California Management Review

For decades, the structure of multinational corporations followed a familiar pattern. Headquarters defined global strategy and controlled core resources. National subsidiaries executed locally, adapting marketing and sales activities to fit market conditions. That model worked well in a world where globalization was steadily deepening. Companies could centralize scale advantages while allowing subsidiaries to respond to local markets. Today, that balance is breaking down. Consider Wilo, a German industrial manufacturer with approximately €1.5 billion in revenue and around 7,800 employees operating in more than 50 countries. For decades, Wilo followed the classic multinational model with a strong central headquarters in Germany. However, by 2020, CEO Oliver Hermes concluded that this model was no longer sufficient. He argued that the global economy was increasingly splitting into three “tectonic plates” centered around the United States, China, and Europe. As he put it, “Wilo has to keep up with current global developments, whether we like them or not. A stronger regionalization of Wilo is necessary to continue our global success.” In response, Wilo began to fundamentally redesign its structure, establishing a second headquarters in China and planning a third in the United States. Rather than relying on a single global center, decision making shifted closer to regional markets, along with greater responsibility for investments and strategic priorities. Across industries, geopolitical fragmentation is reshaping the operating environment of multinational firms. Trade barriers, sanctions, local-content rules, regulatory divergence, digital sovereignty requirements, and shifting industrial policies are making it harder for companies to operate as fully integrated global organizations. “America First,” “Made in China 2025,” new EU regulatory regimes, India’s localization policies, and similar initiatives across regions share a common theme, i.e., strengthening local markets and reducing dependence on global integration. For multinational firms, this new environment creates a structural dilemma. Traditional models that separate headquarters and subsidiaries often prove too slow. The old formula of centralizing control and decentralizing execution no longer resolves the tensions created by deglobalization. As one executive in our study observed, “any decision needs to be escalated… which makes you very, very slow.” Based on a multi-method qualitative study including 18 interviews with C-level and senior executives, a six-month ethnography, and multiple case analyses across global B2B firms, we observe a recurring organizational response. Wilo’s move toward multiple headquarters is not an isolated case but reflects a broader pattern. Across firms in our study, companies are experimenting with intermediate structures that distribute decision authority across regions while maintaining global coordination. In response, many firms are experimenting with a new organizational solution, which we call confederated communities. These are cross-border, semi-autonomous structures that sit between headquarters and national subsidiaries. They form an intermediate layer that coordinates selected functions across subsidiaries while redistributing authority away from headquarters to a regional or cross-country level. By doing so, they allow companies to balance global coordination with local responsiveness in a more flexible way. In an increasingly fragmented world economy, such structures are becoming a critical mechanism for multinational adaptation. Why Deglobalization Creates Organizational Tension Deglobalization does not simply make multinational operations more complex; it intensifies structural tensions that push companies simultaneously toward greater both centralization and decentralization. Three pressures are particularly important. Competitiveness or Local Presence vs Global Scale Protectionist policies increasingly require firms to source and produce regionally. Companies that previously served markets through centralized production and global exports now face growing pressure to establish regional supply chains. Across our interviews, executives consistently highlighted the increasing importance of local value creation requirements. In our interviews, one executive described how, in the U.S., companies may need to demonstrate “60–70% local regional value” to compete in certain contexts. While this figure reflects practitioner experience rather than a universal rule, it illustrates the magnitude of the shift. As another manager summarized, “In the past, we could produce 100% in Germany and ship to the States. That’s no longer possible.” Decentralization, therefore, becomes necessary to remain competitive in procurement processes and regulatory environments that prioritize domestic value creation. At the same time, fragmentation threatens the scale advantages that global firms rely on. Centralized purchasing and standardized production have long provided cost efficiencies. As one executive noted, “We get better prices if we purchase higher volumes centrally.” If each region localizes independently, purchasing power declines, duplication increases, and efficiency suffers. Firms are thus compelled to consolidate resources centrally to defend scale advantages, even as local regulation pushes them in the opposite direction. A concrete example comes from the case study company in our research, a global pump manufacturer. Facing declining demand linked to trade restrictions, the firm was forced to rethink its global marketing and supply structure, balancing regional responsiveness with centralized cost control. Compliance or Local Regulation vs Global Governance Regulatory regimes are diverging rather than converging. Certification requirements, sustainability reporting standards, cybersecurity regulations, and product approval processes increasingly vary by jurisdiction. Local teams are often best positioned to interpret and implement these requirements. One company in our study encountered this when sustainability requirements became more prominent in procurement processes. It responded by creating a cross-subsidiary sustainability committee that developed global assessment criteria while allowing subsidiaries to adapt implementation locally. This enabled the firm to remain compliant across markets while maintaining a consistent global position. At the same time, global firms must ensure governance consistency and brand integrity. Excessive decentralization can lead to inconsistent practices and reputational risks, while excessive centralization can slow responses to regulatory change. At one multinational company we did an ethnographic study with, illustrates this tension. The firm deliberately retained centralized production in certain areas despite local pressure, fearing that decentralized manufacturing could lead to inconsistencies in globally