Generative AI has moved from research labs into boardrooms with unusual speed. In less than a few years, it has become one of the most consequential technologies facing business leaders. That is not because it is merely another software tool. It is because generative AI has the hallmarks of a general-purpose technology: broad applicability, the potential to reshape multiple industries, and the ability to trigger waves of complementary innovation.
But executives should be careful not to confuse technological excitement with business value.
The breakthrough of large language models is an invention. The harder challenge is innovation: turning that invention into products, services, and business models that create value for customers and allow firms to capture a meaningful share of that value themselves. History shows that those two things are not the same. Many companies participate in technological breakthroughs. Far fewer convert them into durable advantage.
To make sense of the current moment, leaders need a practical way to think about how new technologies create value. One useful lens is what I call the three faces of technology’s value creation: emerging, enabling, and embedding. Each face highlights a different strategic challenge. Together, they explain why some firms benefit from technological shifts while others, despite strong technical capabilities, fall behind.
The Three Faces of Value Creation
1. The Emerging Face: Value Begins in Uncertainty
Generative AI is still in an emerging phase. Its capabilities are improving rapidly, its commercial applications are still being tested, and the boundaries of what customers will ultimately value remain unsettled. That makes this phase both attractive and dangerous.
In emerging technologies, the upside can be enormous, but so is the uncertainty. Performance improves unevenly. Costs can be high. Standards are unsettled. And no one yet knows which product architecture, interface, or delivery model will become dominant.
That uncertainty forces leaders into a familiar but difficult tradeoff: exploit what is available now or explore what may become possible later.
An exploitation approach emphasizes immediate commercialization. Companies move quickly to launch products, test demand, and capture early revenue. The advantage is speed. The risk is that firms optimize around today’s version of the technology and miss where the performance frontier is actually heading.
An exploration approach emphasizes learning. Companies invest in experimentation, partnerships, and capability-building while delaying major commitments to a fixed product form. The advantage is flexibility. The risk is that competitors move first and shape the market.
Executives should be careful not to confuse technological excitement with business value.
This tension is best understood through the technology S-curve. Early in the curve, progress can feel unpredictable. Improvements arrive, but not always in ways that translate cleanly into market demand. For generative AI, this means firms should resist simplistic bets. The winning move is rarely blind acceleration or passive waiting. It is disciplined experimentation paired with continuous learning.
In practical terms, that means leaders should treat this moment as a period of managed exploration. Pilot aggressively. Learn where the technology performs well and where it does not. Build partnerships that improve access to scarce capabilities. But avoid locking the organization too early into a single architecture, vendor, use case, or interface. The products people use today are unlikely to be the products that define the category five years from now.
2. The Enabling Face: Broad Reach Depends on Complementary Assets
Generative AI is not only emerging. It is also enabling.
An enabling technology is one whose core capabilities can be applied across many different domains over time. That is exactly what makes generative AI so powerful. The same underlying advances can support customer service, drug discovery, design, software development, marketing, semiconductor engineering, and much more.
This broad reach is the source of the technology’s promise. It is also the source of its complexity.
Enabling technologies do not create value through software alone. They require complementary assets: computing infrastructure, specialized hardware, high-quality data, workflow redesign, skilled talent, governance mechanisms, and often entirely new operating routines. Without those complements, the technology remains impressive but economically underpowered.
The history of enabling technologies makes this clear. A technology can have enormous theoretical applicability and still struggle commercially if the surrounding ecosystem is not ready. The problem is not the core invention. The problem is that the devices, channels, standards, or user habits needed to unlock value have not yet matured.
For large organizations, this creates an important strategic discipline. The right question is not, “Where can we use generative AI?” That question produces a long and expensive list. The better question is, “Where can we deploy generative AI in ways that share data, infrastructure, capabilities, or workflows across applications?”
The logic is simple. When each use case requires its own customized stack, implementation costs rise quickly and organizational friction follows. When multiple use cases can draw on common data assets, common model infrastructure, and common governance, the economics improve. Synergy matters.
That is why the smartest enterprise deployments will not be the noisiest ones. They will be the ones designed around shared complements.
The products people use today are unlikely to be the products that define the category five years from now.
3. The Embedding Face: Technology Creates Value Only When It Fits a Business Model and Ecosystem
Even a powerful enabling technology does not create value in isolation. It must be embedded in a business model and within an ecosystem of suppliers, partners, complementors, regulators, and users.
