A structural analysis of why partnership governance models uniquely constrain enterprise AI deployment, and what changes when firms consolidate from federated to unified profit structures.
The global professional services market exceeds $1.1 trillion in legal services alone, with the combined legal, consulting, and accounting sectors representing one of the largest concentrations of proprietary expertise in the global economy. These firms are simultaneously among the most sophisticated producers of knowledge work and among the slowest to deploy AI at enterprise scale. The gap between experimentation and operational deployment is larger in partnership-structured firms than in any other organizational category.
The binding constraint is governance. Generative AI can draft contracts, summarize case law, analyze financial statements, and identify regulatory risk patterns today. The Thomson Reuters 2024 Future of Professionals survey found that 73% of law firms were actively experimenting with AI tools. Yet the same research identified that only 12% had achieved firm-wide deployment. The gap between experimentation and deployment is a governance gap—the structural inability of partnership organizations to authorize, fund, and sustain shared technology investment across autonomous profit centers.
Partnership structures create a governance problem that corporate enterprises do not face. Partners are autonomous revenue generators whose compensation is tied to individual or practice-level profitability. Shared AI investment requires pooled capital from people who have no structural obligation to subsidize technology infrastructure that benefits another practice group. The free-rider problem, well documented in commons economics, applies with particular force to knowledge-work organizations where the “commons” is a shared AI capability and the “riders” are equity partners with the mobility to leave if the cost allocation feels inequitable.
This brief extends the analysis in Enterprise AI Investment 2026 Outlook (RB-AI), which identified the 93/7 spending inversion—93% of enterprise AI budgets allocated to technology acquisition, 7% to organizational capacity to absorb it. In partnership structures, the inversion is even more extreme: technology investment concentrates in firm-level IT budgets, while the organizational capacity to deploy AI—practice-level workflow redesign, partner adoption, client-facing use case development—receives almost no structural funding.
In a corporate enterprise, the CEO can mandate technology investment. The board approves capital allocation. The CIO or CTO executes. The authority chain is vertical, and dissenting business unit leaders can be overruled or replaced. Partnerships possess no equivalent mechanism. The managing partner leads by consensus, not by fiat. Technology investment above a threshold—typically defined in the partnership agreement—requires a partner vote. And the partners who must approve that investment are the same people whose individual profitability will be reduced by the cost allocation.
Three structural features of partnership governance uniquely constrain AI deployment. First, profit distribution is tied to individual or practice-level performance. Partners in most professional services firms are compensated through a combination of base draws, performance-based distributions, and equity stakes. The Am Law 100 data consistently shows that partner compensation is the single largest expense category for law firms, typically consuming 35–45% of gross revenue. Any shared technology investment that reduces distributable profit faces direct opposition from the compensation mechanism itself.
Second, partner vote requirements create a collective action problem. Major capital expenditures in most partnerships require approval by a supermajority of equity partners. Each partner evaluates the investment against their own practice economics, not the firm’s aggregate return. An AI system that transforms contract review delivers enormous value to the litigation and transactional practices but near-zero value to the tax or regulatory advisory groups. The partners who receive no direct benefit have a rational incentive to vote against the investment, or to demand that costs be allocated only to the practices that benefit—which fragments the shared infrastructure required for AI to function at scale.
Third, lateral partner mobility functions as a structural constraint on investment policy. Partners in professional services firms are not employees. They are mobile capital. If a firm’s cost allocation feels inequitable, high-performing partners can move to competitors—and take their client relationships with them. ALM data shows that lateral partner movement in the Am Law 200 has averaged over 3,000 moves per year since 2019. This mobility creates an implicit upper bound on shared investment: any cost allocation that pushes profitability per partner below competitive benchmarks risks triggering departures that unravel the economics of the investment itself.
The free-rider problem manifests with particular clarity in AI investment. AI infrastructure—training data curation, model fine-tuning, integration with practice management systems, prompt engineering for domain-specific applications—is inherently shared. The value compounds across practices: a document analysis capability trained on one practice’s work product becomes more capable when it ingests another’s. But the cost must be allocated somewhere. And in a partnership, “somewhere” means specific partners whose distributable income is reduced. The economic logic of AI demands unified investment. The governance logic of partnerships demands distributed authority over spending. These logics are in structural opposition.
