Match Your AI Strategy to Your Organization’s Reality

Idea in Brief The Problem Many companies invest heavily in AI, but their initiatives aren’t delivering tangible business value. The Reason Technology itself isn’t the core obstacle. Often there’s a misalignment between an organization’s ambitious innovation goals and its underlying value chains, operating models, and technology stacks. The Solution Use the framework in this article to determine where your company falls along two key dimensions: value-chain control and technological breadth. Then choose the right strategy to realize AI’s potential: focused differentiation, vertical integration, collaborative ecosystem, or platform leadership. In 2018 two global giants set out to harness artificial intelligence to reshape the way they designed and brought products to market. General Motors applied generative-design software using Autodesk’s Fusion 360 to reimagine a humble but critical component—the seat bracket. The AI generated a structure that resembled something nature might have grown—an airy, latticelike form that was 40% lighter and 20% stronger than the original. Yet the part never made it into production. Why? GM’s supply chain and manufacturing system—built for stamped steel—couldn’t handle the complex geometry of the AI-generated design. Retooling the system would have taken years. The innovation stalled. At the same time, Apple began experimenting with metalenses—ultra-thin, AI-optimized optical components capable of replacing traditional camera lenses. The technology required integrating machine learning, materials science, and semiconductor manufacturing. Within two years, Apple had filed dozens of patents and (at the time of this writing) is reportedly preparing to embed the breakthrough in its Face ID sensors—first in the iPad Pro and then in upcoming iPhone 17 models as well. Unlike GM, Apple didn’t just have a bold idea—it had the system to execute it. The two stories capture an important truth about AI: The problem isn’t usually with what AI can and can’t do. More often, it’s the misalignment between what leaders want to achieve and what their value chains, operating models, and technology stacks can realistically support. The issue is pervasive—62% of companies cite poor cross-functional fit and 63% flag the need to adjust workflows as leading barriers to successful AI adoption. Only 25% of CEOs say they feel fully prepared to deploy AI organization-wide, according to a survey by Kearney and the Futurum Group. Unfortunately, the vast majority of companies are more like GM in this regard than like Apple. Studies suggest that many AI initiatives fail to deliver tangible business value: According to S&P Global Market Intelligence, 42% of companies abandoned the majority of their AI initiatives in 2025 (up from 17% in 2024), and on average, 46% of proof-of-concepts were scrapped before reaching production. Moreover, a recent survey of 1,600 enterprise leaders and employees by the AI firm Writer found that only one-third of organizations achieve significant ROI from their AI investments, even though 73% of them spend more than $1 million on AI annually. In this article we offer a practical framework to help companies improve their return on AI investment. Drawing on our research and experience with firms across sectors—including consumer goods and advanced manufacturing—we identify the two key dimensions that shape AI success: value-chain control and technological breadth. These forces define the landscape of possibility—and point to four distinct approaches companies can adopt to realize AI’s potential: focused differentiation, vertical integration, collaborative ecosystem, and platform leadership. Each comes with its own risks, requirements, and potential for breakthrough innovations. But when the approach fits a firm’s reality, the payoff is substantial. Let’s begin by taking a closer look at the two dimensions that form the foundation of this framework. Find Yourself on the Framework The first dimension, value-chain control, refers to the degree of influence a company has over the journey from idea to market. Firms with high value-chain control can test, iterate, and scale up innovations quickly because they own or strongly influence product design, manufacturing, distribution, and customer engagement. Samsung, for example, can roll out AI-powered display or camera improvements across its entire product portfolio because it controls everything from chip fabrication to global retail outlets. At the other end of the spectrum are companies with low value-chain control (think of tier-two suppliers in the automotive sector, for example, or brand licensors in consumer goods). Because these organizations must rely on others to validate, adopt, or distribute their innovations, it’s hard for them to move quickly. They have little influence on how other companies operate and a relatively narrow space to innovate within. The second dimension, technological breadth, refers to the range and interdependence of the technologies a company must integrate to compete. In high-breadth sectors—such as semiconductors, autonomous vehicles, and life sciences—AI rarely works in isolation. It must be woven into a fast-moving web of other technologies, such as sensors, robotics, materials science, cloud architecture, and edge computing. Organizations in high-breadth sectors must manage constant convergence. By contrast, low-breadth industries—such as food processing, building materials, and basic logistics—tend to operate with more-stable technology stacks. Companies in these sectors can use AI to drive big gains, but often they do so by refining existing processes rather than redefining the landscape. These dimensions aren’t fixed points on a map—they’re dynamic forces that evolve across functions, geographies, and time. A company may have high technological breadth in R&D but low breadth in customer engagement. It may exert strong value-chain control in one region but depend heavily on intermediaries in another. Companies should focus on those places where they’re best positioned to act with focus and confidence. Pick the Right Strategy Four AI innovation strategies emerge from our framework. Each one represents a different way to align an organization’s position with a viable path to innovation. Now let’s take a closer look at each strategy—its logic, the specific kind of innovation it tends to unleash, and, of course, its risks. 