required quality standards. Competence or Local Knowledge vs Strategic Coherence Local subsidiaries often possess the deepest knowledge of their markets and the closest relationships with customers. In many fast-growing regions, particularly in parts of Asia, market dynamics and innovation cycles move faster than headquarters decision processes. As one executive noted, headquarters R&D teams “do not have the understanding of the local market” required to respond effectively. Granting subsidiaries more autonomy can therefore improve decision quality and speed. Across our interviews and case observations, firms increasingly relied on such localized expertise to improve responsiveness and decision quality. Firms increasingly rely on this local expertise. For example, one company in our study designated its Chinese subsidiary as a center of competence for “value products,” leveraging its deep understanding of cost sensitive customer segments. However, autonomy also creates risks. Too much decentralization can fragment strategy, as subsidiaries may pursue initiatives that duplicate investments, diverge from global priorities, or weaken brand consistency. One executive described how excessive decentralization led to strategic drift, with local initiatives emerging that did not align with the overall direction of the firm. Competitiveness, compliance, and competence all require both centralization and decentralization at the same time. The result is a structural tug of war. At the same time, it is important to recognize that these tensions do not play out uniformly across the organization. The appropriate balance between centralization and decentralization often differs significantly by value chain function. Decisions in marketing and sales, for example, tend to require stronger local responsiveness and therefore lend themselves more naturally to regional or cross-border coordination structures. In contrast, functions such as production, R&D, IT, finance, or HR may follow different logics, often driven more strongly by standardization or global governance requirements. As a result, multinational firms increasingly adopt differentiated structural solutions across functions rather than a single, uniform model. Traditional organizational models struggle to resolve this paradox. The Rise of Confederated Communities In response, many multinational firms are creating intermediate organizational structures that sit between headquarters and national subsidiaries. We conceptualize these forms as confederated communities. A confederated community is a cross-border group of employees drawn from headquarters and multiple subsidiaries who collaborate to manage specific functions, such as marketing, innovation, procurement, or regulatory coordination, across a region or group of markets. These communities centralize certain activities across subsidiaries while at the same time decentralizing authority away from headquarters. From the perspective of national subsidiaries, decision authority moves upward into a regional or cross-market structure. From the perspective of headquarters, authority moves downward into that same structure. What distinguishes confederated communities from traditional coordination mechanisms is that they redistribute decision authority and occupy a distinct organizational layer between global headquarters and national subsidiaries. The logic behind confederated communities closely mirrors what companies like Wilo are implementing at the corporate level. By establishing multiple headquarters across key regions, Wilo effectively redistributed decision authority, investment responsibility, and strategic influence closer to regional markets. For example, the China headquarters was given authority over regional investments, R&D hiring, and supply chain decisions, while remaining connected to global leadership structures. This reflects the same underlying principle: shifting from a single center of control to a more distributed, yet coordinated, organizational model. Across interviews, case studies, and ethnographic observations, we repeatedly observed similar structural responses across firms operating in Europe, China, and emerging markets. Despite differences in size and industry, companies converged on comparable intermediate structures that balance global coordination with regional autonomy. A striking example from our study is a European industrial firm that established a regional marketing hub in China. This unit consolidated marketing activities previously handled independently by Asian subsidiaries, while also taking over responsibilities from European headquarters. The result was faster regional decision making and better adaptation to local customer and regulatory requirements. A similar pattern can be observed in the broader manufacturing landscape. Volkswagen, for instance, created the Volkswagen China Technology Company to consolidate development and decision making for China specific vehicles, significantly shortening time to market. Siemens and Bosch have also strengthened regional structures in China and India, giving local units greater authority over product adaptation and customer engagement. Confederated communities are particularly visible in marketing and sales functions, where firms must balance global brand coherence with local adaptation. At the same time, similar structures are emerging in areas such as sustainability management, regulatory compliance, digital platforms, and product development. What Confederated Communities Look Like in Practice Confederated communities appear in several forms, ranging from informal to highly institutionalized structures. Cross-Border Networks At the informal end, firms rely on cross-border networks of managers who collaborate across subsidiaries to address shared challenges. For example, product specialists from several national markets may jointly develop training programs or customer solutions tailored to regional needs. In one multinational corporation we studied ethnographically, product-specific training is delivered through informal collaboration among regional experts rather than centralized programs. This allows the company to maintain technical consistency while adapting training to local market conditions and customer requirements. These networks enable companies to share knowledge quickly without waiting for headquarters to standardize global processes. Cross-Subsidiary Committees More structured forms include cross-subsidiary committees that address organization-wide challenges such as sustainability reporting, cybersecurity policies, or regulatory compliance. These committees combine local expertise with global coordination. By integrating perspectives from multiple markets, firms can develop policies that are globally aligned but locally feasible. The sustainability committee mentioned earlier is one example. Another firm created a COVID-19 committee that initially relayed headquarters guidance