Daybreak - India's IT sector bet its future on AI access it doesn't actually control

Episode Date: June 23, 2026

On June 11th, TCS announced an exclusive partnership with Anthropic — 50,000 employees trained on Claude, early access to new models, a dedicated business unit. The next day, a US export control order cut off access to Anthropic's most advanced models for users worldwide, including the very partner that had just signed up for early access.India's IT sector has spent years building its AI pivot on the assumption that access to frontier models would stay open. That assumption just cracked. And while India accelerates its own sovereign AI mission, analysts say that alone won't be enough to fix what just happened.So what would?Tune in.Daybreak is produced from the newsroom of The Ken, India’s first subscriber-only business news platform. Subscribe for more exclusive, deeply-reported, and analytical business stories.

If you're someone who uses Claude, then you've seen the little note that pops up about Fable 5. When you try to switch models on the platform, the top option, which is usually the best one and which right now claims to be for your toughest challenges, is greyed out. And it says that it is currently unavailable. It has been that way for users around the world since June 12th. Ever since a US export control order that targeted Anthropics' newest models, Fable 5 and Mythos 5 was released. The initial order required the company to take it down for foreign nationals only. But because Anthropic couldn't confidently identify the citizenship of its users,

it ended up blocking up those models for every user around the world. Now, everyone has been talking about this shutdown. But something happened just the day before June 12th that got completely overshadowed by these headlines. On June 11th, Tata Consultancy Services, or TCS, announced a huge partnership with Anthropic. 50,000 employees would be trained on Claude. A dedicated business unit would be set up so that TCS could package Anthropics models into its enterprise offerings and tailor make it to suit different clients from different sectors. Also, TCS would get early access to Anthropics' latest models all before public. release. Now, this is the kind of exclusive partnership that allows TCS the opportunity to

reliably say that it understands a model far better than any competitor who only got access to it on release. And that's important for a company like TCS because ever since Agenic AI really blew up earlier this year with Cloud Co-work, IT service companies like TCS are under an existential threat. And they even have a name for it. At the end of April, this year, this year, HCL Tech, Inphesis, PCS and WIPRO all announced their quarterly results. See Vijay Kumar, who is the CEO of HCL, called it AI deflation, which he explained as a dip in future revenue, of about 3 to 5% in the coming year and maybe even more in the future. Even PCS saw its revenue slip 0.5% year over year and Kikritivasan, its CEO, admitted that even he is

seeing the deflation, though he called it de-growth. The only part to survival for them? To reposition themselves as AI integration partners. And that's what the TCS Anthropic Partnership was all about. And it's not just TCS that's been making these moves either. For example, in mid-May this year, Xcel Tech invested $150 million in Sarvam AI, becoming the lead investor in Sarvum Series B round. Also, in February this year, as Anthropic was opening its first office in India, it announced

a partnership with emphasis. So clearly, for TCS, this strategic repositioning was in full swing. Except then, the fable mythos shutdown brought that swing to an uncomfortable pause and a realization. The frontier models that it had just caught in exclusive access to were suddenly gone. The thing is, India's IT sector has all. always depended on their access to the best of global technology. And AI has only intensified that same dependency.

So, when a single US decision limited access to two of the most advanced AI models in the world, it revealed that that access was a privilege with no protection. And while India focuses on building out its sovereign AI capacity, its IT industry has already built its survival pivot onto an inherent vulnerability and sovereign AI models alone are not a solution. Welcome to Daybreak, a business podcast from the Ken. I'm your host, Tretjevergis, and every day of the week, my co-host, Nikita Shramana will bring you one news story that is worth understanding and worth your time.

Today is Wednesday, the 24th of June. An exclusive deal with one of the world's leading AI companies should have been a breadth of relief for DCS. Instead, tech leaders across India are highlighting the sheer vulnerability of India's situation with regards to AI. And this time, it's not just about digital sovereignty for security reasons. It's about how one of India's largest exports, the arbitrage of IT companies like TCS, Infasis and HCL Tech

is entirely dependent on a foreign stack of systems. Nikhil Narendran, a tech policy thought leader and partner at law firm Trilegal, told E.T. in a report that India's IT services industry was built on the assumption of open access to the best global technology. That assumption has now developed a serious crack, which in turn is putting IT firms at a competitive disadvantage and also a deeper risk of strategic dependency. And the thing is, while sovereign AI is in fact the right long-term answer, it's hardly something that could have been of help in this particular situation.

