Shared Services Still Matter in an AI World

While AI won’t replace shared services, it will redefine how they're sequenced and scaled.

著者：Mark Krafft, Evren Sungur, and Praveen Bhardwaj

min read


At A Glance AI raises the ambition for support functions, but scale and standardization still matter. For many enterprises, shared services are the fastest path to the process and data foundations AI requires while accelerating labor arbitrage and process improvement. The right answer to shift first or AI first depends on process-level economics, AI maturity, data readiness, and capacity. The bigger issue isn't whether AI replaces shared services but how they will evolve from a transaction factory into an AI deployment and capability hub. Artificial intelligence has moved from experimentation to executive mandate. AI agents and tokens could represent 20% to 30% of operating expenses by 2028–2029 for AI pioneers vs. just 1% to 2% today. Boards and leadership teams are asking how quickly AI can transform enterprise cost and productivity. This issue is especially urgent in support functions such as finance, HR, and IT, where companies have spent two decades using a proven playbook: centralize, standardize, and offshore. Mature shared services organizations have reduced costs by 20% to 40% while improving quality and control. AI is challenging the traditional shared-services playbook: If technology can eliminate work altogether, why move it into shared services first? The answer is more nuanced than the debate suggests. For many large enterprises, especially those early in their global business services (GBS) journey, shared services are not a detour from AI transformation. They often are the fastest path to the process discipline, data foundations, scale, and funding that AI requires. The real question is not whether AI replaces shared services; it is how companies should sequence shared services, offshoring, process redesign, and AI, and how shared services must evolve in an AI-enabled enterprise. AI ambition meets operational reality Over the past decade, robotic process automation and workflow digitization produced meaningful cost savings, with returns strongest when paired with process redesign and centralization. AI goes further, introducing decision support, pattern recognition, and generative capabilities. Roughly a quarter of CFOs surveyed by Bain expect AI to deliver more than 35% cost savings plus quality benefits. But realizing that potential requires the right foundations. Compared with traditional automation, AI initiatives typically: Require greater upfront investment in infrastructure, governance, and data engineering Depend on clean, structured, and accessible data foundations that many enterprises lack Carry greater model risk, explainability requirements, and regulatory scrutiny Face more uncertain and often longer payback periods As a result, only 15% to 25% of finance organizations have scaled machine learning, generative AI, or agentic AI. The limiting factor isn’t the technology; it’s the process and data foundations AI requires. Why scale still drives ROI in the AI era Despite the technological shift, one structural truth remains: Scale improves economics. Fragmented processes across regions, business units, and systems dilute improvement potential and digital investment returns. When organizations centralize work into shared services, often coupled with offshoring, they create conditions that make both process improvement and AI more viable: Standardized workflows reduce AI model complexity. Consolidated data improves training quality and model accuracy. Concentrated volume increases the economic value of automation. Dedicated teams build repeatable, scalable digital capabilities. Savings from centralization and offshoring help fund the next wave of investment. For example, a global consumer packaged goods company tried for years to standardize its monthly volume and revenue forecast process. Once these activities were moved into shared services, the company deployed a machine-learning solution that reduced forecast creation from two weeks to several hours and improved accuracy. Shared services created the foundation; AI delivered the ROI. Siemens offers another instance. After two decades building a mature, standardized, globally centralized GBS, the company had the foundation to deploy AI and intelligent automation at scale. Its journey reinforces the sequencing point: Technology delivers more value when the foundation is strong. SSON Research & Analytics recognized Siemens Global Business Services as the World's Best GBS 2026. The sequencing dilemma: AI first or shift first? Conventional wisdom cautions "Don't outsource your mess for less." In theory, companies should standardize first, then move the work. In practice, fixing processes in a decentralized structure is slow and politically complex. Without central authority, standardization stalls, and benefits dissipate. Shifting first to shared services can unlock momentum faster. Immediate savings create economic headroom, central visibility exposes variation, and governance becomes clearer. Critically, the move builds the process discipline and data foundations that AI needs. For many companies, shared services are often the fastest path to AI readiness. AI adds a new dimension to this debate. Many executives now argue for an AI-first approach by eliminating work with technology before moving it. But several factors complicate that strategy: AI initiatives are still difficult to scale. Bain's CFO survey found under-delivery in roughly 25% of companies. Organizations rarely have the capacity to execute structural reorganization and enterprise AI deployment at the same time. Back-office use cases frequently lose the competition for AI funding and talent against revenue-generating initiatives. In digital-native or mature companies with standardized environments, strong AI capability, and robust data foundations, automation first can make sense. For most enterprises, the question is not whether to skip the foundation but how to build it fast and use it as the shortest runway to an AI-native operating model. A five-factor framework for sequencing decisions Leaders should apply a nuanced sequencing strategy, evaluating each major process area through five lenses. Different profiles across these factors will lead to different sequencing answers, including cases in which AI first is clearly the right call. A large healthcare services company learned this lesson firsthand. Concerned that offshoring could slow AI momentum or cede too much value to outsourcers, the company initially prioritized AI across many major processes. While some areas advanced, others saw far less progress. Eager to realize value quickly, leadership adopted a more nuanced approach: standardizing and offshoring the areas where process and data foundations were not yet ready, using labor arbitrage savings to fund the broader transformation, and accelerating AI in areas where the foundations were already in place. It was not a choice between offshoring and AI but a disciplined decision about where each approach applied. Over time, AI will become more plug and play through lower costs, pretrained models embedded in enterprise resource planning and human capital management platforms, and demonstrated large-scale ROI. This will shift the equation. But that inflection point is uneven across processes and industries, and leaders who assume it has arrived risk forgoing tangible structural gains. For now, the key is rigorous, process-level evaluation. What shared services become in an AI enterprise Some argue that AI will make shared services obsolete. But we think AI will redefine shared services, not replace it. In the early stages of AI adoption, shared services help scale AI deployment. As AI agents execute more end-to-end processes, the human role shifts from task execution to orchestrating intelligent systems, managing exceptions, and driving continuous improvement. At scale, shared services become an enterprise capability hub for AI, trusted data products, and digital expertise. Shared services organizations that cling to legacy expectations on role, value add, and operating models will find themselves caught between cost pressure from AI-native alternatives and demands for strategic value from business partners. What leaders can do now Sequence at the process level, and build the digital foundations to support it. Evaluate labor arbitrage potential, AI maturity, process stability, and data readiness by function. And invest in data governance, process ownership, and digital talent regardless of where the work sits. Fund AI through structural moves. Shared services and offshoring can generate the capital required for sustainable AI investment. Be rigorous about AI ROI. Distinguish between pilots and enterprise-scale impact. Many AI initiatives succeed in controlled environments but stall at scale. Focus investment on use cases with proven, hard-dollar value. Plan future-back. Knowing that AI will become a high-ROI option within five years, ask yourself what building blocks do you need to lay today. AI builds on structure, not around it AI will reshape shared services, but it does not eliminate the value of structural discipline; it amplifies it. The same elements that made shared services effective—namely, scale, standardization, process ownership, and data governance—are increasingly the conditions that allow AI to deliver meaningful value. For most large enterprises, the path forward is not choosing between shared services and AI; it is sequencing them thoughtfully. Shared services build the operational foundation; AI builds on top of it. Companies that succeed know when to shift, when to standardize, and when to automate.