but later developed region-specific policies as conditions diverged. In areas such as sustainability or data governance, where regulations increasingly vary across jurisdictions, these committees often become central to organizational coordination. Strategic Focus Groups Some firms establish strategic focus groups composed of experts selected for their functional expertise rather than geographic representation. These groups collaborate on cross-market strategic initiatives, such as digital platform development, portfolio management, or commercialization strategies. One company in our study used such a group to decide which products to prioritize globally, avoiding slow hierarchical decision processes. By locating decision authority closer to those with relevant information, focus groups reduce escalation bottlenecks and accelerate innovation. Regional Hubs and Centers of Excellence Centers of excellence or regional hubs constitute more permanent confederated communities. These semi-autonomous units concentrate specialized capabilities in one particular region and influence decisions across multiple markets. A regional hub might focus on developing cost-optimized products for emerging markets, managing regional digital infrastructure, or coordinating regulatory compliance across neighboring countries. Unlike purely local subsidiaries, these centers of excellence often shape global strategy rather than merely implementing it. Regional Powerhouse The most formalized form is the regional marketing department or “regional powerhouse.” Here, firms institutionalize a dedicated regional structure responsible for coordinating marketing and sales activities across multiple countries. Authority over selected functions, such as regional branding or channel strategy, shifts from headquarters to the regional level. At the same time, activities previously fragmented across individual subsidiaries become centralized within the region. This configuration enhances speed and regulatory responsiveness in strategically important markets while preserving global standards. In practice, many companies refer to such structures as regional headquarters, although their role increasingly extends beyond administrative coordination toward active decision-making and strategic integration. Across these forms, the defining characteristic is the creation of a shared organizational space where global and local logics intersect. Confederated communities institutionalize both global coordination and local responsiveness at the structural level. Why These Structures Work in a Fragmented World In a world of geopolitical fragmentation, multinational firms must constantly adapt their operations to shifting regulatory environments and competitive dynamics. Confederated communities improve organizational performance in three ways. First, they improve organizational sensing. Because these communities connect managers across markets, they allow companies to identify emerging opportunities and risks earlier. Second, they improve decision speed. Locating decision authority closer to regional markets reduces delays associated with escalating issues to global headquarters. Third, they improve organizational flexibility. Confederated communities enable coordinated adjustments across multiple subsidiaries without requiring large-scale global restructuring. One multinational company in our study illustrates this well. By relying on cross-border communities for training, sustainability, and coordination, it was able to respond to protectionist pressures while maintaining global consistency. Managers reported faster response times and improved coordination across regions compared to prior centralized structures. Similarly, Wilo’s multi-headquarters approach allowed the company to respond more directly to regional regulatory environments, customer requirements, and supply chain disruptions, particularly in Asia and North America, where local responsiveness became critical for growth. Instead of repeatedly shifting between centralized and decentralized models, firms develop a stable yet flexible intermediate layer capable of adapting to regional variation. Design Implications for Leaders For executives, the emergence of confederated communities raises important organizational design questions. These structures can significantly help multinational firms navigate fragmentation. But if poorly designed, they may add complexity. Clarify decision rights: Leaders must clarify which decisions remain global, which are handled at regional level, and which remain with national subsidiaries. Ambiguity slows decision-making and weakens accountability. Align incentives across organizational levels: If headquarters rewards global standardization, while regional communities are evaluated on local growth, structural conflict will persist. Performance metrics must reflect shared objectives across global and regional layers. Develop boundary-spanning leadership: Leaders within confederated communities play integrative roles. They must navigate political dynamics and translate strategy across contexts. These roles require strategic judgment rather than administrative coordination. Institutionalize gradually: Shared digital platforms and data infrastructures increasingly allow firms to maintain global transparency and coordination even as decision authority becomes more distributed. When Confederated Communities Make Sense Confederated communities are particularly valuable in environments where regulatory divergence is high, regional growth patterns differ significantly, innovation speeds vary across markets, and local competitors are gaining strength. In such contexts, purely centralized models struggle to respond quickly enough, while fully decentralized models risk fragmentation. In contrast, industries with highly standardized products and homogeneous regulatory environments may still benefit from traditional centralized structures. Confederated communities are, therefore, not a universal solution, but rather a contextual response to the realities of geopolitical and economic fragmentation. Toward a Confederated Multinational The classic debate in international management has long revolved around subsidiary autonomy. How much freedom should local units have? Deglobalization shifts the focus of this question. The challenge is no longer merely distributing autonomy but redesigning the architecture through which global and local forces interact. Confederated communities represent one response to that challenge. Rather than relying solely on hierarchical control from headquarters or granting complete autonomy to national subsidiaries, multinational firms are increasingly building layered structures that combine global coordination with regional authority. In this emerging model, the multinational corporation begins to resemble less a rigid hierarchy and more a confederation of interconnected communities. Deglobalization is reshaping not