1. Focused differentiation: Sharpen your edge. Companies with limited value-chain control and low technological breadth typically operate in a mature industry. They have deep expertise in one part of the value chain—product formulation, for example, or upstream sourcing or consumer insight—but they don’t control the full path to market. They’re not chasing the frontier of AI, but they have data, and they know how to use it. They can’t redesign the system, but they can make it smarter. They use AI to fine-tune and optimize products or processes within a defined domain. This is the realm of precise, high-impact use cases: a better label, a smarter sensor, a more adaptive formula. Companies here win with AI by going deep, not broad. See more HBR charts in Data & Visuals For example, the food and beverage giant PepsiCo applied AI upstream in its potato supply chain, where it has tight control through grower programs. By deploying drones and using machine learning to evaluate early indicators of crop health, PepsiCo helped farmers optimize irrigation and fertilizer use, leading to reduced carbon footprints, higher yields, and more-resilient harvests—a critical edge in a commodity business. PepsiCo Europe has partnered with Yara, a global crop nutrition company, to equip farmers with precision-farming digital tools in order to drive the adoption of regenerative agricultural practices. In a similar vein, McCormick & Company narrowed its focus to a specific point of leverage: flavor development. In 2019 the global spice manufacturer partnered with IBM to build SAGE, an AI system trained on decades of sensory data, recipes, and consumer insights. The tool has since become central to McCormick’s product development process, helping the company accelerate innovation and double the net sales contribution from new products between 2022 and 2024. At Fonterra, the New Zealand dairy giant, leaders didn’t try to digitize the whole business. They zeroed in on a choke point—predicting milk quality before it left the farm gate—and applied AI with surgical precision. Working with partners in agri-tech and analytics, Fonterra integrated sensor data from farms (such as weather, pasture conditions, and herd health) with historical milk collection records. Machine-learning models were then trained to detect patterns that signaled potential drops in milk quality—flagging issues like elevated somatic cell counts or bacterial contamination risks before the milk reached the processing plant. This allowed Fonterra to optimize milk collection routes in real time, prioritize high-quality sources, and reduce waste. The chief risk for companies in this quadrant is an excess of ambition. In late 2021 Zillow’s home-flipping initiative, Zillow Offers, which relied on its AI-derived “Zestimate” pricing model, spectacularly failed to scale. Zillow bought 27,000 homes based on the AI’s valuation, but sold only 17,000. The Zestimate prices were off by as much as 6.9% for off-market listings. The result: a staggering $304 million inventory write-down, 2,000 layoffs, and cancellation of the entire Zillow Offers business. 2. Vertical integration: Wire the machine. Companies with strong value-chain control but relatively limited technological breadth are often ideal candidates for enterprisewide AI adoption. They may not need to track every frontier tech trend, but by embedding AI into the processes they already own, they can generate a large impact. AI can link insights across internal systems—revealing synergies and efficiencies between data, departments, or processes. This is the space where AI enhances operational excellence: predictive maintenance, dynamic pricing, and demand-driven logistics, to name a few. Here, scale acts as a multiplier of impact—the larger the operations, the greater the cumulative gains from even small efficiency improvements. JD.com, the Chinese e-commerce giant, embedded AI across its logistics network, using real-time data to optimize everything from warehouse inventory and delivery routing to labor scheduling and demand forecasting. During the pandemic lockdowns, JD.com’s intelligent system rerouted deliveries based on containment zones, automated warehouse operations with AI-driven robotics, and dynamically reassigned inventory to match regional surges in demand. While competitors struggled with bottlenecks and delays, JD.com maintained uninterrupted service because it had made a smarter system. In the energy sector, ExxonMobil used AI to interpret seismic data and optimize drilling paths in Guyana. Algorithms trained on terabytes of historical data cut average well-drilling time by 15%, saving millions per site. Because Exxon owns the infrastructure—from exploration to production—it could move fast, without waiting to receive external validation. And consider Walmart. With control over the supply chain, store operations, and pricing systems, Walmart used AI to optimize logistics, inventory, and labor deployment. In Florida, for example, the company ran a pilot program that used local weather forecasts and social media trends to anticipate demand spikes. Ahead of Hurricane Ian, the system automatically reallocated emergency supplies. After the storm, the system rerouted shipments to avoid a damaged distribution center and kept stores stocked with essentials. Ann Cutting draws on her background in photography, technology, and science to create AI artwork that explores how we work, collaborate, and create. Ann Cutting Companies pursuing vertical integration often excel at unlocking novel connections—linking data and processes across silos to uncover patterns and efficiencies that others can’t see. When AI is deployed across a tightly controlled system, the ability to connect the dots—between logistics and pricing, or operations and forecasting—becomes a unique source of competitive advantage. It’s important, once again, not to overreach. Even firms with strong operations can stumble when ambition outpaces execution. GE, for example, sought to become the Microsoft of industrial AI with its Predix platform, which it designed to connect analytics across machines—from turbines to locomotives to medical devices. But siloed data, internal resistance, a lack of connecting software, and shifting leadership crippled the rollout. After spending more than $4 billion, GE scaled back its platform ambitions and spun off much of GE Digital. 