For years, India has been a favoured market for most tech firms. From email to search to social media, even back-offices, India’s population and large pool of talent has been a magnet for foreign companies. Artificial intelligence is now following a similar script: India has the talent, digital public infrastructure, and one of the world’s largest pools of users and, yet, most of the AI systems used in the private and public sectors in India have been developed by foreign firms.
That might be changing. In what is at best a first step, HCLTech earlier this week acquired a 10.46 per cent stake in homegrown Sarvam AI. The Series B round—which saw a $150 million injection from HCLTech —infused a total of $234 million into Sarvam, catapulting it into unicorn territory.
The funding announcement comes at a time when the US government last week blocked exports of Anthropic Claude Fable 5 and Claude Mythos 5 models on grounds of ‘national security’, effectively cutting off access to what is increasingly seen as critical technology.
The question now is whether HCLTech’s bet can convert India’s sovereign AI ambition into commercial infrastructure, or whether Sarvam will remain a rare domestic success story in a market still dependent on foreign models.
What HCLTech aims to gain from Sarvam investment
Many Indian business groups, including Reliance, Adani and Tata, have entered into AI-related partnerships with global players, ranging from building AI infrastructure and data centres to enterprise AI services and consumer-facing AI access. HCLTech itself has partnered with OpenAI, where it will leverage the latter’s advanced models for its consulting, engineering and cloud services.
With Sarvam, HCLTech wants to go beyond the traditional Indian IT services playbook. In an interview with Business Standard’s Shivani Shinde, HCLTech CEO C Vijayakumar said Sarvam will allow the tech major to target the “sovereign AI” opportunity, especially in areas such as enterprise adoption across banking, insurance, GovTech and large commercial firms, and the use of AI to reimagine citizen services.
Globally, he said, enterprises may increasingly look beyond frontier models for business-specific use cases, where small language models or domain-specific models could prove more efficient.
This is the real playbook behind HCLTech’s Sarvam deal. In the software era, Indian IT services firms built scale by implementing, customising and maintaining systems created by global technology companies. AI threatens to rearrange that value chain. If large language models and agentic platforms become the foundation of future enterprise technology, the firms closest to that foundation may capture more strategic value than those merely implementing it.
Dr Sorabh Bajaj, Director of the Centre for Digital Learning at Pune’s FLAME University, said, “The current Western large language models (LLMs) are structurally limited by a non-Indian knowledge base, introducing inherent cultural and linguistic biases that fail India’s complex reality.
“True sovereign AI requires local LLMs built from scratch to reflect Indic nuance, native speech patterns, and local datasets,” he added.
Kunal Khanna, founder of AI-powered professional networking platform Match It Up, told Business Standard that the deal allows HCLTech to secure intellectual property at the deepest layer of the technology stack.
“This strategic ownership enables them to offer highly customised, deeply integrated, and fully data-sovereign solutions to their enterprise and government clients,” Khanna said. “It empowers them to guarantee privacy and security natively, shifting their role from a systems integrator to a foundational architect of enterprise intelligence.”
Rishabh Sagar, CEO and Co-founder of AI-powered video editing platform CRAON.AI, said owning a stake in Sarvam gives HCLTech access to technology, talent, and the ability to influence products tailored for Indian and enterprise needs.
Why is the Anthropic episode a sovereignty wake-up call?
Industry leaders said the US government’s export-control order blocking foreign access to Anthropic’s top AI models is a test case for the need for AI sovereignty.
Vineet Moroney, Chief Transformation Officer at digital engineering firm Xoriant, told Business Standard that HCLTech’s investment in Sarvam is the ultimate test case for India’s sovereign AI policy because it forces a shift “from theoretical self-reliance to hard commercial execution”.
“In a world where access to frontier models can be throttled overnight by foreign jurisdictions, true digital equity requires owning the underlying weights, data, and infrastructure,” Moroney said.
Khanna said the Fable 5 episode had underlined why India should think about AI sovereignty beyond local data storage. “If an Indian AI agent must route through a foreign server just to securely verify its identity or message another domestic agent, we miss out on true digital independence,” he said.
Can national ambition become business use cases?
But India does not merely need a symbolic AI champion. It needs models that are usable, affordable, secure and good enough for enterprises and governments to trust in everyday operations.
Experts said a sovereign model is useful only if it can be taken into the messy environments where real work happens, such as loan processing, claims handling, customer support, multilingual public-service delivery, procurement, compliance, document review and frontline operations.