If partnership governance creates a collective action problem for AI investment, federated partnership structures compound it. The verein—a Swiss-law association that links legally separate national partnerships under a single brand—has been the dominant organizational form for the world’s largest law firms for two decades. Firms including Baker McKenzie, DLA Piper, Norton Rose Fulbright, and Dentons have used verein structures to achieve global scale without requiring profit-sharing across borders. Each national member firm maintains its own profit pool, its own partnership agreement, and its own technology infrastructure.
The implications for AI deployment are severe. A verein is a collection of independent firms that share a brand and a referral network—separate legal entities with separate profit pools, separate data, and separate regulatory obligations. The London partnership and the New York partnership are distinct organizations in legal fact. An AI system trained on London client data is legally inaccessible to New York—constrained by GDPR, attorney-client privilege doctrines that vary by jurisdiction, and conflict-of-interest rules that prevent commingling of client information across entities that are, structurally, separate firms.
Baker McKenzie’s 2024 decision to dissolve its verein structure and consolidate into a single global partnership is the definitive precedent case. The firm, with approximately 4,700 lawyers across 74 offices in 45 countries, announced the dissolution following a vote of its global partnership. The stated rationale included the need for coordinated technology investment, unified data governance, and a single profit pool that could fund enterprise-scale capabilities. The subtext was explicit: the verein structure had become an impediment to the kind of integrated technology deployment that a $3.3 billion-revenue global firm required to remain competitive.
The Baker McKenzie restructuring is instructive precisely because it was not driven by financial distress. It was driven by the recognition that federated governance and enterprise-scale AI are structurally incompatible. When your AI strategy requires unified training data, consistent model deployment, and coordinated workflow redesign, and your governance model prohibits data sharing across entities, prevents coordinated capital allocation, and maintains separate technology stacks by jurisdiction, the governance model must change before the technology strategy can execute.
Attorney-client privilege, GDPR data residency requirements, and conflict-of-interest rules create jurisdictional data silos that no amount of technology can bridge without governance reform. In a verein structure, client data from the London office may be legally inaccessible to an AI system operated by the New York office. Even within a unified firm, cross-border data use for AI training requires jurisdiction-by-jurisdiction legal analysis. Professional services firms face data governance constraints that technology companies, manufacturers, and financial institutions do not—because the data itself is the client relationship, and the client relationship is the asset that generates partner compensation.
The contradiction can be stated precisely. AI at enterprise scale requires three things: unified data (to train and fine-tune models on the full breadth of firm expertise), shared infrastructure (to deploy, maintain, and govern AI systems consistently), and coordinated investment (to fund capabilities that benefit the firm as a whole, not individual practices in isolation). Partnership governance requires three different things: autonomy (partners control their practice economics), distributed authority (major decisions require collective approval), and practice-level accountability (compensation reflects individual contribution, not firm-level outcomes).
These requirements are not merely in tension. They are structurally opposed. Every AI deployment decision in a partnership is simultaneously a governance decision about who pays, who benefits, and who has authority over shared resources. When a firm deploys an AI-powered contract review system, the decision is not “which AI tool should we use?” The decision is: Who funds the implementation? How is the cost allocated across practices? Who governs the training data? Which partners must adopt the system? What happens to the associates whose work the system partially automates? And who has authority to make these decisions in an organization where authority is, by design, distributed?
McKinsey’s 2024 research on generative AI in professional services estimated that 20–30% of billable hours in legal, consulting, and accounting could be augmented or automated by current AI capabilities. But “could be” and “will be” are separated by the governance gap. The technology exists. The business case is compelling. The constraint is the organizational structure through which the investment decision must pass—and in partnerships, that structure was designed to preserve autonomy, not to enable coordinated investment.
The Thomson Reuters finding that 73% of firms experiment while only 12% deploy firm-wide reflects a governance outcome, not an adoption lag. Experimentation requires one partner’s initiative and a modest budget. Firm-wide deployment requires collective agreement on cost allocation, data governance, workflow redesign, and change management—precisely the coordinated decisions that partnership structures are designed to avoid.
If AI requires shared infrastructure and partnerships require distributed authority, what governance model allows both?
No single research program connects partnership governance to AI deployment failure. Technology analysts study adoption curves. Governance scholars study partnership structures. Management consultants study organizational transformation. But when these three independent evidence streams are placed in sequence, a convergence pattern emerges that none of them state individually.
