The Real Challenge of Tech Business Schools: Scientists Understand Business, So Why Hasn't Their Behavior Changed?

marsbitPublished on 2026-10-10Last updated on 2026-10-10

Abstract

The core challenge discussed in the article is the gap between acquiring business knowledge and changing real-world behavior for scientist-entrepreneurs. It begins with an anecdote where a robotics founder with a technical PhD and an MBA sought an introduction to a CEO using vague, non-actionable language. This illustrates that merely understanding commercial concepts does not automatically translate into effective action in complex business scenarios. The article argues that forming true business capability involves three layers: cognitive understanding, the ability to execute specific actions in context, and the development of stable behavioral patterns. A significant issue is that scientist-founded startups often fail to conduct "commitment design"—transforming a technology's potential value into concrete, low-risk next steps that various stakeholders within a client organization are willing to undertake. This involves addressing not just technical performance but also integration costs, workflow changes, and risk allocation. The role of technology managers is redefined beyond simple matchmaking. They are categorized into three responsibility levels: facilitating transactions, managing conversion projects, and building viable enterprises. For deep-tech startups, the most critical need is for operational partners who co-own business decisions, not just advisors. True complementarity between a scientist and a business partner requires the scientist to delegate authority o...

I once engaged with a founder of a robotics company who holds a Ph.D. in engineering. After starting his business, he realized the importance of commercial acumen and went on to study business. From our conversation, I could sense a genuine breakthrough in his understanding.

Later, he wanted to connect with the CEO of a listed company to explore collaboration opportunities. Since I knew the CEO, he sent me a message, which roughly read: "Could you help arrange an introduction? Let's meet, chat, and have a meal." Upon receiving this message, I found myself unsure of how to help.

Knowing a CEO and facilitating a valuable industry collaboration are separated by significant work. Why would the other party agree to meet? In which business scenarios might they collaborate? What assessments need verification beforehand? What rationale should I provide for the introduction, and for which information can I vouch? None of this was addressed in that message. This incident prompted me to reconsider a question often overlooked in business education: Why does an individual, already equipped with business knowledge, revert to their original modes of action in real-world scenarios?

It also pushed the value of technology managers to a deeper level: Are we merely supplementing scientists' knowledge and finding resources for them, or are we building a sustainable organization around a technology?

I. A Dinner Invitation Exposes Incomplete Collaboration Design

The founder's issue cannot be simply attributed to poor communication skills, nor can his overall business acumen be judged based on this single instance. However, in this request, he framed a collaboration that required thoughtful design as merely a meeting that needed scheduling. *Who* to meet was decided; *why* the meeting was worthwhile and *what* might be advanced afterward remained unformed. This meant he left the tasks of defining the collaboration opportunity, assessing the other party's interests, and structuring the discussion agenda entirely to the introducer.

The introducer also faces another cost: credibility. If I forwarded this request, the recipient would likely interpret it as: I have judged this matter worth discussing, having at least confirmed its relevance to his business. Without sufficient supporting material, this introduction would require using personal credibility to fill the gaps in commercial rationale. Relationships can lower the cost of contact, but they cannot replace the reason for collaboration. This is why having a name in your contacts is different from having a genuine client opportunity.

There are distinct thresholds from acquaintance, to willingness to meet, to internal resource allocation, budget approval, pilot responsibility, and technical validation. A CEO's endorsement does not automatically generate a business unit's budget, technical team's validation plan, or field personnel's cooperation. If a collaboration plan is not yet formed, the higher it is pushed up the chain, the more likely it is to receive a polite "we can look into it," only to stall at the next level. What truly needs preparation is often a one-page collaboration proposal that enables the other party to make a judgment: What initial hypothesis do we have about which business scenario? What value might the technology create, and what evidence is still lacking? Which role do we hope to verify with first? What are the next steps for validation? What do both sides need to invest, and to what extent before deciding to proceed?

These points do not require the founder to know all the answers in advance. They require organizing the unknowns into a problem worth jointly verifying. Only then does the meeting gain purpose, the introducer knows what they are introducing, and the other party has a reason to mobilize internal resources.

II. Cognitive Change Completes Only Part of Capability Formation

After years in business education, I have become increasingly wary of an overly smooth assumption: that if a person understands customer orientation, they will naturally act customer-centrically the next time they face a client.