only markets but also organizational design. The traditional multinational, structured as a hierarchy with headquarters at the top, is evolving into a more networked system with empowered regional layers. In this emerging model, the multinational corporation begins to resemble less a rigid hierarchy and more a confederation of interconnected communities. This shift does not imply the end of globalization. Nor does it signify a retreat into national silos. Multinational firms will continue to operate across borders and benefit from global integration. But as geopolitical fragmentation reshapes markets and regulations, the organizational architecture of the multinational enterprise is evolving. Wilo’s trajectory illustrates this transformation in practice. What began as a single European headquarters has evolved into a multi-headquarters structure spanning China, the United States, and, in February 2025, an additional regional headquarter in Dubai. This progression reflects a broader shift from centralized global control toward distributed regional authority. In a fragmented world, competitive advantage will increasingly depend not only on strategy, but on organizational architecture. The key question is no longer whether to centralize or decentralize. It is how to do both at the same time. Confederated communities offer one answer and may become a defining feature of the next generation of multinational organizations.

Most corporate venture capital is solving the wrong problem. The core challenge is not finding promising startups, but ensuring that gained insights actually change decisions inside the parent company. Bolt-on AI — tools layered onto legacy workflows — will not fix a broken operating model; it will accelerate its weaknesses. Drawing on a decade of longitudinal research on European CVCs, we introduce AI-native CVC: a fundamental redesign, built around five organizational shifts, in how knowledge flows between startups and the parent organization. The question is not whether to adopt AI. It is whether leaders will redesign the organization around it. The $229 Billion Blind Spot Corporate venture capital has become a primary mechanism through which established firms access external innovation.1 In 2025, more than 3,000 corporations invested in startups across more than 5,000 deals worth a combined $229 billion—up 75 percent year on year—and more than half of all dollars deployed into startups worldwide included a corporate investor.2 CVC units participated in 68 percent of global AI deal value, a historic high, driven by the conviction that proximity to startups is the fastest route to AI capability.3 Salesforce Ventures has deployed over $850 million through a dedicated AI fund into 35 enterprises whose combined valuations now exceed $270 billion.4 Intel Capital—before spinning off as an independent fund in January 2025—had invested over $20 billion across more than 1,800 companies and orchestrated roughly 1,300 curated customer introductions for its portfolio in 2024 alone.5 The ambition is clear. The execution is not. Behind the deal volume lies a persistent, awkward problem. Most CVC units generate more information than their parent organizations can absorb. They surface promising startups, produce detailed investment memos, and sit on portfolio company boards. And then, quarter after quarter, the knowledge stays trapped inside the venture team—never reaching the business units that could act on it.6 AI makes this problem worse before it makes it better. Today, 85 percent of private capital dealmakers already use AI to automate daily tasks, up from 76 percent a year earlier, and roughly 82 percent of PE and VC firms are actively using AI across their operations.7 The Data Driven VC Landscape now tracks 235 data-driven investment firms globally—up from 151 two years ago—with roughly a third generating more than 40 percent of their deal flow through algorithmic tools.8 But these tools have been built, bought, and benchmarked for financial venture capital, not corporate venture capital, and the distinction matters profoundly. A financial VC succeeds by picking winners. A CVC succeeds by connecting external innovation to internal capability. AI optimized for the first objective can quietly undermine the second. That is the blind spot. The Alignment Trap The most common assumption in boardrooms today is that AI-powered scouting helps CVC units cast a wider net—and that a wider net catches better fish. A decade of longitudinal research on European corporations tells a more uncomfortable story.9 For capital-intensive firms—those whose competitive advantage rests on manufacturing plants, logistics networks, or infrastructure—CVC investments that match the parent’s core business produce meaningful performance spillovers. Under performance pressure, these “sticking-to-the-core” investments help firms extract more value from existing assets. But when the same firms pursue non-matching investments in distant domains, performance declines. The operational structure that makes these firms powerful also makes them rigid: they lack the flexibility to absorb genuinely unfamiliar knowledge. Asset-light firms face the opposite dynamic. Software companies, professional services firms, and platform businesses can benefit from non-matching investments precisely because their lighter structures allow broader adaptation. The implication for AI is direct and underappreciated. When a CVC unit deploys AI-powered scouting that dramatically expands the opportunity set, it also increases the probability of surfacing opportunities the parent cannot absorb.10 The CVC team sees a brilliant robotics startup. The AI confirms strong signals. The investment goes through. But the chemical manufacturer that owns the CVC lacks the technical talent, IT infrastructure, and organizational culture to integrate what the startup offers. The deal becomes a financial bet dressed up as a strategic one. Recent research on CVC as a gateway for corporate AI adoption reinforces and extends this point. A multiple-case study of CVC-enabled AI collaborations across Europe identifies four distinct AI-adoption archetypes—internal adoption, customer-facing adoption, product integration, and co-creation—and suggests that each triangular relationship among corporate, CVC, and startup can be unique, with firms pursuing different adoption types depending on the AI sector, the investment strategy, and the technological fit between corporate and startup.11 Without that organizational infrastructure, investments accumulate but knowledge does not, as shown in non-CVC contexts by Ericsson’s decade-long pursuit of AI value: profiting from AI requires co-specialized complementary assets in data, capability, and organization. Without them, the technology itself is rarely the constraint; the surrounding system is.12 The Bolt-On Illusion Most organizations apply AI to CVC the way they apply it elsewhere—as a tool layered