is that traditional control models need to be rebuilt around trust rather than bolted on after deployment: dynamic governance that embeds automated checks for bias, privacy and compliance directly into workflows, rather than reviewing them after the fact. Her underlying point is unambiguous. Without trust in the data, the outputs and the decisions that AI models and agents produce, none of the investment upstream generates value.
That is the throughline across Gartner’s entire framework. Governance is not a constraint on the other five shifts. Gartner positions it as the mechanism that makes the return on the other five shifts realisable at all.
What the four-times figure actually rewards
It’s worth being precise about what Gartner’s headline statistic is measuring. Organisations reporting successful AI outcomes are not simply spending more overall. They are spending up to four times more, as a share of revenue, specifically on the foundational layer: data quality, governance, AI-ready people and change management.
Model selection and use-case identification, the parts of an AI programme that tend to dominate the conversation in a board meeting, sit downstream of that investment, not upstream of it.
This is consistent with Gartner’s broader six-shift framework.
Building toward an AI-first D&A function, redesigning teams for human-agent collaboration, establishing context as infrastructure, scaling connected engineering practices, embedding trust-based governance, and moving beyond a single ROI calculation toward compounding value all depend on the same foundation being in place first. None of the six shifts are achievable as a bolt-on to an AI initiative that has already been scoped, built and shipped without them.
Why this validates a foundation-first approach to AI delivery
Gartner’s data gives a name and a number to something we see consistently in enterprise AI engagements: the organisations getting a return are the ones that treat data quality, governance and context architecture as the starting point of the engineering effort, not a parallel workstream that catches up later.
That is the thinking behind Sprint Zero in our own delivery methodology, establishing the data, context and governance foundation deliberately before scaling AI capability, rather than retrofitting it once a dozen use cases are already live and the gaps have become expensive to unwind.
It is also why data engineering, context architecture and connected engineering practices sit at the centre of how we build AI platforms for clients. We achieve this through our DAPA (Digital AI Platform Accelerator) approach, rather than infrastructure work that happens quietly in the background.Â
Gartner’s research makes the commercial case for that sequencing explicit: the 61-percentage-point gap between organisations investing in foundations and organisations that aren’t is not a rounding error.
It is the difference between the 39% of leaders confident in their AI returns and the 61% who aren’t.
For CIOs and CDOs planning the next phase of an AI programme, Gartner’s research offers a fairly clear diagnostic question.
Not “which model should we use next,” but “is our data quality, governance, context layer and engineering practice built to the standard that the organisations getting a return on AI have already built theirs to.”
FAQ
What did Gartner find about AI investment and success? Gartner found that organisations reporting successful AI initiatives invest up to four times more, as a percentage of revenue, in foundational areas such as data quality, governance, AI-ready people and change management, compared with organisations reporting poor AI outcomes.
How confident are technology leaders in their AI investments? Only 39% of technology leaders are confident that their organisation’s current AI investments will have a positive impact on financial performance, according to Gartner’s research.
What are Gartner’s six shifts for data and analytics leaders through 2030? Gartner identifies six shifts: building toward AI-first data and analytics, redesigning the organisation for human-agent collaboration, establishing context as critical infrastructure, scaling connected engineering practices across data, AI, context and software, establishing trust as a catalyst for value, and moving beyond ROI to value compounding.
Why does Gartner say context is now critical infrastructure? Gartner finds that organisations with the highest maturity of AI-ready data and analytics capabilities achieve up to 65% greater business outcomes, including revenue growth and cost optimisation. Gartner’s research concludes that AI success depends on giving agents governed, contextual access to the right data, not on using better models alone.
How confident are IT leaders in managing AI governance? Only 23% of IT leaders said they were very confident in their organisation’s ability to manage security and governance when deploying generative AI tools, according to a separate Gartner survey of 360 IT leaders fielded in the second quarter of 2025.
What should organisations prioritise before scaling AI? Gartner’s research points to the same conclusion as our own Sprint Zero delivery approach: data quality, governance and a connected context and engineering foundation should be established before AI initiatives scale, rather than retrofitted afterward once gaps in data and governance have already become expensive to correct.