Reality is often not this smooth. An individual can agree in class that "we must understand the customer," yet return to the field and still lead with their own technology. They can comprehend that "collaboration requires mutual benefit," yet when making a request, only express their own needs. They can know that "organizations need delegation," yet when encountering important decisions, still centralize all judgment.

There are at least three levels here. Knowing what is important is cognition; being able to recognize and execute specific actions in concrete situations is action capability; and continuing to do so consistently without reminders forms a stable behavior pattern.

Wood and Neal's research on habits indicates that frequently repeated behaviors become associated with situational cues; new goals or understandings do not necessarily directly override established responses. This provides a mechanism for understanding "knowing but not doing," but it shouldn't lead to diagnosing a founder as having a habit problem. The real gap may also stem from not knowing the sequence of actions, lacking on-the-ground feedback, or having insufficient time and organizational support.

Therefore, changing behavior requires further inquiry: Regarding this collaboration request, does he know what to research first, whom to verify with, what materials to prepare, and what scope to propose for the next step? If there's no response, can he discern whether the issue lies in value proposition, evidence, timing, or contacting the wrong person? These judgments are difficult to make by merely remembering the phrase "customer orientation."

Training transfer research also divides post-training application into two tests: whether the learning can be applied to work, and whether it can be sustained; it also discusses influences from the learner, training design, and work environment. For tech business schools, the problem lies here: if assessment stops at whether learners understood and reports were complete, cognitive progress is prematurely equated with operational capability formation. A deeper misjudgment is: writing all capabilities needed by the enterprise into the founder's personal learning agenda. Even if this scientist gradually changes his behavior, the enterprise cannot wait forever for him to simultaneously become the product lead, sales lead, organizational lead, and financing lead. Personal growth has its own pace, but enterprise survival has deadlines dictated by cash flow, delivery, and client windows.

III. Productization Encompasses the Entire Process of Client Organizational Adoption

A major challenge for scientist-founders is turning achievements into products clients need. However, the phrase "client need" can also be interpreted too shallowly. A client may acknowledge the technology's advancement, admit a problem exists, and even be willing to watch a demo, yet still hesitate to purchase.

This is because between technological value and adoption value lies a set of operational calculations. Taking robots entering industrial sites as an example—this is a mechanistic illustration, not necessarily the aforementioned company's experience—performance advantages must be placed within real operating conditions: Does it change existing workflows? Does it require line shutdowns for retrofitting? Who restores production during anomalies? Who is responsible for maintenance? Are field personnel willing to use it?

Beyond purchase price, there are costs for integration, training, downtime, maintenance, and responsibility for failures. Technological gains may be offset by the costs of organizational adoption. More complexly, these benefits and costs may not fall on the same person. Management seeks cost reduction and efficiency gains, while on-site managers first bear the risk of production disruption. The R&D department may be willing to try new technology, but procurement needs clear specifications and delivery responsibilities. For a startup, proving technical feasibility is different work from making these roles willing to jointly advance.

This requires someone to perform what I term "Commitment Design" work: transforming the value technology can create into concrete next steps each party is willing to commit to within defined boundaries. It must at least clarify: Who is responsible? What is invested? What evidence constitutes acceptance? How are losses controlled if it fails? Under what conditions does collaboration expand? Under what conditions does it stop?

Good design doesn't necessarily seek a large order upfront. Sometimes it's verifying operating conditions, sometimes a clearly bounded test, sometimes confirming collaboration isn't suitable at this stage. The key is that each step reduces uncertainty for the next. If this work is missing, founders tend to repeatedly add technical presentations and seek higher-level connections without increasing concrete commitments clients are willing to undertake.

Financing faces a similar issue. What has the technology proven? What have clients validated? Which specific risk will the next round of funding address? These need to be articulated separately. Capital can provide time and resources for validation, but it cannot automatically complete validation for the enterprise. When operational responsibility is unassigned, injecting funds may only prolong the repetition of existing actions.

IV. Technology Managers Need Differentiation Based on Responsibilities Undertaken

The value of technology managers should be understood precisely within these gaps. The interpretation of relevant policies in Beijing for 2022 already covers the work of technology managers in technology productization, commercialization, resource organization, and helping scientists find co-founders and build startup teams, among other aspects. Thus, understanding this profession cannot focus solely on information brokerage.