“In the end, AI can only add true value to enterprises when it is embedded into the core of how products are built, workflows are run, and decisions are made, what we call as applied intelligence,” Moroney said.
That is where HCLTech’s enterprise reach matters. Sarvam has positioned itself as a full-stack AI company rather than a narrow model lab. It has released models trained from scratch in India, including a 105-billion-parameter reasoning model and a 30-billion-parameter model designed to run on consumer-grade hardware. Its work spans speech, vision, conversational AI, enterprise use cases and government applications.
Sarvam may build the model layer, but HCLTech can help carry those models into large organisations, integrate them with existing technology systems and shape them around industry-specific problems.
Khanna said public adoption will also depend on Indian AI firms moving beyond conversational interfaces.
“The next major milestone for Indian AI is evolving from conversational interfaces to foundational, autonomous infrastructure,” he said. “Public adoption will truly accelerate when Indian AI firms integrate directly into the daily operational workflows of our MSMEs and enterprises.”
Can private capital complete what policy has started?
India’s sovereign AI push has so far relied heavily on state support. The IndiaAI Mission, with a five-year outlay of ₹10,371.92 crore, is designed to support compute access, indigenous model development, data quality, startup funding and AI adoption. India has also unveiled homegrown AI models focused on Indian languages, voice use and local data control.
Sagar said government procurement may help early adoption, but the longer-term test lies elsewhere.
“Government support can help in the early stages, but long-term adoption will come from building products that solve real problems for businesses and consumers,” he said. “Indian AI companies need to focus on sectors like healthcare, education, media, financial services, and enterprise software.”
Khanna said government support and private capital are complementary. “Government support under the IndiaAI Mission serves as the incredible catalyst. It helps de-risk massive initial compute investments and establishes the vital regulatory and standards frameworks,” he said. “Private capital, like HCLTech’s, then brings the enterprise distribution, market discipline, and operational scale necessary to take these technologies global.”
What happens after the unicorn round?
While the HCLTech funding may give Sarvam capital and visibility, it does not ease the tough road ahead. Training and serving large models require expensive graphics processing unit infrastructure. Model performance has to improve continuously. Inference costs have to be controlled. Enterprise clients will compare Sarvam not only with global closed models but also with fast-improving open-source alternatives. Government departments may support domestic AI, but enterprises will pay only if the product delivers.
In simple terms, Sarvam will have to show that sovereignty can be sold as a business advantage, not just a national aspiration.
“This deal will test whether India can scale a native, multilingual AI ecosystem that satisfies strict local governance while remaining competitive enough for global enterprise deployment,” Moroney said.
Sagar said HCLTech’s investment is an encouraging signal, but one deal alone cannot define success.
“Real validation will come when Indian AI companies achieve large-scale adoption, build globally competitive products, and create sustainable businesses,” he said. “We’re still in the early innings.”
Bajaj said success five years from now would mean Sarvam’s models becoming default infrastructure for mission-critical citizen workflows, localised e-governance and domestic enterprise platforms at a fraction of current computing costs.
“Conversely, failure would look like Sarvam being relegated to building shallow application wrappers or generic consulting layers on top of foreign foundation models due to an inability to sustain the massive capital expenditures required for frontier R&D,” Bajaj said. “If India remains structurally dependent on foreign APIs for its primary intelligence infrastructure, its sovereign AI policy will have failed.”
That warning is useful because it separates the two possible futures. In one, Sarvam becomes infrastructure. In the other, it becomes another AI services layer.
Can this become a template for corporate India?
If the bet works, the implications could extend beyond HCLTech and Sarvam.
Bajaj said HCLTech’s investment breaks a historical pattern in which Indian conglomerates relied on foreign technology vendors rather than nurturing domestic deep-tech.
“A successful Sarvam demonstrates that Indian capital can successfully scale indigenous foundational research, establishing a template for corporate India to back domestic AI firms,” he said.
Khanna said: “As these models succeed, we will see a natural evolution where Indian corporates move beyond merely licensing AI to actively investing in and co-creating domestic intelligence infrastructure.”
The next few years will show whether Sarvam can move from model launches to daily use inside enterprises and government systems. If it can, HCLTech’s $150-million bet may become more than a successful startup investment. It may become a proof point for India’s attempt to build sovereign AI as working infrastructure.
