Data is destiny: the AI winners (and losers) are decided by data strategy.

Technology alone won’t fix an organization’s data problems – people and processes are the linchpin. At this week’s Gartner Data & Analytics Summit, analysts drove home that an often-overlooked obstacle to AI success is cultural resistance ([1]). According to Gartner, by 2027, 60% of organizations that fail to address the human side of data governance will be unable to achieve effective AI governance at all ([2]). In practice, many enterprises have poured millions into data warehouses and analytics platforms, yet struggle because employees don’t trust data or know how to use it.

The focus on culture means fostering data literacy, accountability, and cross-functional ownership of data quality. Gartner’s analysts noted that many companies remain fixated on technical policies and compliance checklists, while neglecting to engage the workforce around why data governance matters ([3]) ([4]). This gap between IT and the business can derail even well-funded AI initiatives. For instance, if front-line employees bypass cumbersome data entry processes or ignore new analytics tools, the fanciest algorithms won’t have solid ground to stand on.

Leading organizations are therefore broadening their AI readiness agendas to include “AI-ready stakeholders,” not just AI-ready data ([5]). They are tying data governance goals to tangible business outcomes that employees care about, and embedding data quality responsibilities into everyday workflows. The takeaway: building a culture that values data is now as critical as the technology stack. Companies that cultivate trust in data, clarity in ownership, and shared responsibility are pulling ahead, while those that don’t are finding their AI ambitions stymied by human roadblocks.

Even as organizations double down on AI, many are discovering that technology investments alone aren’t translating into returns. A new Accenture survey of over 3,000 C-level leaders finds 82% are increasing spending on AI, yet only 23% report achieving wide-ranging business value from these investments ([1]). In fact, the percentage of firms seeing significant AI-driven value has dropped from earlier in the year. This “AI value gap” is prompting hard questions in the boardroom: If models are more powerful than ever, why aren’t they delivering the expected ROI?

One root cause is coming into focus – an absence of mapped and optimized business processes and data flows for AI to learn from. Much of an organization’s real-world knowledge lives in undocumented workflows, workarounds, and institutional know-how that never makes it into any database ([2]). When AI systems don’t have access to this tacit knowledge – the exceptions, the hand-offs, the human decisions that occur outside structured systems – they struggle to produce relevant and reliable outcomes. It’s like asking a new hire to optimize a process without ever giving them the training manual: errors and misfires abound.

In response, new tools and partnerships are emerging to illuminate these “shadow” processes and feed the findings back into AI models. One example is Accenture’s investment this week in Within, a platform that records how employees actually get work done across apps and offline steps ([3]). Within automatically compiles this operational data into a continuously updated “Work Brain” – a rich context layer that AI agents can tap into for decision-making ([4]). By capturing the nuance of real business operations, such solutions aim to close the AI ROI gap. Leaders who invest in understanding and digitizing their core processes can give their AI initiatives the missing context they need to truly scale impact.

With each passing week, it becomes more evident that the only competitive edge in the AI era comes from proprietary data assets. A new analysis published today underscores that as foundation-model AI becomes ubiquitous and cheap to obtain, companies will differentiate themselves through the unique data they control ([1]). In one study, the top 25% of AI-performing enterprises were distinguished not by access to better algorithms – they often used the same off-the-shelf models as everyone else – but by having 3.4 times more proprietary, domain-specific data feeding those models than the median company ([2]). In contrast, the laggards largely relied on generic public data and saw little competitive lift.

Industry leaders from cloud and AI providers have been echoing this point in recent months: models are becoming commodities, available via API to anyone at low cost, while the real long-term value lies in the “data gravity” of an organization ([3]) ([4]). That gravity comes from years’ or decades’ worth of customer interactions, transactions, and operational insights that no rival can easily replicate. As one AI executive put it, an AI strategy focused solely on picking the right model is misguided – the pressing question is, what proprietary knowledge do you have that nobody else does? ([5]) In practical terms, this means enterprises should aggressively secure, integrate, and leverage their unique datasets – from customer behavior and supply chain patterns to domain-specific sensor data – as critical intellectual property.

Forward-thinking firms are already reorganizing around data as a strategic asset. They are breaking down data silos via modern architectures like data lakehouses and data fabrics, ensuring that information flows freely yet securely to where AI applications need it. By treating data governance and architecture as board-level priorities, companies not only mitigate risks but also create a virtuous cycle: more high-quality data leads to better AI models, which yield better services and products, attracting even more data. In the end, this compounding data advantage becomes a self-reinforcing moat around the business. The race to innovate in AI is quickly becoming a race to amass and activate proprietary data.

Even as businesses scramble to harness data for AI, governments are reevaluating how to balance innovation with individual privacy. This week in Europe, a controversial proposal to amend data protection rules for AI has ignited a fierce debate ([1]) ([2]). A leaked European Commission draft would create new "legitimate interest" exceptions in the GDPR, effectively allowing companies to use vast troves of personal data collected over decades to train AI systems – without seeking users’ consent ([3]) ([4]). Policymakers argue that loosening the rules could boost competitiveness in the AI race ([5]), but privacy advocates, led by campaigner Max Schrems, warn this amounts to sacrificing fundamental rights for profit. Schrems’ organization noyb (None of Your Business) slammed the plan as a “digital expropriation of Europeans,” accusing EU leaders of caving to Big Tech’s demands at the expense of privacy ([6]).

The outcome of this policy battle remains uncertain – it may face challenges in the EU courts if enacted ([7]). However, the implications for data strategy are already clear. Companies operating in or with Europe need to stay agile in their data governance, as the regulatory ground beneath AI is shifting. On one hand, more permissive data usage laws could unlock new AI training datasets and encourage data sharing across borders. On the other, any misstep on privacy could result in reputational damage or legal penalties. Responsible AI leadership today requires a dual focus: pushing for data-driven innovation while also doubling down on ethics, compliance, and transparent data management. Striking that balance will be crucial for sustaining public trust and unlocking AI’s full potential in a world increasingly concerned with data rights.