Proof over Promise Insights on Real World AI Adoption from 2025 MINDS Organizations 2026

MINDS organizations recognize that data is a fundamental driver of AI impacts. Yet, “data quality” remains the most oft-cited barrier to achieving such impact. To address it, organizations are actively building differentiated data advantages – centralizing structured and unstructured data, and augmenting it with synthetic, real-time and physics-based data to unlock new frontiers of innovation. This shift signals that scaling AI depends on mastering data at greater depth and precision than before. Digital natives and industry incumbents have different data advantages when scaling AI Whether through modern design or legacy data transformed into intelligence, or through strategic alliances, MINDS organizations are finding distinct paths to scale AI effectively. –Digital natives: Many digital-native organizations had early structural advantages, including integrated data ecosystems allowing high-velocity expansion. They have converted this foundation into intelligence platforms that continuously generate insights and value. –Industry incumbents: Established industry players are converting their own unique assets into differentiators – particularly their deep reservoirs of proprietary and unstructured data accumulated over decades (see Spotlight 5). Several incumbent MINDS organizations are also extracting new value from proprietary and legacy data through code and workflow assistants, a proven use case for large language models (LLMs). When such assistants are grounded in an organization’s own archives of knowledge, they become significantly more effective, with specific responses aligned with internal standards. 2.3 Insight 3: Strengthening data foundations and augmented data sources to advance impact and scale Enabling an agile supply chain with AI BOX 8 Black Lake Technologies exemplified this approach. Building on a robust data foundation and a unified technology architecture, Black Lake combines structured, unstructured and synthetic data through a single platform to support flexible supply-chain coordination. By using AI to automate design, modelling and part decomposition that was traditionally manual, creators can bring new products to market faster, while SME factories with surplus capacity can take on more high-quality orders. This enables more flexible production: cycle times drop from 6–12 months to 1–3 months, minimum order sizes shrink from 10,000 units to 100, and capacity use rises from 65% to 83%, demonstrating how a digital native translates data and process innovation into measurable impact with AI. SPOTLIGHT 5 CATL’s transformation of terabytes of data into an intelligent battery cell design system CATL (Contemporary Amperex Technology) turned a deep reservoir of proprietary, multimodal data into a sustained AI advantage. The company developed an AI-powered battery cell design platform that fuses physics-informed ML with transformer models and a unified pipeline for time series, text, image and graph data. Harnessing 600 terabytes (TB) of historical testing across hundreds of thousands of design cases, the platform automates data collection, cleaning, feature engineering and model training while enabling reuse of data, models and compute across products and plants. It replaces manual trial-and-error design cycles (which often last weeks) with AI-driven recommendations generated in seconds and parameter optimization completed in minutes. Impact: CATL compressed design from two weeks to minutes, cutting prototype cycles from 24 to 13 months and raising the accuracy of new designs from 70% to 95%. Savings of $140.6 million per year were achieved. Scaling AI depends on mastering data at greater depth and precision than before. Proof over Promise: Insights on Real-World AI Adoption from 2025 MINDS Organizations 17