However, it's equally misguided to expect every technology manager to become a CEO. I prefer differentiating three types of work based on actual responsibility. The first type is transaction facilitation. Identifying supply and demand, finding partners, assisting with technology licensing or transfer. This work has independent value, bounded by a specific transaction.

The second type is organizing conversion projects. Coordinating technology, intellectual property, validation, industry partners, and delivery arrangements to help an achievement cross multiple stages. Responsibility extends to agreed project milestones.

The third type is enterprise building. Participating in selecting application directions, defining products, developing customers, organizing teams, and arranging funding, enabling a viable operating entity to gradually grow around the technology.

These three types can overlap but should not be addressed with identical competency requirements, compensation methods, or evaluation criteria. The operational complementarity scientists need when starting a business primarily occurs in the third type. Some individuals are suited to be CEOs, others to take on COO roles, while in some phases, a deeply involved operational partner can first assume specific tasks.

Becoming a co-founder means sharing the burden of trade-offs. Technology can develop in multiple directions, but which one gets the limited funding first? A client is willing to trial but requests extensive customization—should it be accepted? The most valuable R&D topic conflicts with a product that must be delivered soon—how to prioritize? These issues cannot be resolved by merely translating technical terms into business language. They require someone to continuously organize evidence, coordinate interests, and take responsibility for actions after decisions are made.

This also distinguishes a mentor from an operational partner: a mentor can offer valuable advice, while an operational partner must also translate advice into product priorities, staffing, client advancement, and cash expenditures. The rarity lies in the ability to organize action amid incomplete information and withstand the feedback resulting from that action.

V. Whether Complementarity Works Depends on Which Decisions the Scientist Is Willing to Cede

Finding someone with business experience does not automatically form operational complementarity. Excellent managers in mature companies are often adept at improving efficiency within existing products, clients, and processes. Early-stage hard-tech ventures may not even have defined application directions, client budgets, or delivery boundaries. The experience needed in these two phases is not identical.

When evaluating a candidate, ask less about *who* they know, and more about whether they have *done* these things: identified specific scenarios amid vague demand; modified products based on client feedback; organized pilots through to delivery; handled conflicts between technological promise and resource constraints. Also observe how they handle bad news. Can they honestly report client rejection? Can they recommend pausing a direction? Can they explain reasons for lack of progress, rather than constantly using new connections to mask old problems?

But the real difficulty may lie on the scientist's side. If the operational partner is responsible for clients but lacks authority to adjust product priorities; responsible for cash but cannot participate in spending decisions; responsible for the team but cannot define roles and delivery responsibilities, then they are merely given tasks, not the power needed to fulfill them. If the scientist verbally accepts business logic but still makes all key decisions based on personal research preferences, complementarity degrades into "someone to sell my technology for me." Ceding some operational judgment to the complementary partner is itself the most critical behavior change for a scientist-founder.

This does not mean research judgment must submit to short-term sales. Both parties need clarity: which technological boundaries cannot be breached, which application directions can be validated, which decisions are whose responsibility, and how major disagreements are handled. A co-founding relationship should, at minimum, align responsibility, authority, sustained investment, and benefit arrangements. Giving a title without delegation, granting equity without mechanisms for collaborative work, are insufficient to form true complementarity.

A more reliable matching method is to jointly advance a real project first. Experiencing client rejection, resource scarcity, and priority conflicts together reveals whether collaboration is possible. A pleasant meal cannot show these things. Simultaneously, commercial uncertainty cannot be entirely shifted onto the technology manager. Validation costs still need to be borne by funding, and operational roles also require basic support. Demanding profit-sharing only upon success, expecting long-term unpaid input, tends to filter out those capable of assuming operational responsibility, leaving those inclined toward short-term brokerage.

VI. Tech Business Schools Should Connect Action Training with Team Building

Looking back at tech business schools, courses remain valuable, but they must serve a more complete capability formation process. The first layer is helping learners change how they interpret problems: understanding how customers, organizations, and capital each make judgments. The second layer is training actions around real projects: conducting a needs interview, preparing a collaboration proposal, advancing a bounded validation, then revising judgments based on external feedback. The third layer is identifying who the enterprise lacks and how to genuinely involve this person in decision-making.

