CKGSB Professor Leon Yang Zhu and his co-authors found that successful disruption depends less on being first to leverage the new technology than on choosing the timing that alters the competitor’s optimal response.
When a potentially disruptive technology appears, managers often frame the decision as a simple choice: move now or wait. But CKGSB Professor of Operations Management Leon Yang Zhu’s new paper Disruptive Timing, published in Management Science argues that this framing misses the strategic core of disruption. The key question is not simply whether to act early or late. It is how the timing of one firm’s move changes the other firm’s options.
In the paper, Leon Yang Zhu and his coauthors Guang Li (Smith School of Business, Queen’s University), Andy Wu (Harvard University) and Brian Wu (University of Michigan) examine disruption as a timing game between entrants and incumbents.
Their central insight is that successful disruption depends on the race between two forces: how quickly the new technology improves outside the firm, and how quickly firms accumulate experience with the technologies they use.
The paper’s core mechanism rests on two forces.
The first is external technological improvement: the way a technology gets better over time because of broader advances in science, engineering, customer demand or the surrounding ecosystem.
The second is firm-specific learning: the know-how a company builds only by using a technology over time.
External improvement benefits any firm using the technology. Firm-specific learning depends on when a firm enters, stays, or switches.
When external technological improvement is slow, waiting can help an entrant deepen the incumbent’s commitment to the old technology; when improvement is fast, the incumbent’s option to switch may force the entrant to move earlier than it otherwise would.
The central contrast in the paper is therefore counterintuitive but clear: slow relative speed can favor strategic delay, while fast relative speed can favor preemptive entry.
The paper then broadens the meaning of disruption. Rather than treating disruption as a single story in which a startup moves first and an incumbent collapses, such as smaller disk drives from Seagate disrupting larger drives from IBM, or digital photography disrupting Polaroid’s analogue film business, the authors organize the phenomenon into five types.
Their typology begins with a simple matrix: the entrant can race into the market or dawdle, and the incumbent can choose no response, a delayed response, or an immediate response.
From those combinations come five distinct disruption paths.
The paper is careful about how it uses real-world cases. The authors do not claim that their model is the only explanation of these examples. Instead, they use them to show how the theory can plausibly describe what managers observe in practice. Read that way, the case material makes the differences among the five paths much easier to see.
Seen through the paper’s framework, the OpenAI-Google search case resembles passing disruption. In public debate, Google is often cast as a slow incumbent that failed to recognize the threat. The paper offers a more disciplined interpretation. Generative AI may still be economically immature for core search at scale, even if it is technologically promising. In that context, delaying a full switch can be rational. The key issue is not whether Google can respond, but when switching becomes attractive enough to justify the move.
The Akamai-Cloudflare case can be read as an example of postponed disruption. By the time software-defined networking had become a serious alternative in content delivery networks, Akamai had already spent years building a global network around specialized hardware. That history may have left it too enmeshed in the old architecture to justify a full switch. The paper’s point is not simply that Cloudflare was later. It is that late entry can change the incumbent’s incentives.
The Samsung-Apple example illustrates phased disruption and shows why both timing and ecosystem maturity matter. A foldable smartphone may be conceptually attractive long before it is commercially ready. Samsung did not enter at the earliest imaginable moment, but it did enter before Apple, and the paper argues that this early learning may itself help deter Apple’s response.
Robinhood and Schwab, by contrast, show that disruption does not require the incumbent to disappear. Robinhood introduced commission-free trading, and incumbents such as Schwab adopted the same model and continued operating alongside the entrant, which is what the authors would call partial disruption. The old model can be displaced even when both firms survive.
The managerial implications follow directly from this logic.
For entrants, the paper’s advice is not to ask only “Can we enter now?”. Instead, they need to ask:
They have to decide explicitly whether to enter immediately or strategically delay. In some settings, preemptive entry before the technology is fully ready can help build a learning advantage and delay the incumbent’s response. In others, strategic delay can entrench the incumbent in the old technology and make retaliation less attractive.
For incumbents, the paper suggests a different discipline. Rather than asking only “Should we respond?”, executives should ask:
Waiting is not always a mistake, but it is only useful if the firm understands the economics behind that waiting. The authors argue that incumbents can improve their position by reducing switching costs, building barriers to entry, and continuing to invest in the broader ecosystem around the existing technology. Those actions can preserve flexibility and, in some cases, deter an entrant from entering at all.
Disruption depends on how technologies improve, how firms learn, and when firms move relative to one another. That is why the most important question is not simply whether to move first or wait. It is when action changes the other side’s options.
Sometimes waiting helps the entrant. Sometimes rushing helps the entrant. The difference lies in the relative speed of technological improvement and learning.