At A Glance

AI raises the ambition for support functions, but scale and standardization still matter.
For many enterprises, shared services are the fastest path to the process and data foundations AI requires while accelerating labor arbitrage and process improvement.
The right answer to shift first or AI first depends on process-level economics, AI maturity, data readiness, and capacity.
The bigger issue isn't whether AI replaces shared services but how they will evolve from a transaction factory into an AI deployment and capability hub.

Artificial intelligence has moved from experimentation to executive mandate. AI agents and tokens could represent 20% to 30% of operating expenses by 2028–2029 for AI pioneers vs. just 1% to 2% today. Boards and leadership teams are asking how quickly AI can transform enterprise cost and productivity. This issue is especially urgent in support functions such as finance, HR, and IT, where companies have spent two decades using a proven playbook: centralize, standardize, and offshore. Mature shared services organizations have reduced costs by 20% to 40% while improving quality and control.

AI is challenging the traditional shared-services playbook: If technology can eliminate work altogether, why move it into shared services first?

The answer is more nuanced than the debate suggests. For many large enterprises, especially those early in their global business services (GBS) journey, shared services are not a detour from AI transformation. They often are the fastest path to the process discipline, data foundations, scale, and funding that AI requires. The real question is not whether AI replaces shared services; it is how companies should sequence shared services, offshoring, process redesign, and AI, and how shared services must evolve in an AI-enabled enterprise.

AI ambition meets operational reality

Over the past decade, robotic process automation and workflow digitization produced meaningful cost savings, with returns strongest when paired with process redesign and centralization. AI goes further, introducing decision support, pattern recognition, and generative capabilities. Roughly a quarter of CFOs surveyed by Bain expect AI to deliver more than 35% cost savings plus quality benefits.

But realizing that potential requires the right foundations. Compared with traditional automation, AI initiatives typically:

Require greater upfront investment in infrastructure, governance, and data engineering
Depend on clean, structured, and accessible data foundations that many enterprises lack
Carry greater model risk, explainability requirements, and regulatory scrutiny
Face more uncertain and often longer payback periods

As a result, only 15% to 25% of finance organizations have scaled machine learning, generative AI, or agentic AI. The limiting factor isn’t the technology; it’s the process and data foundations AI requires.