This is where many firms stumble.
Leaders cannot assume that once the technical capability exists, value will follow automatically. In practice, business value depends on whether a firm can place the technology inside a system that not only creates value, but also allows the firm to capture a meaningful share of it.
For generative AI, embedding raises two immediate challenges.
First, firms must persuade customers, partners, employees, and regulators that their AI-enabled offering is trustworthy, useful, and appropriate. This is especially important in settings where privacy, intellectual property, bias, explainability, and accountability are central concerns.
Second, once a successful model or approach becomes visible, rivals can imitate it. That means firms must think early about how to retain advantage. In practice, the open-versus-proprietary and big-versus-small model questions are not technical side debates; they are strategic choices about where value will commoditize and where competitive advantage will reside. Some firms will differentiate through proprietary data, workflow integration, customer relationships, or distribution. Others will benefit from shaping standards, orchestrating ecosystems, or controlling the application layer rather than the model itself.
This tension is especially important in ecosystems. In the early stages, firms often need partners to co-create value. But over time, those same partners may commoditize core components, capture a larger share of profits, or move into adjacent layers of the stack. Today’s complementor can become tomorrow’s competitor.
That means executives must balance two goals that are often in conflict: co-creating value in the near term while preserving their ability to capture value over the long term.
Why Strong Companies Still Lose During Technological Shifts
If this sounds abstract, history makes it concrete.
Nokia invested heavily in research and development and introduced important mobile innovations early. But technological sophistication in devices was not enough. The smartphone market was ultimately shaped by ecosystems, platforms, developer networks, and business models that extended beyond devices.
Kodak pioneered digital imaging technology yet struggled to thrive in the digital era. The issue was not a lack of awareness or innovation. It was the difficulty of adapting to a new ecosystem and business model architecture that could match or exceed its century-long success.
There is no standalone generative AI strategy. There is only business strategy, strengthened by generative AI.
By contrast, companies such as Amazon, Google, Meta, and Netflix did not merely adopt the internet. They built business models and ecosystems around it. They understood that the technology’s value came not from access alone, but from how it was organized, monetized, and embedded in platforms, workflows, partners, and users.
The lesson is clear: Technological leadership does not by itself translate into sustainable commercial leadership.
What Executives Should Do Now
For leaders navigating the generative AI era, five strategic realities stand out:
First, bottlenecks still matter. Generative AI may be advancing quickly, but value creation will still be constrained by compute, energy, specialized hardware, proprietary data, and talent inside the organization.
Second, net utility matters more than novelty. Customers and employees will adopt AI when the benefits clearly outweigh the economic and organizational costs. In many firms, implementation complexity will be as important as model quality.
Third, monetization matters more as costs fall. As model costs fall, the question will shift from who has access to the technology to who can price it, package it, and embed it in ways that create measurable customer value.
Fourth, regulation matters more than many firms expect. Rules around privacy, copyright, transparency, liability, and industry-specific compliance will influence where and how value can be captured. The complication for executives is not regulation alone, but regulatory fragmentation across regions, which may force firms to operate with different data, governance, and deployment models in different markets.
Finally, strategy matters most. There is no standalone generative AI strategy. There is only business strategy, strengthened by generative AI. Companies that chase AI for its own sake will accumulate pilots, vendors, and headlines. Companies that use it to sharpen a clear strategic position will build advantages that last.
Generative AI is likely to be transformative. But its value will not come simply from the power of the models. It will come from how firms navigate uncertainty, assemble complements, and embed the technology into business models and ecosystems that make value creation sustainable.
That is where the real competitive game begins.
References
Adner, R., & Kapoor, R. (2016). Right tech, wrong time: How to make sure your ecosystem is ready for the newest technologies. Harvard Business Review.
Kapoor, R., & Klueter, T. (2017). Organizing for new technologies. MIT Sloan Management Review.
Kapoor, R. (2018). Ecosystems: Broadening the locus of value creation. Journal of Organization Design.
Kapoor, R., & Eklund, J. (2018). Research: Self-disruption can hurt the companies that need it the most. Harvard Business Review.
Kapoor, R., & Klueter, T. (2020). The uncertainty factor. MIT Sloan Management Review.
Kapoor, R., & Teece, D. J. (2021). Three faces of technology’s value creation: Emerging, enabling, embedding. Strategy Science.
