Technology Business Research, Inc.Infosys Moves From AI Experimentation to Enterprise Orchestration as Infosys Topaz Takes Center Stage in the Company’s Next Chapter
Infosys hosted more than 200 clients, partners, analysts and advisers during its flagship event, Infosys Americas Confluence 2026. The packed agenda included Partner Day and Analyst Day tracks, client stories, partner presentations and Living Labs demos of both Infosys- and partner-developed solutions and services. AI dominated the agenda, but the conversations were notably different than those at Confluence 2025. Last year, Infosys and its clients largely focused on the prerequisites for enterprise AI adoption, including data readiness, governance, architecture and talent. This year the discussion shifted toward what happens when enterprises actually begin scaling AI; the need to address rising complexity, agent governance, token economics, cybersecurity and model choice; and the ever-difficult task of proving business value. Infosys used the event to reinforce its ambition to become a leading enterprise AI services provider, positioning Infosys Topaz — and increasingly Infosys Topaz Fabric — as the enabling layer connecting models, applications, data business processes and partner technologies.
Infosys Topaz is evolving from an AI portfolio into an orchestration layer, putting Infosys’ integration strengths back at the center of the AI opportunity
Infosys’ AI strategy has matured considerably since Confluence 2025 and even more significantly compared to two years ago. In 2024, Infosys was positioning Infosys Topaz as a collection of AI services and solutions that could sit on top of the company’s Cobalt cloud foundations, as the company was taking a more pragmatic approach to increase its value with clients, using the idea that a client could not meaningfully scale generative AI (GenAI) without first addressing data, governance, architecture and workforce readiness.
In 2026 these prerequisites remain, but Infosys is now trying to operationalize them through a more coherent AI architecture. Infosys’ AI-first value framework spans areas including strategy and engineering, data, process AI, agentic legacy modernization, physical AI and AI trust, with the company tying these capabilities to three broader market shifts: automated intelligence, AI-augmented workforce and AI-powered predictive agility. Infosys Topaz Fabric provides a common layer through which the company can expose models, tools and AI assets while protecting client data and managing the economics of model consumption.
Supporting both open and closed public models through Infosys Topaz Fabric will give the company the flexibility to strengthen its position as an AI orchestrator. We believe supporting both models is table stakes for the services provider, which, for years, has played the technology-agnostic game with its alliance partners. The true value will come from Infosys’ ability to deploy its frontier engineers at speed and capture the right balance of AI-first and AI-infused revenue opportunities, as the former will help it drive higher pricing and the latter will increase client stickiness. Managing the expectations of impatient shareholders will be key, especially as peers pursue similar strategies.
An event session with an energy client brought some of that strategy to life. Infosys argued that the only part of the AI stack the enterprise truly owns is the context, including processes, institutional knowledge, operating history and domain expertise. The model can be rented or swapped, but context can not. The client described context not as a static database but as a supply chain containing years of engineering knowledge, operational data, procedures, decisions and best practices.
This framing aligns with Infosys’ strategy as the company does not need to own the client’s context but rather make the context usable across applications, data and business workflows. Understanding the context is key to Infosys’ value proposition. The client described Infosys connecting those dots while leaving ownership of the underlying knowledge to the client. Executing well against the strategy could help Infosys elevate in the context-orchestration assurance value chain, a step above its traditional implementation services realm, but it could also fall short if the company fails to remain proactive in the “what’s next phase” of the client’s business growth conversations.
AI-centric alliances allow Infosys to build its position as a tech orchestrator, but differentiation will require agility as the tech requires a multiparty setup
Although alliance partners remain central to Infosys’ go-to-market strategy, AI is changing their function. In a traditional cloud or enterprise application relationship, partners often provided the technology platform while Infosys supplied the integration, implementation and managed services. AI, however, makes the relationship a bit more fluid as clients may use several model providers, hyperscalers, SaaS platform and AI-native tools simultaneously. This will test Infosys’ traditional vendor-agnostic stance as the ability to move between technologies and govern them consistently becomes as important as expertise in any one platform.
During the event, Infosys’ executives including company’s CEO Salil Parekh highlighted partnerships with cloud, SaaS and frontier AI companies while noting that Infosys Topaz Fabric can select the appropriate model for the task and cost requirements. Meanwhile, the company continues to use its large install capabilities around vendors such as SAP and Microsoft serve as the foundation for AI-enabled transformation, particularly as those vendors increasingly embed AI into their own platforms. The 360-degree arrangements of some of these relationships has historically benefited Infosys, as clients often view Infosys as a trusted execution provider.
AI has now raised the stakes, and Infosys has an opportunity to elevate that trust further and become as independent orchestrator across a fragmented AI stack. Over the last six months, Infosys’ alliances with Anthropic, OpenAI, Intel, Harness and Cursor suggest the company is trying to capture the emerging opportunity between AI creation and enterprise execution. Anthropic gives Infosys access to frontier reasoning and agent design for complex, regulated workflows, which help Infosys move upstream into operating-model redesign rather than be confined to downstream implementation.