The U.S. National Science Foundation's I-Corps program provides a concrete reference: teams include a technical lead, entrepreneurial lead, and industry mentor; training emphasizes Customer Discovery through engagement with potential customers and partners. It doesn't prove a specific team configuration guarantees success, but it places role complementarity and real commercial interaction within the same training process.

For tech business schools, a more worthwhile teaching unit is also "a team engaged in a real task," not just the founder attending class alone. Faculty should include those who can observe on-the-ground actions, not just lecture on frameworks. Assignments should include actual collaboration materials sent and client feedback received, not just polished pitch reports. Assessment should observe whether learners can adjust actions based on evidence and proactively repeat effective actions in similar subsequent scenarios. Team-level changes should also factor into assessment: Are there clear responsible parties for product definition, customer validation, and operational advancement? Do they possess corresponding authority?

If learners merely become more articulate about business in class but never complete an effective validation or establish operational division of labor, the curriculum must continue addressing this gap. Tech business schools and conversion platforms can participate in finding operational partners and organizing joint training and phased validation. But as collaboration enters the enterprise, decision-making and responsibility must gradually be handed over to the founding team. Otherwise, the more capable the platform is at helping, the less likely the enterprise is to develop its own operational capabilities.

Returning to that robotics founder. He was already willing to step out of the lab, learn business, and seek industry collaboration. This progress deserves recognition. But the next level of capability will not form automatically because he knows more CEOs. It manifests in: Can he transform a wish into an opportunity others are willing to evaluate? Can he advance an opportunity into a bounded joint validation? Can he then configure the capabilities needed during validation into his team?

Therefore, the support system for scientist-founders needs to answer a more specific question: Beyond helping scientists understand business, have we ensured someone in this enterprise is continuously responsible for commercial action? The long-term value of a technology manager as an operational partner lies precisely in this sustained responsibility. Education can help people see gaps, training can help change actions, but complementary organizations ensure the enterprise doesn't have to wait for one person to learn everything.

This article is from the WeChat public account "Focus and Periphery" (ID: Insight-ABC), author: Chang Yunlei.

Related Questions

QWhat key issue does the example of the robotic entrepreneur's meeting request highlight in business education for scientists?

AThe example highlights that business education for scientists often fails to bridge the gap between theoretical knowledge and practical action. While the entrepreneur had gained business knowledge, he still used his original operational style, treating a complex business opportunity as a simple social meeting and outsourcing the core work of designing the partnership to the intermediary.

QAccording to the article, what are the three distinct levels involved in truly forming a new capability, beyond just cognitive change?

ABeyond cognitive change (knowing what is important), the three levels are: 1) Action Ability: the capability to recognize specific situations and execute appropriate actions. 2) Stable Behavior Pattern: consistently applying the new actions even without external reminders. The gap often lies in not knowing the sequence of actions, lacking real-time feedback, or having insufficient time and organizational support.

QWhat is 'commitment design' as described in the article, and why is it crucial for commercialization of technology?

A'Commitment design' is the work of translating the potential value of a technology into specific, bounded next steps that various stakeholders within a client organization are willing to commit to. It is crucial because it moves beyond proving technical feasibility to addressing the organizational costs, risks, and responsibilities of adoption, thereby generating concrete commitments (like trials or tests) instead of endless demonstrations.

QHow does the article differentiate between three types of roles for technology managers (or commercial partners) in supporting scientist-led ventures?

AThe article differentiates them by their scope of responsibility: 1) Transaction Facilitators: focus on identifying supply/demand and facilitating deals like technology licensing. 2) Transformation Project Organizers: coordinate multi-stage projects across tech, IP, and validation up to agreed milestones. 3) Enterprise Builders: participate deeply in building the company itself—defining products, acquiring customers, building teams, and managing funding to create a sustainable business entity around the technology.

QWhat critical internal challenge must scientist-founders overcome to achieve effective collaboration with a business partner, according to the article's analysis?

AThe critical challenge is for scientist-founders to be willing to genuinely delegate and share authority over key business decisions with their complementary partner. This involves granting the partner real power over areas like product priorities, customer strategy, and spending decisions, rather than retaining all judgment based on personal research preferences. Without this delegation of authority, collaboration devolves into mere task execution, not true co-founding.

