Infosys’ Future: Scaling GenAI and SLM Innovation to Drive Growth and Stakeholder Trust

Infosys’ proven engagement and delivery strategies continue to pay off, evidenced by accelerated sales during 4Q24 and an increase in FY25 revenue guidance for the third consecutive quarter. The company’s approach of underpromising and overdelivering, rooted in the company’s humble culture, allows it maintain trust with ecosystem stakeholders while it continues to expand and enhance its portfolio in emerging areas such as agentic AI and industry-aligned small language models (SLMs).
 
Infosys remains well positioned to surpass India-centric peer Cognizant for the No. 2 spot in revenue size, despite the inorganic boost of over $800 million that Cognizant will receive over the next year from its purchase of Belcan. Its’ relentless execution, backed by investments in talent development and partner-enabled solutions, will continue to be the company’s key to success as it gradually increases its share of value-based selling efforts, which are also bolstering its profitability, making otherwise impatient shareholders happy.

Scaling GenAI use cases through the development of prebuilt, industry-specific SLMs, and relying on highly skilled talent and resource management

As noted in TBR’s 4Q24 Infosys Earnings Response report, developing a client-ready AI-first portfolio is not a strategy unique to Infosys, but keeping pace with the rapidly evolving generative AI (GenAI) market highlights the company’s appetite for innovation and helps it strengthen stakeholder trust. Over the past 24 months, a large portion of vendor-client discussions focused on experimenting with developing and running large language models (LLMs), often fed with either public or nonessential data. Growing adoption of the technology has introduced the need for developing SLMs that are either function or industry specific.
 
While cloud-deployed models have far fewer resource constraints, there are still significant drawbacks with an LLM approach. Additionally, LLMs’ massive size leads to downsides in efficiency, cost and customizability, presenting serious hurdles over the long term, especially as contextualization improves. Moreover, when looking at specific use cases, SLMs built to perform particular tasks can outperform broader LLMs. These SLMs can be pretrained on smaller datasets, enabling developers to be more selective with training data and opt only for high-quality data sources pertinent to the desired use case.
 

 

Ecosystem partners remain a critical component of Infosys’ GenAI strategy

Infosys and NVIDIA co-launched three NVIDIA-enabled GenAI solutions, which, according to Infosys’ press release, use “NVIDIA NIM inference microservices, NVIDIA NeMo Retriever embedding models, and NeMo Guardrails to customize and deploy generative AI telco domain-specific LLM models.” Infosys also launched NVIDIA-enabled SLMs for Infosys Topaz BankingSLM and Infosys Topaz ITOpsSLM, targeting clients through core industry and horizontal offerings and allowing them to use their own data on top of the prebuilt SLMs.
 
Further, Infosys launched the Finacle Data and AI Suite of solutions to support banking clients seeking to enhance IT systems and customer experience using AI. The solutions include Finacle Data Platform, Finacle AI Platform and Finacle Generative AI Offerings. We see these capabilities as a prerequisite to enhance the core Infosys Finacle platform and enable Infosys to remain a formidable competitor in the banking space.
 
Despite the modularity of these offerings, we do not expect the company to change its commercial model and continue to use the suite of offerings to drive services opportunities. Infosys’ SLM portfolio expansion strategy closely mimics the company’s build-out of industry cloud offerings that address client pain points with prebuilt models for specific functions. The difference is the added complexity around building and managing the prebuilt SLM models with their massive number of parameters.
 
Developing and supporting these prebuilt models will require the right-skilled bench and, more importantly, retention programs enabled by unique career paths for programmers who are involved in such tasks. Infosys’ Power Programmers group of engineers consists of highly skilled professionals who are responsible for developing products and ensuring that the intellectual property they create and use meets the cost-saving requirements Infosys pitches to clients. The Power Programmers group is much leaner than the traditional software developers pyramid and resembles the business models that many vendors, including Infosys, may aspire to implement in the future.

Enhancing its chip-to-cloud strategy via acquisitions can bolster cloud performance, but only if Infosys accounts for and aligns with OEM and cloud vendor priorities, including edge and 5G

With its cloud business reaching 30% of Infosys’ total sales and growing 25.1% year-to-year in 4Q24, Infosys continues to invest across its portfolio to expand its addressable market opportunities. For example, Infosys Engineering Services remains among the fastest-growing units within Infosys as the company strives to get closer to product development and minimize GenAI’s disruption of its content distribution and support position.
 
Since the 2020 purchase of Kaleidoscope, which provided a much-needed boost for Infosys to infuse new skills and the IP needed to appeal to the OT buyer, Infosys has enhanced its value proposition to also meet GenAI-infused demand. Infosys’ investments in recent acquisitions including in-tech and InSemi have expanded the company’s addressable market opportunities around product engineering and silicon design services, further strengthening its chip-to-cloud strategy.
 
We do not expect growth of Infosys’ cloud business, Infosys Cobalt, to slow down anytime soon, given the company’s market position for infrastructure migration and managed services as well as its well-run partner strategy with hyperscalers. Adding semiconductor design services bolsters that value proposition as buyers consider whether to use price-attractive CPUs or premium-priced GPU data centers. The latter currently dominates the marketplace, and we expect that trend will not change for at least the next 18 to 24 months. But having semiconductor engineers on its bench can help Infosys start supporting CPU-run models, appealing to more price-sensitive clients.
 
Additionally, expanding its product engineering services also enhances Infosys’ edge and 5G value proposition, which we believe are two of the next frontiers for AI-enabled growth.

 

Follow Infosys’ performance throughout 2025 with data and analysis in TBR Insight Center. Start your free trial today.