You see, while sovereign AI in a very specific narrow sense will be able to. out soon in India in about three to five years, it won't really be able to replace the capabilities that the IT sector companies need, like cybersecurity, advanced coding, scientific work and complex agentic tasks. And the dependence on these capabilities have already been built out by these companies. At this stage, AI has become a necessary part of these companies' workflows. It's part of customer support systems, software development pipelines, analytics platforms, internal knowledge bases, content production workflows, and decision support systems. Timelines for projects involving all of these things get affected by complications in access.

And like Narendrin mentioned, affect the competitive advantage IT service companies have enjoyed over their American counterparts based on their low prices and quick work for years. And it also goes beyond AI that helps you with getting your work done quickly. models like Mythos 5, which is an advanced model that allows user to test for complex cybersecurity flaws, have been rolled out very exclusively and intentionally because of the fear of what could happen if it falls into the wrong hands. In fact, Mythos was live for about two months before a select few Indian companies were allowed to access it. In the meantime, E.T had reported in May that essential services providers such as banks, telecom operators, and power utility providers had started stress testing their codebases to prepare their defenses

against advanced cyber threats. These companies are still testing their customer-facing assets using the live AI models such as Opus 4.7 and GPD 5.5. And while those models are still live and these companies aren't entirely handicapped, imagine how bad things could get if the US were to decide arbitrarily that these models could also be cut off from foreign nationals. The foreign national aspect, by the way, raises even more complications. Abishar Prakash, an author and geopolitical strategist at Canada-based advisory firm,

told E.T. in a report that this rule determines who gets to access critical technologies. And that could have far-reaching implications for research and development, hiring, funding and collaboration. In the future, it could mean that American companies could hire fewer foreign nationals. It could mean less partnerships like that of Anthropic and TCS. For Indians working in the IT sector, that's a scary thought. And as a country, India doesn't really have many assurances to give. Stay tuned. In the first few days following the shutdown, the then-British Prime Minister Kare Stammer called Washington.

He was busy lobbying for a UK exemption to the ban, reportedly seeing the technology as vital. The argument Stalmers administration made was that access to these models was crucial for innovation and operations, especially for SMEs or small to medium-sized enterprises which are pretty dependent on cutting-edge technology. Those same dependencies echo in India, perhaps more so because the IT sector is extremely consequential to the country. To put that in perspective, here's a number for you. The India Brand Equity Foundation estimates that, India's IT industry is likely to hit the $350 billion mark by 2026 and contribute 10% toward the country's GDP. Still, there was no dialogue that was triggered between India and the US when the block began.

And this after India had gone in pretty heavily on tech diplomacy with the US. There was a trust initiative launched by Trump and Modi in February 2025, and even the India-US AI Opportunity Partnership signed in the same month. Also, at the G7 summit in France on June 17th, which was just last week, Modi even told world leaders, including Trump, that all democratic nations should have access to such AI models so that they can protect their critical information infrastructure and counter-growing cyber threats.

Is this a case of India simply not understanding the depth of the dependency, one of its most important industries has built over the last couple of years? It's hard to tell. Because another national response was also in the fact that India's India AI mission approved 12 projects in the same week as a shutdown. These 12 projects will focus on building different types of AI systems that will include language models, voice assistants and multimodal AI that can work with text, images and videos.

So clearly, the awareness of the vulnerability is there. and it has gained an urgency. But as analysts write, just building sovereign AI models is not enough because the dependency has already been built out. Still, there are solutions here. For companies, this means not depending on a single AI provider, because doing so exposes them to operational risks

that can occur exactly like the fable mythos shutdown, without warning. Chandrakant Agarwal, the co-founder and C. CEO of an AI solutions company told NDTV in a report that decisions like this can disrupt entire business processes that have been redesigned around AI-assisted execution. His lesson is this. Enterprises have to diversify critical technology dependencies. Anushri Varma, a senior director-analyst at Gartner, a research and advisory firm, explained in a Forbes piece that companies need to build something called model optionality as a

foundational requirement, which is basically using multiple models with primary fallback and sovereign options. For India, on the other hand, the solution is to go beyond building sovereign AI models. It needs to go into negotiating bilateral access agreements so that any disruption like this has a diplomatic fix. It also needs to make compulsory continuity clauses in any sort of agreement or partnership, so that there is a surefire plan for what happens when access is taken away. The thing is, India has leverage. It's both Anthropic and Open AI's second largest market. It already has deals it has signed for billions of dollars worth of data centers. All it needs to do now is build the right channels that can use that leverage to its advantage and

protect the interests of its companies while it builds out independence. Thank you.