3. Collaborative ecosystem: Work the network. Some companies operate in technologically complex ecosystems but lack control over how their solutions ultimately reach the market. In this quadrant AI success comes not from going it alone but from partnering strategically—sharing innovation risk, infrastructure, and expertise. Companies here often work in high-tech, fast-moving industries, but the levers of execution aren’t fully in their hands: They depend on regulators, researchers, or platform partners to reach the end user. Their advantage lies in partnership: shared platforms, codeveloped tools, and alliances that align incentives, not just timelines. Take Novartis and Microsoft, for example. They didn’t just form a partnership—they created an AI innovation lab aimed at accelerating drug discovery and development. At the heart of the effort was a suite of machine-learning models trained to predict molecular behavior, identify potential therapeutic targets, and optimize clinical trial protocols. In one case, the lab’s tools helped Novartis identify new biomarker combinations for oncology trials—cutting trial design time by more than 30%. The success of the endeavor didn’t come from shared technology alone but from shared purpose, codeveloped infrastructure, and a commitment to scientific and operational integration. BMW Group’s alliance with Intel and Mobileye is another compelling example. Each player contributed distinct capabilities—processing power, computer vision, vehicle integration—to develop autonomous driving solutions. Their jointly developed prototype was tested on a German autobahn, where it demonstrated safe, hands-free driving in controlled settings. None could have achieved it alone. And consider Pfizer’s collaboration with BioNTech during the Covid-19 pandemic. BioNTech’s AI models screened more than 10,000 mRNA candidates in days, selecting the BNT162b2 formulation that would become the vaccine. Pfizer’s global regulatory and manufacturing capabilities—augmented by AI tools—accelerated production and approval. Together, the companies turned an audacious bet into a historic success. As that example shows, this quadrant is best positioned to unlock fundamental discovery—the kind of early-stage breakthrough that reshapes science, technology, or medicine. But discovery requires coordination. When governance is unclear, cultures clash, or goals diverge, even the most promising tools can falter. When alignment runs deep—when knowledge, incentives, and execution are codesigned—the payoff is transformative. AI in this quadrant isn’t just about collaboration. It’s about coherence and how the partners work together. For AI partnerships to unlock powerful synergies, alignment must go beyond the press release. IBM’s high-profile collaboration with the cancer center MD Anderson aimed to revolutionize cancer care with its Watson-powered Oncology Expert Advisor. Despite promising technical results, the project ran into organizational and integration challenges. Implementation proved difficult, and the system never moved beyond the pilot phase. The effort ultimately stalled—not for lack of ambition but because the partnership struggled to align technology, data, and clinical practice. 4. Platform leadership: Shape the norms. At the apex of both dimensions—high technological breadth and broad value-chain control—are companies that don’t just adapt to change; they shape it. They create infrastructure and ecosystems as well as build products. This is the domain of orchestration: setting standards, opening APIs, and designing systems that others want to build on. Take Bloomberg’s launch of BloombergGPT, a finance-specific large language model trained on more than 700 billion tokens. It wasn’t just a technical upgrade—it was a strategic move to define the next generation of financial AI. Unlike general-purpose models, BloombergGPT was trained on a unique blend of financial documents, earnings calls, and proprietary datasets, enabling it to summarize earnings reports, automate news classification, and assist with risk modeling—all within Bloomberg’s tightly integrated terminal ecosystem. Bloomberg’s offering wasn’t just about generative AI. It was about setting a new industry standard. Siemens Healthineers has achieved a similar platform-leadership position in medical imaging. Its AI-Rad Companion suite integrates directly with hospital systems, automatically analyzing X-rays, CT scans, and MRIs to highlight anomalies. Trained on more than 400 million scans, the system delivers FDA-cleared results in more than 60 countries—improving diagnostic accuracy while streamlining clinical workflows. Siemens doesn’t just sell AI tools. It shapes how hospitals use them and sets new standards for care. Microsoft’s platform approach to AI combines infrastructure, tooling, and ecosystem orchestration. GitHub Copilot, trained on billions of lines of code, now contributes up to 40% of code written in supported languages. In Microsoft 365, embedded AI copilots in Word, Excel, and Teams are redefining how work gets done. And with Azure OpenAI Service, Microsoft has become the enterprise backbone for generative AI. Crucially, it has