The easiest way to misunderstand Capgemini is to call it a consulting firm and picture a neat row of people carrying laptops into a conference room. The reality is noisier. Somewhere in its orbit, a bank is replacing software that predates the smartphone, a factory is connecting machines to a cloud platform, a government office is redesigning a citizen service, and a customer-support operation is deciding which questions an AI agent may answer without a human. Capgemini sits in the middle, translating ambition into systems that must work on Monday morning.
That is what the Paris-headquartered company sells: the ability to change a large organization without bringing it to a halt. It can advise executives on strategy, write and integrate software, migrate infrastructure, engineer connected products, secure the result, and keep operating pieces of the business after the consultants leave. The work is often unglamorous. It is also difficult to replace because big organizations accumulate decades of data, permissions, vendor contracts and exceptions.
The enterprise plumber
Clients usually arrive with one of two problems. They need growth from new products and better customer experiences, or they need efficiency from systems and processes that have become slow, expensive or brittle. Frequently they need both. Capgemini's portfolio follows that tension. "Customer First" covers commerce, marketing, sales and service. "Enterprise Management" deals with the machinery of finance, HR, procurement and core applications. "Intelligent Industry" connects software to physical products, factories and supply chains. Cloud and data sit underneath all three, with cybersecurity and sustainability threaded through them.
The buyers are not startups looking for a weekend prototype. They are banks, manufacturers, retailers, utilities, telecom operators and public agencies managing risk at national or global scale. Recent named work includes helping Bentley Motors modernize data, AI and manufacturing systems at its Crewe factory, orchestrating technology operations across BAE Systems, and extending a transformation relationship with McDonald's. Capgemini also works with the technologies its clients already own, whether those come from SAP, Salesforce, ServiceNow, Microsoft, Google Cloud or AWS.
“Together we can give technology that human touch the world needs.”Aiman Ezzat, chief executive officer
This makes Capgemini less like a single product and more like a switchboard. The company does not need to persuade a chief information officer to abandon every vendor. It gets paid to make competing products coexist, move old data into new structures, wrap controls around them and train people to use the result. In a market full of powerful software, integration remains the stubborn tax.
AI meets the old wiring
Generative AI made demonstrations cheap. A team can summarize documents, draft code or produce a chatbot in days. Production is different. An agent that touches a payment, a patient record or a factory schedule needs permission rules, reliable data, audit trails, security and a clear handoff to a person. That gap between what a model can do and what an enterprise can safely deploy is Capgemini's immediate market.
The company has wrapped its approach in named offers. Its Resonance AI Framework is a sequence for choosing priorities, changing operations and moving from pilots to scale. RAISE is a modular engineering foundation for building governed AI systems. Under those labels sits familiar systems work: modernize the technology stack, prepare data, connect models to workflows, measure outcomes and keep humans accountable.
The cash consideration, excluding net financial debt, for WNS. The 2025 acquisition added business-process expertise and operating teams to Capgemini's AI proposition.
WNS matters because agents require processes in which to act. The acquired company brought deep experience running finance, customer service, analytics and industry operations. Capgemini brought global technology delivery, consulting and partnerships. Put together, the offer moves beyond installing an AI assistant. It proposes to redesign and then operate an end-to-end process, with software agents doing more of the routine work.
The alliances provide the model layer. Capgemini became a founding member of OpenAI's Frontier Alliance in February 2026 and later invested in the OpenAI Deployment Company. It also created a Google Cloud AI Enterprise Hub with joint teams intended to work inside client environments. Microsoft, AWS, Salesforce and others fill out an ecosystem deliberately broader than any single platform.
A business made of contracts and people
Capgemini earns money through project fees and multiyear contracts. A strategy engagement may lead to an implementation, which may lead to cloud or application management, which may lead to operating a business process. That ladder explains the appeal of being end to end. Each step reveals more of the client's systems, makes the next assignment easier to win and deepens the switching cost.
