Clear Soup and Utility Curve
On November 13, 2025, NVIDIA CEO Jensen Huang received the Stephen Hawking Memorial Award. During an informal exchange at the event, he once again found a memorable metaphor for his management philosophy: rather than a "Consommé"-like elite organization that is pure but bland, he prefers a "Stew"-like noisy and diverse mix.
In the context of modern corporate governance, Huang's remarks are a gentle challenge to classic efficiencyism. The traditional bottom-ranking elimination system (as advocated by Jack Welch) essentially follows a clear utility curve: removing the least efficient 5% to ensure optimal resource allocation and maximize short-term net profit.
However, Huang's philosophy points directly to a core contradiction: the trade-off between short-term efficiency and long-term innovation returns. The management model he advocates—"letting management serve people"—can only be realized if the company is willing to bear huge institutional costs. These costs far exceed what a good-looking profit and loss statement can reflect.
Definition of Moral Hazard and Boundary Conditions
The biggest risk of Huang's "Stew" theory lies in Moral Hazard.
He assumes that the 5% "failures" are high-value "explorers", not inefficient "redundant personnel". However, in an organization with information asymmetry, the primary challenge for managers is to define boundary conditions for failure within an inclusive framework.
This management model requires leaders not to rely solely on quantitative output indicators for decision-making. They must delve into the input side of projects: does failure stem from an active choice of high-difficulty, high-risk paths (valuable trial and error), or from insufficient capability or process negligence (inefficient mistakes)?
Without precise differentiation, tolerance for failure can be transformed into institutional incentives for mediocrity. This would lower the team's average effort level, thereby harming overall innovation capability. Therefore, Huang's management model is less about tolerance and more about extreme trust in the leader's personal discretion.
The Costly "Leadership Premium"
The "Stew" is expensive, with its price reflected in the manager's energy cost.
Traditional "Clear Soup" management achieves management outsourcing through standardization and process. The system itself becomes the decision-maker, and managers only need to execute established rules. But the "Stew" requires leaders to frequently intervene and provide personalized guidance, consuming enormous energy and time.
In an organization that allows chaos and trial and error, the team needs a solid "container" to prevent chaos from turning into internal friction. Managers must become this container, bearing all psychological safety responsibilities.
This model cannot be scaled and replicated because it depends on the leader's personal talent and energy investment. Only managers with extremely high emotional intelligence, strong control, and decisive decision-making can handle this high-risk, high-effort management structure. For organizations generally lacking this leadership premium, pursuing the "Stew" is nothing short of a governance disaster.
Institutional Alienation from Agents
However, the most fundamental institutional resistance does not come solely from the external market. A more corrosive force stems from incentive misalignment within the organization—a variation of The Principal-Agent Problem.
As agents of the organization, managers' actions are not always aligned with the organization's long-term, sustainable efficiency. They must bear pressure from company shareholders, boards, investors, etc., regarding profitability, stock price, short-term financial performance. These external indicators inherently do not focus on the team's long-term innovation capability but directly influence managers' decisions and behaviors.
A more subtle issue is that these external indicators are often internalized by managers, either actively or passively, becoming core standards for evaluating their own performance and planning career paths. When this internalization occurs, indicators that appear reasonable but actually serve interests outside the team begin to dominate managers' judgments:
- Personal career signaling: To pursue faster promotions or higher short-term cash returns, leaders actively avoid the chaos and high-risk projects brought by the "Stew," as they may create unpredictable fluctuations in annual evaluations.
- Short-termism: Leaders view "short-term gains" or "surface compliance" as core values, rather than treating "sustained innovation capability" as the most important asset. These self-set indicators, serving personal cash flow or career trajectory, ultimately become constraints on the team.
When leaders prioritize personal interests over the organization's long-term health, they actively push the management model toward "Clear Soup." "Clear Soup" is easy to quantify, easy to report upward, and easy to prove the agent's "effective control." This behavior of "letting the team serve management" rather than "letting management serve the team" is a covert action where leaders use institutional tools to sacrifice the team's long-term potential in exchange for their own short-term security or interest maximization.
Therefore, the challenges to Huang's management philosophy come not only from the greed of capital markets but also from agents at all levels within the organization. They use these internalized external indicators to serve their own sense of security and controllability, pushing the team toward a more manageable but less creative "Clear Soup" direction.
If an organization wants to practice his management philosophy, it must possess: the courage to accept failure, the willingness to pay high energy costs, and the skill to resist external short-termist pressures. These conditions stem from the manager's willingness, not ability.
A Call for Long-termism
Huang's "Stew" is a governance philosophy that places human capital value above short-term financial indicators.
In the AI era, with technology updating on a monthly cycle, the performance review window is extremely compressed. The "result pressure" managers face no longer comes from a single annual goal but from continuously converging short-cycle indicators. These indicators often have nothing to do with the team's long-term efficiency, technical capability accumulation, or organizational resilience, yet they profoundly influence managers' behavioral choices.
In this environment, agents are more inclined to view uncertainty as risk rather than opportunity. AI technology has not yet formed a stable business model, but the narrative around its potential is rapidly expanding, squeezing managers' decision-making space with "technology cycle pressure": the more uncertain it is, the more they need to prove "control" in the short term. Consequently, the space that should be for team experimentation, iteration, and exploration is replaced by stricter processes, more quantifiable goals, and more frequent checks.
The result is that the "Stew-like chaos" that can truly foster innovation becomes harder to maintain in the AI era. Managers must invest higher discretion costs to protect the team's trial-and-error space within densely packed performance cycles. Most organizations lack the capacity to bear this management effort, ultimately sliding further toward "Clear Soup": replacing governance methods that require patience, wisdom, and long-term judgment with structures that are reportable, measurable, and mass-producible.
The AI era should enhance an organization's exploration capability, but under the distortion of incentives, it may instead make it harder for organizations to maintain a truly innovative environment.
If a company cannot meet these prerequisites, its efforts to emulate the "Stew" will ultimately lead to the collapse of its governance structure. Therefore, Huang's speech is less a management secret and more a call to all companies and investors: To achieve true innovation, we must first adjust our incentive structures and reassess the institutional costs we are willing to pay for long-term value.