onto existing processes supposedly leading to better scouting platforms, automated screening, and faster due-diligence summaries. A 2026 World Economic Forum report, citing Accenture research of 450 executives across five sectors, confirms that only about 15 percent of organizations use AI to fundamentally redesign how work is performed.13 In CVC, the legacy workflow is sequential and siloed: source → screen → diligence → decide → manage → report.14 Each handoff loses context. The scout understands the technology but not the business unit’s pain points. The analyst builds a financial model but has never spoken to the VP of engineering who would integrate the startup’s product. The investment committee sees a polished deck, not the evolving evidence that shaped it. Current applications of AI tend to accelerate individual tasks, yet often fail to improve overall system performance because they overlook interdependencies across tasks.15 The scouting platform finds more startups, but the same bottleneck at the screening stage remains. The due-diligence tool generates faster summaries, but the committee still decides from static presentations. The knowledge flows that determine whether a CVC creates strategic value— outside-in transfers from startups to business units, inside-out deployment of corporate resources, inside-in linkages across silos, and outside-out ecosystem orchestration—are still constrained by the same organizational barriers.6 Research on capturing value from AI shows why. Profiting from AI requires overcoming data, capability and value appropriation challenges through complementary assets and co-specialization—the technology itself is rarely the constraint.12 Without these assets, productivity gains at the task level evaporate at the system level. Corporates, notably, are behind. While financial VC firms use AI across sourcing, screening, portfolio support, and LP reporting, many CVC units have been slower to integrate AI across workflows—often because they operate with smaller technology budgets and less organizational autonomy than independent VC firms. A vanguard of CVCs, including NGP Capital and Salesforce Ventures, shows what is possible once those constraints are addressed.16 The lag is not primarily technological. It is organizational. Designing an AI-Native CVC System An AI-native CVC program shifts from adding tools to existing tasks and legacy workflows towards redesigning the operating model, in which agent-based systems integrate reasoning, decision-making, and execution to enable dynamic workflows—provided appropriate governance and oversight.17 The shift can be expressed as five coordinated redesigns. From episodic to continuous scouting. Traditional CVC relies on networks, conferences, and inbound flow.14 AI enables more systematic monitoring of signals such as patents, hiring, and ecosystem activity via platforms like Harmonic, Specter, and PitchBook, and can be configured to adapt as corporate priorities shift. From sequential decisions to parallel evaluation. Legacy CVC processes are linear; AI tools enable overlapping technical, market, and business-unit assessments. Over time, this shifts investment committees toward interpreting dynamic evidence rather than static decks—though the content they interpret still needs to be checked and validated by humans.18 The committee no longer asks, “Does this pass our criteria?” It asks, “What does the evolving evidence tell us about strategic fit?” Human judgment becomes more important, not less because the questions become harder, demanding the experienced pattern recognition and organizational heuristics that cannot be algorithmically replicated.19 From sampling to broader due diligence. Traditional due diligence examines document subsets under time constraints. AI enables broader analysis of contracts, codebases, and transaction data, helping reviewers identify patterns across larger datasets. For technical assessments, AI agents can flag signs of technical debt, scalability issues, and security risks; for legal diligence, they can parse clauses across many customer agreements; for commercial diligence, they can reconstruct cohort dynamics from transaction logs. The shift is from “What documents do we have?” to “What do all of these documents mean together?” From portfolio management to knowledge orchestration. Here lies the greatest unrealized potential—and the dimension that most clearly separates corporate from independent venture capital. Research on knowledge flows in CVC identifies four types: outside-in transfer from startups to the parent, inside-out sharing of corporate resources with startups, inside-in connections across corporate silos, and outside-out ecosystem-enriching exchanges among portfolio companies.6 In practice, CVC units often focus on the outside-in flow, while the other three receive uneven attention. AI-native systems can help make all four flows more visible and structured, supplementing (not replacing) the personal networks on which CVC teams already. The system can help surface candidate connections, for example: this startup’s sensor technology maps to an R&D challenge documented by that engineering team in another country. The CVC unit evolves from an investment team into a knowledge-orchestration function—connecting capabilities across organizational boundaries that neither startups nor business units can see on their own. From periodic to more frequent portfolio review. Strategic steering of CVC portfolios already happens between annual reviews—board meetings, quarterly committees, and deal-by-deal discussions all shape direction. AI can support that rhythm by helping teams track startup milestones, burn rates, and competitive shifts more frequently and with less manual effort. Monitoring becomes useful when it reduces the lag between a meaningful change in a portfolio company and the parent’s ability to notice and discuss it—not when it becomes an always-on time.20 Early-warning dashboards can flag material changes in a portfolio company’s situation earlier, giving the CVC team time to convene the right people—rather than constraining the startup’s path. The aim is to inform human judgment between review cycles, not to turn the portfolio into a cage. Taken together, these five shifts constitute a different operating model, not a better toolkit. They also expose the organizational preconditions most CVC units have not yet built: cross-functional review cadences, shared data infrastructure between ventures and business units, investment-committee cultures that tolerate living evidence, and incentive structures that reward knowledge flow as well as financial return. The Human Question As AI absorbs more analytical work, what remains for the human investor? Research on the automation–augmentation paradox identifies a core tension: the same technology that enhances human capabilities can also erode them if deployed carelessly.21 A field experiment of 758 management