The numbers make the model plain. In 2025, Applications and Technology accounted for 63 percent of revenue. Operations and Engineering contributed 29 percent. Strategy and Transformation was just 8 percent. Manufacturing was the largest client sector at 25 percent of revenue, followed by financial services at 22 percent and the public sector at 15 percent. This is not a boutique adviser. It is an industrial service network.
The network is global by design. At the end of 2025, 279,200 employees were classified as offshore, 66 percent of total headcount. Work can move among teams and time zones, mixing proximity to the customer with lower-cost delivery centers. The model supports scale and margin, but it also leaves the company exposed when automation reduces the labor required for coding, testing, support or process work.
AI makes some hours cheaper. Capgemini's bet is that it makes the transformation larger.
Management is responding on both sides. Employees averaged 97 hours of learning in 2025, much of it aimed at AI skills. At the same time, the company expects roughly €700 million in restructuring costs across 2026 and 2027 as it adapts roles and capacity country by country. The juxtaposition is the strategy in miniature: teach a huge workforce to use AI, remove work that no longer fits, and redirect investment toward larger programs.
Where Capgemini fits
The competitive set is crowded. Accenture is the obvious global comparison. IBM Consulting, Deloitte, Cognizant, Tata Consultancy Services, Infosys, Wipro and NTT DATA chase many of the same transformation budgets. EPAM and Globant compete for digital engineering. Cloud providers increasingly offer professional services, while software companies cultivate their own implementation armies.
Capgemini's difference is a particular combination rather than a magic ingredient: European roots, a large offshore base, industrial engineering inherited and expanded through Altran, a credible strategy arm in Capgemini Invent, and enough platform relationships to meet clients where they are. Its research institute and sports partnerships create visibility, but the durable advantage is institutional memory. People who know how a bank's ledger, an airline's booking flow or a factory's production line actually behaves are useful when a new model meets old wiring.
That advantage can also become weight. A company with more than 420,000 employees must coordinate incentives, skills and quality across dozens of countries. It must remain neutral while partners compete. It must promise efficiency without making clients wonder whether billed effort should fall faster. And because it operates sensitive systems, failures of governance can travel quickly from a local contract to the global brand.
The culture beneath the scale
Serge Kampf founded the business as Sogeti in Grenoble in 1967, in a converted two-room apartment. His seven values still sound more like house rules than a compliance framework: honesty, boldness, trust, freedom, fun, modesty and team spirit. They have survived acquisitions, renamings and the growth from five people to a workforce larger than many cities.
Sogeti opens in Grenoble with Serge Kampf and a small founding team.
Ernst & Young Consulting expands the group's management-consulting reach.
Capgemini Engineering unifies the group's engineering capabilities after the Altran acquisition.
WNS joins, giving intelligent operations a much larger process engine.
OpenAI and Google Cloud alliances push AI deployment closer to client workflows.
The values are unusually human for a firm selling automation. That may be useful. Enterprise AI is not only a technical migration. It rearranges authority: what a machine may decide, when a person must intervene and who owns the mistake. Capgemini's mission says it wants to unleash human energy through technology for an inclusive and sustainable future. The test is whether those words influence thousands of small design decisions inside real systems.
The practical promise
For a client, Capgemini can be helpful when the problem crosses organizational borders. Moving one application to the cloud is a project. Redesigning how customer orders travel through marketing, commerce, inventory, finance and support is a transformation. The latter requires process knowledge, software engineering, change management and someone willing to be accountable after launch.
Capgemini reported €22.465 billion in 2025 revenue, a 13.3 percent operating margin and €24.356 billion in bookings. In the first quarter of 2026, generative and agentic AI made up more than 11 percent of bookings. Those figures show demand, not destiny. The more revealing question is whether the company can turn AI from a new label on familiar work into a better way of delivering it.
Nearly sixty years after Kampf opened that Grenoble apartment, Capgemini is still selling managed change. The vocabulary has moved from data processing to digital transformation to agentic operations. The client anxiety underneath it is recognizable: the future is arriving, the current system cannot simply be switched off, and somebody has to connect the two.