Through its relationship with OpenAI, Infosys is combining Infosys Topaz Fabric and OpenAI Codex to jointly drive AI adoption across enterprises, with a focus on transforming software development and modernization. We see the relationship with OpenAI as both an opportunity and a threat for Infosys, highlighting leadership’s willingness to disrupt the company rather than be disrupted. By tying Infosys Topaz Fabric to Intel Xeon, Gaudi and AI PC environments, and by emphasizing rightsized architectures, benchmarking and total cost of ownership, Infosys is positioned to shape enterprises’ infrastructure decisions before they select hyperscaler-led architectures. Harness then tackles the production bottleneck that follows AI-enabled code generation by standardizing the path from code to governed deployment across hybrid and multicloud estates.
Cursor adds another layer of significance because it pushes Infosys deeper into the developer workbench. If Infosys can standardize these partner technologies across its own delivery base, it could convert internal productivity gains into a proprietary delivery advantage by using its workforce as both a proving ground and a distribution channel. With Infosys’ ©Human Intensity Reduction Index (HIRI) trending toward 0% in the last few quarters, the company has a long way to go before nonlinear growth materializes. We believe the best path forward for Infosys will be less about monetizing agents through direct sales and more about bundling them in fixed-priced-wrapped contracts that can help it generate higher volume, but only after it pivots from IT-based to business-centric outcomes.
Client use cases demonstrated that Infosys’ AI value proposition centers on embedding AI into transformations already understood
Client examples throughout the event were more mature than the productivity pilots that dominated early GenAI discussions, with some of the more concrete examples revolving around embedding AI into broader modernization, engineering and operating model changes.
For an energy client, Infosys built an operations assistant that began as a conversational interface for engineering procedures but evolved into a tool that helps engineers aggregate data, conduct analysis, and move from the client reactively addressing issues toward more proactive operations. The client linked the approach to improved maintenance and reliability; lower operating costs; and extended useful life of assets, potentially by several years. Infosys is also codeveloping the client’s enterprise AI platform, where multiple agents support both revenue-generating and cost-cutting use cases.
A financial services client offered a different modernization example. Instead of approaching legacy IT environments as systems that must be replaced wholesale, the company described using AI to automate regression testing, document legacy applications and incrementally modernize older code. This use case shed light on the shift AI is creating around cost versus value, potentially making incremental renovations more attractive than large-scale replacement programs, especially as the overuse of tokens can backfire and slow progress. For Infosys, this could expand its application modernization opportunity, even as automating coding reduces the labor needed for individual tasks.
A personal-care provider illustrated the relationship between platform implications and AI readiness. Using the Infosys Equinox commerce platform to support its IT transformation initiative, the company is in the process of consolidating seven disparate commerce platforms into two. The client emphasized that fragmented technology and data environments become even more problematic as AI scales, as they amplify wherever complexity already exists, further reinforcing Infosys’ position that AI does not replace the need for modernization but makes it more urgent.
Other examples showed similar progression from isolated pilots to scale programs. One client said it had moved roughly 24 AI use cases into production within several months, with another 27 ready to follow, and described Infosys Topaz as helping turn large volumes of operational events into actionable outcomes. Infosys is also extending Infosys Topaz Fabric into developer environments and operations tools, including lightweight embedded apps and purpose-built applications for modernization, software development life cycle, data engineering and process engineering.
Protecting relevance while creating a path toward higher-value orchestration work will allow Infosys to maintain client stickiness as the company continues to use AI to deepen and accelerate transformation programs across technology estates it is already familiar with. Although one can argue that forums such as Confluence often have a predetermined agenda and talking points, TBR’s attendance at the conference the last several years and ongoing coverage of the company for the better part of the last two decades have allowed us to see and hear clients’ firsthand accounts and consistent feedback about the company’s culture built around humility. Many of the clients presenting on stage are not just frequent presenters but also expand on the relationship story and discuss what’s new and what’s changing in their work together.
Governance is moving upstream as agentic AI expands both the opportunity and the complexity of enterprise transformation
One of the consistent themes of the event was that AI becomes more complicated as it scales. Enterprises are discovering that moving several controlled pilots to hundreds or thousands of agents changes the governance problem significantly. Data lineage, agent identity, model selection, security, token consumption and accountability must increasingly be considered at the architecture and operating-model levels rather than handled on an ad hoc project basis.
Infosys’ clients, both on and off stage, repeatedly emphasized the importance of addressing governance earlier. For example, an energy equipment and manufacturing client described an approach in which AI ideas move through the governance process that asks questions about ROI, subject matter expert (SME) availability and data readiness before development begins.