consultants found that AI users completed significantly more tasks at higher quality—inside the “jagged frontier” of AI competence. Outside that frontier, AI-assisted consultants performed worse than unassisted peers. The users who did best maintained clear boundaries between their own judgment and the model’s contributions; “centaurs” who alternated between domains outperformed “cyborgs” who fused their work with AI in continuous, undifferentiated co-creation.22 For CVC, this suggests an asymmetric design. High-stakes decisions—investment approval, founder assessment, strategic positioning—should remain centaur processes, where experienced professionals use AI as an analytical instrument but retain full decision authority. Exploratory tasks—market mapping, pattern discovery, trend identification—benefit from cyborg interaction, where iterative human–machine dialogue generates insights neither side could produce alone. The deeper risk is not that AI replaces human judgment but that it quietly degrades it. If junior CVC professionals never learn to evaluate a startup without algorithmic support, the organization loses the experiential foundation on which strategic judgment depends.23 Building human expertise must remain a deliberate organizational investment, even—especially—as AI handles an increasing share of analytical work.24 Three Risks Leaders Must Confront Algorithmic convergence. If CVC units adopt similar AI platforms and draw from the same data sources, their scouting and screening outputs will converge. Capital crowds into the same startups faster, valuations inflate, and the proprietary strategic advantage that justifies CVC’s existence erodes. Financial regulators have taken notice. Financial regulators have flagged AI-driven convergence as a systemic risk, warning that reliance on a small number of dominant models and data providers can amplify correlated behavior across markets.25 Research on organizational heuristics highlights the role of simple, firm-specific decision rules in guiding choices under uncertainty.19 Standardizing those rules through shared AI infrastructure surrenders that advantage. Behavioral resistance and trust dynamics. AI systems that monitor performance, surface biases, and recommend decisions also function as control mechanisms.15 Research documents that organizational members respond to algorithmic oversight by resisting, gaming inputs, manipulating digital footprints, or withdrawing altogether—feeding systems inaccurate data to manage impressions rather than improve decisions.26 In CVC, where relationships and intuition are deeply valued, the introduction of AI can trigger trust dynamics that undermine the very system it was designed to improve. Four distinct trust configurations—full trust, full distrust, uncomfortable trust, and blind trust—each produce different organizational pathologies. The productive configuration requires both cognitive and emotional trust; both must be earned. The illusion of precision. AI generates outputs that feel authoritative: scores, rankings, probability distributions. In venture investing, where outcomes are fundamentally uncertain, this precision is seductive and misleading. The danger is not wrong predictions. It is displaced attention. Teams that trust the model stop questioning the assumptions behind it. The model identifies which startups match the criteria. It cannot tell you whether the criteria are right. Five Questions Before You Begin Leaders weighing an AI-native redesign of their CVC unit should test their organization against five questions before committing. Is your scouting calibrated to your structure—or just your ambition? Most CVC mandates reflect where the firm wants to compete, not where its asset base allows it to absorb innovation. Before deploying AI-powered scouting, audit the gap between the domains you are scouting and the domains your business units can actually integrate. Capital-intensive firms that point AI toward distant frontiers without the flexibility to act will accumulate optionality they cannot exercise. Where does knowledge die in your process? Every CVC unit has handoff points where context evaporates—between scout and analyst, between diligence team and investment committee, between deal team and business unit, between portfolio board seat and the product managers who would benefit from what was learned. Map these breaks. They are where AI-native redesign creates the most value and where bolt-on tools most reliably fail. Can your investment committee work with evolving evidence? Parallel evaluation only works if decision-makers engage with living assessments rather than demanding a polished final deck. If your committee culture requires certainty before discussion, the organizational change must precede the technological one. What happens when AI and your best investor disagree? If you always defer to the human, AI adds no value. If you always defer to the algorithm, you have automated away the strategic judgment that justifies CVC’s existence. The productive answer is a protocol: stop, investigate the source of disagreement, and learn from it. Each disagreement is a data point about the boundary of the model’s competence. Are you willing to measure knowledge flow, not just deal flow? If CVC performance is evaluated solely on financial returns and deal volume, AI-native redesign will default to faster sourcing rather than deeper integration. The strategic potential of CVC—the reason a corporation maintains a venture arm rather than investing as a limited partner—lies in knowledge orchestration. Measure it. Track business-unit adoption of portfolio-company technology. Track cross-silo introductions that generated pilots. Track time-to-integration, not only time-to-term-sheet. Conclusion: Change the Organization, Not Just the Software AI will reshape corporate venture capital. The transformation, however, is widely misunderstood. The dominant narrative—that AI helps CVC units find better startups faster—confuses the symptom with the disease. The primary constraint in CVC has never been information access. It is the organization’s ability to align investments with strategy, move knowledge across boundaries, and make good decisions under uncertainty. AI amplifies these challenges before it solves them. It puts more information into a system that already struggles to act on what it has. It expands the scouting aperture for firms whose absorptive capacity has not changed. It produces analytical precision in a domain where the most important variables resist measurement. The firms that will benefit most are those that treat AI not as a scouting upgrade but as a reason to redesign the CVC operating model—from how opportunities are identified, to how decisions are structured, to how knowledge moves between the venture team and the rest of the organization. That redesign is organizational, not technological. It requires changing incentives, decision rights, team