Related Reads

When Will the Chip Stock 'Capitulation-Style Selling' End? BofA: Perhaps in Early 2027, Stock Price Pullbacks Typically Lead Fundamental Inflection Points by 6-9 Months

The SOX semiconductor index has fallen 38% (from 14,634 to 9,011) since its June 22, 2026 peak, over 136 days—exceeding the average historical peak-to-trough decline of 32% but less than half the average 253-day duration of past downturns (2011, 2015, 2018, 2022). BofA analyst Vivek Arya expects this "indiscriminate selling," driven by AI capex slowdown fears, to end around early 2027. Historically, SOX declines lead actual semiconductor sales declines by 6–9 months, implying real EPS cuts may emerge from Q1 2027, with capex reductions reflected in H2 2027—though stocks may already price in much of the bad news. SOX currently trades at 13.7x forward P/E versus 17.7x for the S&P 500 (a 22% discount, down from a 15% premium at the AI peak), leaving roughly 12% downside to the historical trough valuation of ~12x. US cloud capex runs at ~$300B/quarter (~$1.2T annualized) versus $1.555T consensus for 2027—about 23% below expectations if spending holds steady, equating to ~30 GW of compute deployment versus BofA's 40+ GW base case. A 20% capex cut would pressure AI semiconductor revenue by 20%+ and EPS by 35–40%. BofA outlines three scenarios: base case ( ~10% downside, bottoming Q1 2027, favor segment leaders), bear case (pressure into mid-2027, favor EDA, high-FCF analog, low-AI-exposure names), and bull case (favor memory, optics, quantum, CPUs). Potential stabilizers include policy clarity in Q1 2027, buybacks (Lam $25B, Nvidia, Micron starting Dec 9), and slower capex extending the upcycle.

marsbit14m ago

When Will the Chip Stock 'Capitulation-Style Selling' End? BofA: Perhaps in Early 2027, Stock Price Pullbacks Typically Lead Fundamental Inflection Points by 6-9 Months

marsbit14m ago

TSMC: October Revenue Up 60%, with Month-over-Month Growth Still Accelerating

TSMC reported October consolidated revenue of NT$586.89 billion, up 59.7% year-over-year and 14.7% month-over-month, signaling accelerating momentum. Converted to RMB, that is about RMB 119.4 billion, with cumulative revenue for the first ten months reaching NT$4.49 trillion, up 43.3% year-over-year. The near-60% growth came on top of an already high base, while the sequential acceleration confirms strengthening demand rather than a slowdown. The report arrived at a delicate moment. The Philadelphia Semiconductor Index had plunged 12.25% over two sessions, its worst single-day drop since March 2020, as political changes in the U.S. prompted new governors in Texas and Ohio to pause approvals for new data center projects. Markets pessimistically priced in a halt across the AI data center supply chain. Sentiment rebounded sharply on November 9, with the semiconductor index jumping 6.86%, and Taiwan's weighted index surging 1,960.06 points, its second-largest single-day gain on record, supported by government stabilization funds and short-selling restrictions. TSMC stands apart from other AI assets. As the only foundry capable of mass-producing 3nm and 2nm chips, it holds about 90% of the global AI chip foundry market, with advanced-node capacity booked through 2028. Its diversified revenue mix—68% high-performance computing, 21% smartphones, 4% automotive and 4% IoT—provides a buffer, allowing it to reallocate capacity to flagship phone chips, automotive computing, and edge AI if data center approvals slow. October's results validate management's guidance of roughly 44% full-year dollar revenue growth and capital spending of US$60-64 billion. Expansion continues, with 2nm ramping and Arizona's second fab moved up to Q3 2027; Goldman Sachs raised its 12-month target price to NT$3,300. Google's expanded eighth-generation TPU lineup adds further demand, with TSMC and other Taiwanese suppliers in the ecosystem. TSMC's ADR trades at about 24x trailing earnings, low relative to its growth. Risks include a prolonged U.S. data center approval freeze and geopolitical or oil-price shocks. In A-shares, domestic substitution names such as SMIC and NAURA remain driven by rising localization rather than Texas approvals.

marsbit15m ago

TSMC: October Revenue Up 60%, with Month-over-Month Growth Still Accelerating

marsbit15m ago

Trading

Spot
活动图片