built that leadership not just through delivering technical performance but by its governance, transparency, and alignment with enterprise customers. Companies in this quadrant are well positioned to move into seemingly distant domains through their ability to detect weak signals and novel patterns across geographies, industries, and domains. Platform leaders have the scale, the data, and the architectural reach to see what others can’t—and act on it before anyone else. But with that reach comes risk. The ability to shape industries brings a heightened responsibility to earn and maintain trust—across partners, regulators, and end users. Failure in this quadrant rarely stems from weak technology. Consider Google’s foray into healthcare AI through its DeepMind Health. The team partnered with UK hospitals to develop diagnostic models trained on real-world patient data. Technologically, the project was promising. But when it emerged that DeepMind had accessed millions of NHS records without proper consent, public backlash erupted. The initiative was then absorbed into Google Health and lost momentum. The failure wasn’t the algorithm; it was the breach of trust. Engage Your People Across all four strategies, the defining challenge isn’t technical. It’s human. Many AI initiatives falter not because the underlying algorithms are weak but because people resist using the tools. In the Writer enterprise survey referenced earlier, 31% of employees admitted to actively pushing back on their company’s AI initiatives—often because they feared being replaced. One in 10 went even further, saying they had tampered with performance metrics or intentionally generated low-quality outputs to undermine adoption efforts. When Rent a Mac, an Apple device rental company, launched an AI-driven inventory management system, it triggered anxiety across its workforce. That led to a seven-week delay in implementation and a loss of about $85,000 in expected efficiency savings. But by appointing AI champions to demonstrate real use cases, the company saw engagement levels triple—from 31% to 89% in just a few months. Transparency, dialogue, and personal experiences changed the narrative. Colgate-Palmolive recognized the importance of employee engagement when it launched its internal AI Hub that empowered employees to develop their own assistants—thousands of them—without coding experience. The result wasn’t just better workflows. It was buy-in. When people feel like participants in the future, they don’t fear it—they help build it. The role of the manager is shifting in AI-powered organizations. Beyond coordinating people, managers now must help teams learn to collaborate with algorithms—interpreting machine insights, redesigning workflows, and translating technical progress into human progress. Doing so often requires a cultural shift: creating space to experiment, to fail fast, and to learn in real time. The most successful organizations treat AI not as an answer but as a question: How can we work smarter, together? And ultimately that means that some of them are present in all four quadrants. Let’s see what that looks like. Leading in the Age of AI At first glance, P&G might seem like a textbook example of vertical integration. (P&G is a client of Iprova, where two of us—Christopher and Julian—work.) With end-to-end control over everything from R&D labs to manufacturing plants to retail execution, it has embedded AI deep into its operational core. Using Microsoft’s Azure IoT Operations, the company cut the time required to deploy new machine-learning models to its factories by up to 90%, enabling predictive algorithms to monitor machine vibration and temperature data, anticipate equipment failures, and continuously optimize production. Beyond operations, P&G applies AI with precision where it matters. The Oral-B iO toothbrush uses sensors and real-time machine learning to coach users on brushing technique, improving oral hygiene outcomes. The team behind Tide used AI to accelerate formulation testing, compressing the time from months to weeks. These are classic focused differentiation plays—narrow in scope but deep in impact. Outside its walls, P&G actively cultivates collaborative ecosystems through initiatives like its long-running Connect + Develop platform. This open-innovation network feeds into AI-powered tools such as Olay Skin Advisor, which analyzes selfies to recommend skincare regimens. P&G has also built a proprietary consumer-pulse system that integrates signals from social media, customer service, product reviews, and retail data—giving it the power not just to respond to market trends but to shape them. What makes P&G distinctive is that it intentionally plays across all four quadrants: It focuses where AI can deliver immediate value; it integrates where scale drives performance; it partners where complementary capabilities matter; it creates platforms that place it at the heart of the key ecosystems in which it operates. And in doing these, it reveals a deeper truth: Strategy may begin in a quadrant, but success is built through a system. . . . The next decade won’t be won by the companies that launched the most pilots. It will be won by the companies that know how to scale. That means translating ambition into action—choosing the strategy that fits your organizational reality, empowering your people, and aligning AI with what you can truly control. Because in the end, AI is not the strategy. It’s a tool that brings strategy to life. So ask the hard questions: Where do we have leverage? Where can we move fast? What kind of innovation are we built to unlock? AI may generate insights, but only your organization can turn them into outcomes. The difference between GM and Apple wasn’t just the technology. It was the system that surrounded it. The companies that build the right systems—the ones where ambition meets execution—will be the ones that stop piloting and start leading.

Match Your AI Strategy to Your Organization’s Reality

Summary.

Partner Center