structures, and performance metrics before it requires changing software. The question is no longer whether to adopt AI in corporate venturing. It is whether leaders are willing to change the organization around it. References Henry Chesbrough, “Making Sense of Corporate Venture Capital,” Harvard Business Review, 80/3 (March 2002): 90–99. Global Corporate Venturing, Corporate Venturing Annual Data 2025 (2026); see also CB Insights, State of CVC Q1’25 (2025); and SVB (in partnership with Counterpart Ventures), State of Corporate Venture Capital 2025 (2025). Corporate-backed deal value rose from roughly $131 billion in 2024 to $229 billion in 2025, a 75 percent increase year on year. Bain & Company, “Global Venture Capital Outlook: The Latest Trends,” March 2026. Salesforce, “Salesforce Ventures Deploys Over $850M of Its $1B AI Fund to Back the AI Revolution,” June 2025. Intel Corporation, “Intel News: January 2025,” Intel Newsroom, January 14, 2025, (cumulative $20 billion invested across more than 1,800 companies); and Intel Capital, “Intel Capital’s Next Chapter: Fueling Tomorrow’s Tech,” January 14, 2025, (approximately 1,300 curated customer introductions in 2024). Intel announced on January 14, 2025 its intention to separate Intel Capital into a standalone fund, with operations formally independent in the second half of 2025. Tobias Gutmann, Christopher Chochoiek, and Henry Chesbrough, “Extending Open Innovation: Orchestrating Knowledge Flows from Corporate Venture Capital Investments,” California Management Review, 65/2 (2023): 45–70. Affinity, The 2025 Guide to Deal Management (2025) and V7 Labs, “5 Applications of AI in Venture Capital and Private Equity,” blog post (2025). The 82 percent figure refers to active AI use across PE and VC operations rather than to deal sourcing specifically. Andre Retterath, Data Driven VC Landscape 2025 (2025). Manuel Hess, Jana Reuther, Dietmar Grichnik, and Joakim Wincent, “Corporate Venture Capital and Strategic Search: Performance Spillovers in Parent Companies from Sticking to the Core Business,” Journal of Business Research, 200 (2025): 115628. Sergey Alexander Anokhin, Manuel Hess, and Joakim Wincent, “Technology Sourcing Ambidexterity in Corporate Venture Capital: Limitations of Learning from Open Innovation,” Small Business Economics, 64/1 (2025): 1–24. Louisa A. Müller, Marc Neubert, and Dominik K. Kanbach, “Gatekeepers of the Future: The Role of Corporate Venture Capital in Corporate Artificial Intelligence Adoptions,” Strategic Change, 35/1 (2026): e70061. Mats O. Pettersson, Johan Björkdahl, and Marcus Holgersson, “Profiting from AI: Evidence from Ericsson’s Pursuit to Capture Value,” California Management Review, 67/4 (2025): 5–20. World Economic Forum (in collaboration with Accenture), Organizational Transformation in the Age of AI: How Organizations Maximize AI’s Potential, March 2026. Dietmar Grichnik, Manuel Hess, Jana Reuther, Alexander Stoeckel, and Michael Hilb, The Corporate Venturing Handbook: A Step-by-Step Guide to the Value Creation Process (London: Kogan Page, 2024). Luis Hillebrand, Sebastian Raisch, and Jonathan Schad, “Managing with Artificial Intelligence: An Integrative Framework,” Academy of Management Annals, 19/1 (2025): 343–375. Global Corporate Venturing, “AI Tools Are Transforming VC Investment—But CVCs Are Behind the Curve,” March 2025. Mohammad Hossein Jarrahi and Paavo Ritala, “Rethinking AI Agents: A Principal-Agent Perspective,” California Management Review Insights, July 2025. Yash Raj Shrestha, Shiko M. Ben-Menahem, and Georg von Krogh, “Organizational Decision-Making Structures in the Age of Artificial Intelligence,” California Management Review, 61/4 (2019): 66–83. Natalia Vuori, Barbara Burkhard, Tomi Laamanen, and Christopher B. Bingham, “Heuristics in Organizations: Toward an Integrative Process Model,” Academy of Management Annals, 18/2 (2024): 670–711. Rebecka C. Ångström, Michael Björn, Linus Dahlander, Magnus Mähring, and Martin W. Wallin, “Getting AI Implementation Right: Insights from a Global Survey,” California Management Review, 66/1 (2023): 5–22. Sebastian Raisch and Sebastian Krakowski, “Artificial Intelligence and Management: The Automation–Augmentation Paradox,” Academy of Management Review, 46/1 (2021): 192–210. Fabrizio Dell’Acqua, Edward McFowland III, Ethan R. Mollick, Hila Lifshitz-Assaf, Katherine Kellogg, Saran Rajendran, Lisa Krayer, François Candelon, and Karim R. Lakhani, “Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality,” Harvard Business School Working Paper No. 24-013 (2023); a peer-reviewed version appeared in Organization Science in 2026. Vegard Kolbjørnsrud, “Designing the Intelligent Organization: Six Principles for Human–AI Collaboration,” California Management Review, 66/2 (2024): 44–64. World Economic Forum (2026); see especially the principles on scalable talent systems and human accountability at scale. On convergence to common models and concentration risk, see Bank of England, Financial Stability in Focus: Artificial Intelligence in the Financial System, April 2025. On related but distinct concerns, see European Central Bank, “The Rise of Artificial Intelligence: Benefits and Risks for Financial Stability,” Financial Stability Review, May 2024 (emphasizing the propagation of pre-trained biases across firms that adopt similar foundation models); and Financial Stability Board, “The Financial Stability Implications of Artificial Intelligence,” November 2024 (emphasizing model risk and reduced explainability when AI is embedded in financial models). Natalia Vuori, Barbara Burkhard, and Leena Pitkäranta, “It’s Amazing—But Terrifying!: Unveiling the Combined Effect of Emotional and Cognitive Trust on Organizational Members’ Behaviours, AI Performance, and Adoption,” Journal of Management Studies, 63/2 (2026): 473–514. Brian R. Spisak and Gary Marcus, “Cutting Through the AI Hype: The Facts Leaders Need to Know About GenAI Adoption and Return on Investment,” California Management Review Insights, June 2025. Acknowledgements The authors used AI for editorial assistance in refining parts of this manuscript. All AI-assisted content was reviewed, edited, and approved by the authors, who take full responsibility for the accuracy, integrity, and originality of the work. No AI tool was used in the underlying research, data collection, analysis, or substantive development of the arguments presented.

