Infosys rolls out 'silicon to application' AI strategy

Bengaluru: Infosys has set its new AI strategy, “silicon to application,” in motion. The playbook aims to span the full AI ecosystem—from hardware to end-user applications—through targeted investments and partnerships across each layer.At the silicon layer, Infosys has partnered with Nvidia and Intel. “These partnerships matter most when clients need AI to run on their own infrastructure, whether for data sovereignty, regulatory compliance, latency requirements, or cost optimization,” an internal note to employees said.Two years ago, Infosys set up an Nvidia Centre of Excellence to train and certify 50,000 employees on Nvidia’s technology, enabling them to deliver GenAI solutions across industries.At the platform layer, the focus shifts to leading cloud ecosystems such as AWS, Microsoft Azure, and Google Cloud, which underpin most enterprise AI development with tools for building, training and deploying models.Infosys has also tied up with AI-native firms including OpenAI, Anthropic and Gemini to develop advanced enterprise AI solutions across telecommunications, financial services, manufacturing and software development, as it doubles down on agentic AI capabilities.The IT firm has partnered with Anthropic to set up the Anthropic Center of Excellence to build and deploy AI agents tailored to industry-specific operations, with plans to expand into financial services, manufacturing and software development. At the core of this tie-up is the integration of Anthropic’s Claude models, including Claude Code, with Infosys Topaz AI offerings.“At the foundation model layer, we work with OpenAI, Anthropic, and Google Gemini. Each model family has genuine strengths in different areas—GPT models in general reasoning, Claude in safety-conscious and long-context applications, Gemini in multimodal processing and cost-efficient inference through Flash variants. Our role is to understand these differences well enough to make honest, informed recommendations for specific use cases, rather than defaulting to whichever model is most familiar,” the note said.At the tools layer, Infosys is working with Cursor and Cognition—two companies building AI-native approaches to software development—along with tools from other AI-native firms such as Claude Code and OpenAI Codex. Cursor offers an AI-integrated development environment that complements tools like GitHub Copilot. “Cognition’s Devin represents an early attempt at autonomous software engineering. These are newer, evolving partnerships, but they represent where developer tooling is heading, and we believe early engagement gives us a better understanding of that trajectory,” the note said.Infosys and Cognition, the maker of AI coding agent Devin, have announced a strategic collaboration to scale the tool across global enterprises. Infosys’ financial services practice is already using Devin to transform engineering delivery across banking, payments, capital markets, insurance and wealth management.At the application layer, Infosys brings its strengths in industry expertise, enterprise integration, platform capabilities and responsible AI governance—helping clients move AI from prototype to production in complex enterprise environments.Infosys said the value of this layered approach is straightforward: real-world AI solutions often span multiple layers simultaneously. “Being able to draw from partnerships across the full stack allows us to architect coherent solutions rather than piecemeal ones,” the note said.