Meanwhile, as agents proliferate across the enterprise and governance complexity increases the need for better cybersecurity, Infosys will create further openings if the company manages to combine capabilities that clients previously bought separately. As AI trust, cybersecurity, data architecture, application modernization and managed services increasingly intersect around the same workflows, the company could use Infosys Topaz Fabric as the connective layer while monetizing governance and assurance through recurring services. With both Infosys’ cybersecurity and Infosys Consulting capabilities cutting horizontally across the company’s entire portfolio, Infosys needs to ensure the internal governance hierarchy does not create a hurdle of greater integration, impacting client relationships.
AI will compress labor intensity faster than it eliminates Infosys’ talent pyramid, increasing pressure on the company to monetize productivity gains
Infosys is redesigning its workforce model for an AI-first services business. The company is building a talent strategy around three pillars: a new talent operating model, a new career model and a new talent development model. According to Infosys executives, these pillars will help the company “augment the whole organization with AI while also building deeper engineering and domain expertise to deliver AI-first services.”
We see these pillars as an evolution rather than a complete delivery model transformation as the company tries to minimize the disruption of its ongoing contracts and broader stakeholder ecosystem. We do not expect Infosys to ever fully change its delivery model and develop a HIRI that resembles that of a software company. However, the company’s workforce redesign will give it the necessary insights into how to position the newly developed skills with clients and tech partners. Executives noted that the company has trained more than 270,000 employees on AI, with a significant number already using AI-powered coding tools.
In addition, Infosys plans to build a pool of 6,000 frontier engineers over the next several years. This is in line with many of its peers, which are looking to pursue similar routes to elevate the value of their offerings. The challenge ahead for both Infosys and its peers is centered on balancing the scale and skills of this bench and on managing the influence of this bench on their staffing pyramids. The Palantir-developed forward-deployed engineer (FDE) model relies on fewer people and more platform-based service delivery centered on outcomes.
Moving away from the mindset and messaging around scale of this bench could be a double-edged sword unless Infosys transforms its commercial model at scale. The current use of fixed-price engagements creates a buffer against potential pushback from shareholders about revenue visibility. Using fixed-price contracts as a wrapper for outcome-based services will likely be the next phase but will compel Infosys to better align its expectations with tech partners.
Leadership continuity limits transition risk but could make it harder for Infosys move up the value chain
Confluence also served as an important marker in Infosys’ leadership transition. Parekh introduced company veteran Ashiss Dash, who was recently appointed by the board as the CEO Designate until April 2027, when he will become the company’s next CEO. Dash used his initial remarks to emphasize client centricity, continuous learning and the company’s long history of reinventing itself through successive technology disruptions. Although we do not expect Dash to make any drastic changes to Infosys’ operating model, we believe he will likely try to put a stamp on his tenure, just like every CEO, by pursuing further industry specialization and deeper alliance alignment.
Leaning on a strong foundation enabled through a pragmatic approach to enhance rather than fully transform its operating and delivery models will allow Infosys to focus on preparing its commercial and partner models to scale AI sales. Initial results suggest Infosys has the proper ingredients to execute on the task. Accelerating that evolution will test Dash’s ability to double the company’s revenue, as Parekh has done during his tenure, with Infosys growing from $10 billion in 2018 to over $20 billion in 2025. The stakes are high for Dash and his team, especially as AI is as much of a revenue tailwind worth messaging as it is a compressor of core areas. Expanding AI-first and AI-infused revenue while maintaining service quality execution will also test Dash’s readiness as the company looks to meet expectations within its stakeholder ecosystem.
Infosys’ next challenge is no longer proving it can build AI but whether it can shape what clients do with the technology
Infosys’ share of AI revenue grew from 5.5% in 1Q26 to 8.2% of total sales in 2Q26, suggesting that the company’s formula is working for now. Generating over $1 billion in AI services revenue on an annual basis is a notable milestone, but pushing the narrative too much could increase the cannibalization of traditional revenue. TBR will continue to monitor and estimate Infosys’ AI performance, especially as AI-related sales and marketing expenses create new margin headwinds. We believe that in two to three years, the company will be less vocal about itemizing AI sales. Sustaining momentum will be measured not just by sales but also by profitable growth, something Dash will have to navigate.
Infosys has a credible opportunity to separate from peers if it can make Infosys Topaz Fabric the operating layer for governed agent delivery while using FDEs to shorten the time to value and embed AI into clients’ operating models. That would strengthen Infosys’ differentiation at a time when many competitors continue to go to market leaning on AI tools plus their services. However, this shift will pressure Infosys’ commercial model. Clients are already expecting AI productivity benefits to flow back to them, and competitive intensity is increasing, raising the risk that Infosys gives away too much value through pricing concessions.
The company will need to carefully balance its use of time-and-materials (T&M)-based, fixed-price and agent-based pricing models, ensuring productivity gains are captured through outcome-based constructs rather than diluted through discounting. Execution complexity, partner dependence, offshore-led margin protection, and the need to recalibrate the staffing pyramid around FDEs and other AI-native roles remain the main risks to watch.
