How TCS is looking to accelerate its AI play in FY27

India’s largest IT services and consulting firm Tata Consultancy Services posted a sequential dollar revenue growth of 1.5%. However, on a full-year basis, the company saw its first decline in revenue, which fell from $30.17 billion in FY25 to $30.01 billion in FY26. While annual dollar revenue fell 0.5%, profitability improved, with margins expanding by 70 basis points year-on-year to 25% in FY26. Notably, the company said that even after excluding the AI component in large deals, the value of its AI-led business transformation revenue has crossed $2.3 billion on an annualised basis, accounting for about 7.7% of its total revenue.  

AI services and opportunities   

At the company’s earnings call, management of TCS said its FY26 order book exit of $40.7 billion was driven by deals spanning vendor consolidation, AI-led modernisation, enterprise-scale AI deployment, digital core, and data platform transformation, operating model changes, and regulatory compliance. K Krithivasan, CEO and MD, said AI is expected to be net accretive, with revenue productivity in AI-led business significantly higher than the company average and traditional segments, both onsite and offshore. TCS sees the ongoing generative AI cycle to drive a structural shift in revenues, with AI-led income rising even as some traditional revenue streams gradually taper off, ultimately offsetting and surpassing declines in legacy service lines. 

Elaborating on where demand for AI services, particularly in transformation, is emerging, Aarthi Subramanian, COO of TCS, said enterprises are investing in cloud adoption and migration, data modernisation, cybersecurity, and enterprise systems upgrades. She said that a distinctly AI-led opportunity lies in modernisation and the reduction of technical debt—areas that companies had long deferred due to time and cost constraints. A third area of growth, she added, is pure-play AI transformation, where AI directly drives measurable business outcomes. 

“In the last two years, AI model capabilities have evolved tremendously and rapidly. Yet there is a gap in enterprises realising the true potential of these technologies. One of the challenges our customers face is that they have invested in these tools but not yet reaping the expected productivity benefits. Our goal is to systematically help our customers address these gaps with our Human + AI service autonomy model,” Subramanian added.

HyperVault execution   

In October last year, TCS incorporated a wholly-owned subsidiary  HyperVault to establish multiple AI and sovereign data centres for providing Infrastructure and technology enabled services, bringing in TPG as an equity partner. TCS CEO Krithivasan said that significant progress was made in Q4FY26 on its capacity buildout. “This includes winning customer commitments, land parcel finalisations, and partnering agreements. We are actively engaging across the full ecosystem of hyperscalers, semiconductor companies, and model providers while TCS is the integration partner across infrastructure, engineering, and AI-led services,” said Krithivasan.  

With an investment of up to $2 billion by both TCS and TPG, the company has earlier said that the 1-GW capacity data centre is expected to reach full capacity in a five-to-seven-year time frame. The first revenue stream from the data centre business is expected to flow in FY27-28.  

As a part of the buildout of the datacentre, the company updated that engagements with hyperscalers and frontier AI model companies had moved beyond early exploration into design alignment, security frameworks, site due diligence, and commercial structuring. For HyperVault, TCS has already partnered with Tata Power, Tata Projects, Tata Communications, as well as GE, Honeywell, ABB, Siemens, and others for EPC power, cooling, controls, networks, and security. It expects anchor AI workloads to be in the range 100-200 MW range per customer, including the OpenAI partnership to build 100-MW capacity with an option to scale to 1-GW. The company has also partnered with AMD to combine the Helios rack-scale AI architecture and tech capabilities of Tata Group entities to create high-density AI capacity in India.  

Apart from OpenAI, the company said it is also working with other frontier model companies such as Anthropic and Mistral. “With all model companies, it's extremely important for us to build a strategic partnership. I would say that we are also trying to shape these partnerships differently than the traditional GTM partnerships of the past. We want to make them 360-degree partnerships, not just one. Because of our investment in HyperVault, we have a very unique opportunity where they can become our customers,” Subramanian added.   

On the back of Q4 results coming in line with Street estimates, top brokerages including Nomura, Motilal Oswal Financial Services, Nuvama Wealth Management, and Centrum Broking have maintained a ‘buy’ rating on the stock. Analysts at Centrum broking noted, “Strong order book, accelerating AI-led opportunities, resilient margins, and improving client confidence underpin medium-term growth visibility. With most macro headwinds likely behind and a gradual recovery expected into FY27, we see favourable risk-reward at current levels, supported by TCS’s leadership in large deals, execution track record, and strengthening positioning in AI-led transformation.” Motilal Oswal analysts said that while the HyperVault strategy has started to take shape, the timeline remains unclear.   

On the TCS results and management commentary, Shubham Rathore, Principal Analyst, Gartner, said the demand environment is increasingly shifting towards complex technologies, echoing Gartner data, which projects that the generative AI models market will grow by 107.3% in 2026. However, despite this momentum, he sees the sector continues to face challenges related to cautious client discretionary spending and the premium pricing required for specialised AI talent. “As we look towards the upcoming fiscal year, the information technology sector's outlook will depend heavily on transitioning clients from simple productivity augmentation to scaled, multi-step AI implementations,” Rathore added.