Thomson Reuters’ 2024 Future of Professionals survey established the 73%/12% gap: near-universal experimentation, minimal firm-wide deployment. Gartner’s AI governance research identifies the absence of centralized decision-making authority as the primary barrier to enterprise AI scaling. The ALM/Am Law Technology Survey shows that legal technology spending averages 3–7% of firm revenue—substantially below the 8–12% technology-to-revenue ratio typical of corporate enterprises. The data shows that professional services firms invest less in technology, deploy less at scale, and have a wider gap between experimentation and operational use than any comparable industry segment.
The academic literature on partnership governance, spanning Oxford’s Saïd Business School research on professional service firms to Gilson and Mnookin’s foundational work on law firm economics at Harvard Law Review, identifies the same structural features this brief describes: distributed authority, compensation tied to individual performance, and partner mobility as the enforcement mechanism. McKinsey’s Professional Services practice reports that 62% of partners in surveyed firms cite governance complexity as the primary barrier to AI investment—not technology readiness, not cost, not talent availability. The governance literature explains WHY the deployment gap exists: partnership structures distribute authority in ways that prevent the coordinated investment AI requires.
Baker McKenzie’s 2024 verein dissolution is the most significant governance restructuring in global professional services in a decade. The firm explicitly cited technology investment coordination as a driver. Deloitte’s separation from its consulting practice (completed 2025 in several jurisdictions), PwC’s ongoing structural consolidation, and EY’s abandoned but instructive “Project Everest” split all reflect the same underlying dynamic: firms are restructuring governance to enable coordinated technology investment. The restructuring evidence shows WHAT changes when governance unifies: the structural barrier to coordinated AI investment is removed, but new governance challenges emerge around centralized authority in a culture built on partner autonomy.
The convergence insight: The 12% firm-wide deployment rate is primarily a governance outcome, not a technology limitation. Partnership structures distribute authority in ways that prevent coordinated AI investment. Firms that unify their profit pool remove the structural barrier—but create new governance challenges around centralized technology authority in a culture built on partner autonomy. The question is not “which AI to deploy” but “who authorizes shared investment when no one is structurally obligated to share.”
Synthesizing the deployment data with governance research and firm restructuring evidence, five distinct models emerge for how partnerships are attempting to resolve the structural contradiction between AI’s requirements and partnership governance. Each model reflects a different trade-off between investment coordination and partner autonomy.
The following diagnostic helps technology leaders in partnership-structured firms assess whether their governance model is prepared to support enterprise AI deployment. It is organized around the Four Capability Bands from The Intelligence Organization™, applied to the specific challenge of AI investment governance in distributed-authority organizations.
| Capability Band | Governance Readiness Question | Score 1–5 | If Score <3 |
|---|---|---|---|
| Band 1: Right-Fit Technology | Is there a defined governance mechanism for approving firm-wide AI investment that does not require a full partner vote for each initiative? | ___ | Establish a technology investment committee with delegated authority below a defined threshold. Partner votes for every AI initiative create a structural veto that prevents enterprise deployment. |
| Has the firm completed a jurisdiction-by-jurisdiction data governance analysis that identifies which client data can legally be used for AI training and deployment? | ___ | Commission a cross-jurisdictional data governance review before any enterprise AI system ingests client data. Privilege, GDPR, and conflict-of-interest rules vary by jurisdiction and cannot be assumed. | |
| Band 2: People & Purpose | Is AI investment cost allocation mapped to the existing partner compensation structure in a way that partners perceive as fair relative to benefit received? | ___ | Redesign the AI cost allocation model to mirror the firm’s compensation logic. If partners cannot see the connection between what they pay and what they receive, resistance is rational, not irrational. |
| Has the firm defined how AI deployment will affect associate staffing ratios, partner-to-associate billing dynamics, and the career development model? | ___ | AI that automates associate-level work without a corresponding workforce model adjustment will face adoption resistance from partners who depend on associate staffing ratios for profitability. | |
| Band 3: Operational Integration | Are AI deployment plans structured as practice-level workflow redesigns with measurable impact metrics, or as technology rollouts with adoption targets? | ___ | Reframe AI deployment as workflow transformation, not technology adoption. Partners respond to practice economics, not technology features. Define success in terms of realization rates, cycle times, and client satisfaction. |
| Does the firm have a mechanism for sharing AI-generated insights across practices without violating client confidentiality or conflict rules? | ___ | Cross-practice AI value requires cross-practice data governance. Without a mechanism for ethical knowledge sharing, AI capabilities remain siloed in the same practice structure that created the governance problem. | |
| Band 4: Adaptive Governance | Is there a single person or body with authority to make binding AI governance decisions (data use, model deployment, ethical guidelines) without requiring consensus of all equity partners? | ___ | Appoint or create a technology governance authority with defined scope and delegated decision rights. Consensus governance and enterprise AI deployment are structurally incompatible. |
| Has the partnership agreement been reviewed and, if necessary, amended to accommodate centralized technology investment authority and cross-practice data governance? | ___ | If the partnership agreement does not contemplate centralized technology investment, any AI strategy that requires it operates on informal authority that can be challenged or revoked by partner vote. |
Total 32–40: Unified Authority readiness. The firm’s governance structure can support enterprise AI deployment. The priority shifts from governance reform to execution: technology selection, workflow redesign, and change management. Focus on the Hybrid Investment Model or Unified Authority Model.