Most companies treat organic waste as a cost to be managed. Denmark treats it as an infrastructure input worthy of institutional capital. The difference is not technology or policy, it is framing. This piece draws on firsthand observations from Denmark and scholarship on Nordic sustainability to offer two reframes that practicing managers can apply today. Of all the iconic landmarks in Copenhagen’s skyline, CopenHill stands out as unlike anything else. Located near the harbor, a steady plume of steam rises from an architectural marvel designed by Bjarke Ingels as an ode to “hedonistic sustainability,” the idea that green infrastructure doesn’t have to be invisible or utilitarian, but can be joyful and functional for the public at the same time. CopenHill is a waste-to-energy power plant that burns waste to generate electricity and heat for roughly 150,000 homes, while also serving as a public recreation space with a ski slope, walking trails, and a climbing wall. I arrived in Copenhagen carrying a specific assumption: that the reason organic waste remained stranded in so many markets, unmonetized, underleveraged, treated as a disposal problem, was fundamentally a policy failure. Pass the right incentives, mandate the right targets, and the market would follow. My background living and working in both the United States and Kenya had given me a front-row seat to the same gap in two very different contexts: waste streams with real value, no financing mechanism, and no political will to catalyze the market. I had seen biogas work at the household scale, powering a few lights or a cookstove, but never at the industrial scale needed to serve urban or commercial energy demand. What I found in Denmark dismantled that assumption. The gap wasn’t policy. It was framing. The System Nobody Designed CopenHill is not an anomaly. It is a visible expression of a design philosophy that runs through Danish industrial history. In the 1970s, a cluster of companies near the town of Kalundborg began informally exchanging waste streams.1 A power plant’s excess heat warmed a pharmaceutical manufacturer’s facilities, steam that would have been vented was piped to a refinery, sludge from one facility became fertilizer for local farms. None of this was planned by a government agency but rather began as practical barter among neighbors and evolved into what is now called industrial symbiosis: a system where one firm’s waste becomes another firm’s input. Kalundborg Symbiosis, as it is formally known, has since become a global reference point for circular economy design, a framework built on the principle that outputs should become inputs, and that waste is a design flaw, not an inevitability. The firms that built it didn’t wait for Denmark to pass an industrial symbiosis policy. They started because it was economically rational at the bilateral level, then built the institutional infrastructure to make it systemic. This is the reframe that changed my thinking. I had been looking for policy as the precondition, but Kalundborg showed me that viability comes first, and policy tends to follow the actors who demonstrate it. When Waste Becomes an Asset Class The most clarifying moment of my time in Denmark was a visit to Copenhagen Infrastructure Partners (CIP), one of the world’s largest dedicated renewable energy fund managers. CIP manages an Advanced Bioenergy Fund, a dedicated infrastructure investment vehicle targeting facilities that convert agricultural and municipal waste into advanced biofuels and biomethane. The waste streams that most companies manage as disposal costs are the feedstocks that institutional investors are now building multi-billion dollar asset classes around. The technology including anaerobic digestion, biomethane upgrading, advanced biofuel production is commercially proven. What Nordic actors have done is reframe agricultural and industrial biowaste as infrastructure inputs worthy of long-term institutional capital. This is what Strand describes as the “Nordic cooperative advantage,” the tendency of Nordic firms to build value through stakeholder cooperation rather than despite it, treating sustainability not as a constraint on commercial success but as its foundation.2 They have built systems that make sustainability the commercially rational path. The Nordic model is not a set of policies layered on top of a market economy; it is a different architecture for how companies, investors, and governments relate to each other over time.3 What Leaders Can Do Without Waiting In the U.S., investors are waiting for policy certainty before committing capital at scale. Federal momentum on clean energy has become unreliable, and uncertainty raises the cost of capital. But the Nordic evidence suggests that waiting carries its own strategic cost. The actors that move first shape the policy environment that follows. Firms that treat policy as a prerequisite for action are effectively delegating their strategic agenda to regulators. Two actions follow directly from this: Revalue waste streams as feedstock inventories. Agricultural and food processing operations generate enormous volumes of organic material accounted for as costs. A feedstock inventory, valued at what bioenergy producers would actually pay for it, changes the conversation from managing a liability to selling an asset. Infrastructure investors speak in the language of long-duration assets, contracted offtake, and return profiles. Translating circular economy proposals into that language unlocks capital that sustainability-framed pitches rarely reach. Start the symbiosis conversation before the system exists. Kalundborg was not designed top-down, it started as conversations between neighbors. Recent research on circular value propositions confirms that the companies most likely to succeed are those that articulate a compelling internal rationale first, then reach out to ecosystem partners before formal structures exist.4 US industrial corridors where agriculture, food processing, energy generation, and municipal waste management exist in close proximity already have the preconditions for similar networks. My own work on California’s bioeconomy has reinforced that the feedstocks, technology, and financing models exist. What is missing is the conversation. A senior leader at an agricultural processor or energy company has the standing to initiate that dialogue today without a policy mandate, and without waiting. CopenHill works because it made something invisible impossible to ignore: tons of waste, converted into heat and light and recreation, rising above a city that chose to see its infrastructure differently. The same reframing is available in every market where organic waste is treated as a problem to be managed rather than a resource to be deployed. The technology exists. The capital exists. What has to change first is how you look at it. References European Circular Economy Stakeholder Platform, Kalundborg Symbiosis: Six decades of a circular approach to production, (2019) European Union. Robert Strand, “Global Sustainability Frontrunners: Lessons from the Nordics,” California Management Review 66, no. 3 (2024): 5–26. Robert G. Strand, Nordic Capitalism: Lessons for Realizing Sustainable Capitalism (Cambridge: Cambridge University Press, 2026). Mattias Axelson, Johan Frishammar, and Lars Nybom, “Going Circular: What Enables Companies to Successfully Move from Ideas to New Circular Value Propositions?” California Management Review Insights, March 12, 2026.