Why scale still drives ROI in the AI era

Despite the technological shift, one structural truth remains: Scale improves economics. Fragmented processes across regions, business units, and systems dilute improvement potential and digital investment returns. When organizations centralize work into shared services, often coupled with offshoring, they create conditions that make both process improvement and AI more viable:

Standardized workflows reduce AI model complexity.
Consolidated data improves training quality and model accuracy.
Concentrated volume increases the economic value of automation.
Dedicated teams build repeatable, scalable digital capabilities.
Savings from centralization and offshoring help fund the next wave of investment.

For example, a global consumer packaged goods company tried for years to standardize its monthly volume and revenue forecast process. Once these activities were moved into shared services, the company deployed a machine-learning solution that reduced forecast creation from two weeks to several hours and improved accuracy. Shared services created the foundation; AI delivered the ROI.

Siemens offers another instance. After two decades building a mature, standardized, globally centralized GBS, the company had the foundation to deploy AI and intelligent automation at scale. Its journey reinforces the sequencing point: Technology delivers more value when the foundation is strong. SSON Research & Analytics recognized Siemens Global Business Services as the World's Best GBS 2026.

The sequencing dilemma: AI first or shift first?

Conventional wisdom cautions "Don't outsource your mess for less." In theory, companies should standardize first, then move the work. In practice, fixing processes in a decentralized structure is slow and politically complex. Without central authority, standardization stalls, and benefits dissipate.

Shifting first to shared services can unlock momentum faster. Immediate savings create economic headroom, central visibility exposes variation, and governance becomes clearer. Critically, the move builds the process discipline and data foundations that AI needs. For many companies, shared services are often the fastest path to AI readiness.

AI adds a new dimension to this debate. Many executives now argue for an AI-first approach by eliminating work with technology before moving it. But several factors complicate that strategy:

AI initiatives are still difficult to scale. Bain's CFO survey found under-delivery in roughly 25% of companies.
Organizations rarely have the capacity to execute structural reorganization and enterprise AI deployment at the same time.
Back-office use cases frequently lose the competition for AI funding and talent against revenue-generating initiatives.

In digital-native or mature companies with standardized environments, strong AI capability, and robust data foundations, automation first can make sense. For most enterprises, the question is not whether to skip the foundation but how to build it fast and use it as the shortest runway to an AI-native operating model.

A five-factor framework for sequencing decisions

Leaders should apply a nuanced sequencing strategy, evaluating each major process area through five lenses. Different profiles across these factors will lead to different sequencing answers, including cases in which AI first is clearly the right call.

A large healthcare services company learned this lesson firsthand. Concerned that offshoring could slow AI momentum or cede too much value to outsourcers, the company initially prioritized AI across many major processes. While some areas advanced, others saw far less progress. Eager to realize value quickly, leadership adopted a more nuanced approach: standardizing and offshoring the areas where process and data foundations were not yet ready, using labor arbitrage savings to fund the broader transformation, and accelerating AI in areas where the foundations were already in place. It was not a choice between offshoring and AI but a disciplined decision about where each approach applied.

Over time, AI will become more plug and play through lower costs, pretrained models embedded in enterprise resource planning and human capital management platforms, and demonstrated large-scale ROI. This will shift the equation. But that inflection point is uneven across processes and industries, and leaders who assume it has arrived risk forgoing tangible structural gains. For now, the key is rigorous, process-level evaluation.

What shared services become in an AI enterprise

Some argue that AI will make shared services obsolete. But we think AI will redefine shared services, not replace it. In the early stages of AI adoption, shared services help scale AI deployment. As AI agents execute more end-to-end processes, the human role shifts from task execution to orchestrating intelligent systems, managing exceptions, and driving continuous improvement. At scale, shared services become an enterprise capability hub for AI, trusted data products, and digital expertise.

Shared services organizations that cling to legacy expectations on role, value add, and operating models will find themselves caught between cost pressure from AI-native alternatives and demands for strategic value from business partners.

What leaders can do now

Sequence at the process level, and build the digital foundations to support it. Evaluate labor arbitrage potential, AI maturity, process stability, and data readiness by function. And invest in data governance, process ownership, and digital talent regardless of where the work sits.
Fund AI through structural moves. Shared services and offshoring can generate the capital required for sustainable AI investment.
Be rigorous about AI ROI. Distinguish between pilots and enterprise-scale impact. Many AI initiatives succeed in controlled environments but stall at scale. Focus investment on use cases with proven, hard-dollar value.
Plan future-back. Knowing that AI will become a high-ROI option within five years, ask yourself what building blocks do you need to lay today.

AI builds on structure, not around it

AI will reshape shared services, but it does not eliminate the value of structural discipline; it amplifies it. The same elements that made shared services effective—namely, scale, standardization, process ownership, and data governance—are increasingly the conditions that allow AI to deliver meaningful value. For most large enterprises, the path forward is not choosing between shared services and AI; it is sequencing them thoughtfully. Shared services build the operational foundation; AI builds on top of it. Companies that succeed know when to shift, when to standardize, and when to automate.