Total 20–31: Hybrid readiness. Some governance mechanisms exist, but gaps remain in cost allocation fairness, data governance, or decision authority. Identify the lowest-scoring Band and address it before scaling AI beyond pilot programs. The Firm Assessment Model or Hybrid Investment Model is the appropriate governance target.
Total 8–19: Practice-Led default. The firm lacks the governance infrastructure to deploy AI beyond individual practice initiatives. Experimentation will continue; enterprise deployment will not occur until governance reform addresses investment authority and data governance. Prioritize Band 4 (Adaptive Governance) immediately—without centralized decision authority, all other capability development is structurally blocked.
Firms operating under verein or other federated structures face an additional diagnostic layer. The questions above assume a single legal entity. In a federated structure, each question must be answered separately for each member firm—and then a meta-governance question must be addressed: is there a mechanism for coordinating AI governance across member firms? If the answer is no, the firm faces the Baker McKenzie dilemma: the technology strategy requires governance unification, and governance unification requires structural reform that may take years to negotiate and execute.
Decision support aligned with The Intelligence Organization · Band 4 (Adaptive Governance) as the primary lever for partnership AI readiness · Governance authority as the structural determinant of AI deployment capabilityThis research informs our partnership governance advisory practice. If your firm is navigating the transition from experimentation to enterprise AI deployment, we should talk.
Schedule a ConversationAI Deployment in Professional Services: Thomson Reuters, "Future of Professionals: Harnessing AI to Reimagine Professional Services," 2024, surveying 1,200+ professionals across legal, tax, accounting, and compliance, finding 73% experimentation and 12% firm-wide deployment rates. Gartner, "AI Governance Frameworks for Enterprise Deployment," 2024, identifying centralized decision-making authority as the primary barrier to enterprise AI scaling. McKinsey & Company, "The Economic Potential of Generative AI: The Next Productivity Frontier," 2024, estimating 20–30% of professional services billable hours augmentable by current AI capabilities. McKinsey & Company, "Generative AI and the Future of Work in Professional Services," 2024, surveying partner attitudes toward AI governance.
Legal Market and Technology Spending: Thomson Reuters, "2024 Report on the State of the Legal Market," Georgetown Law Center for the Study of the Legal Profession and Thomson Reuters Institute, reporting $1.1T+ global legal services market. ALM/Am Law, "Technology Survey and Am Law 100/200 Financial Rankings," 2024, documenting 3–7% technology-to-revenue spending ratios and lateral partner movement data. NALP (National Association for Law Placement), associate survey data on firm selection factors, 2025.
Partnership Governance Research: Gilson, R.J. and Mnookin, R.H., "Sharing Among the Human Capitalists: An Economic Inquiry into the Corporate Law Firm and How Partners Split Profits," Stanford Law Review, 37(2), 1985. Empson, L., "Leading Professionals: Power, Politics, and Prima Donnas," Oxford University Press, 2017, examining governance challenges in partnership-structured professional services. Saïd Business School, University of Oxford, research program on professional service firm governance and organizational design.
Firm Restructuring Evidence: Baker McKenzie, announcement of verein dissolution and global partnership unification, 2024, covering 4,700 lawyers across 74 offices in 45 countries. Reuters, "Baker McKenzie Votes to Unify Global Partnership," 2024. The American Lawyer, coverage of Big Four structural consolidation and professional services governance reform, 2024–2025. Ernst & Young, "Project Everest" consulting/audit separation analysis, 2023 (abandoned).
RBD. Research: Starkey, M.C., The Intelligence Organization, 2026. RBD., "Enterprise AI Investment 2026 Outlook: From Technology-First Budgets to Capability-First Returns," RB-AI, Q2 2026. RBD. cross-industry partnership governance and AI deployment synthesis, 2026.