The Anthropic-Indian IT Partnership Is a Two-Way Bet

Anthropic’s S-1 filing spends a long passage on competition. It names model developers such as OpenAI and Google, and first-party application providers. IT services firms do not appear.

The passages on partners cite systems integrators (SIs) such as Accenture, Deloitte, KPMG and PwC. None is Indian.

Yet the filing describes much of the work Indian IT bills by the hour. Claude Code writes, tests and debugs code across production codebases. Claude Cowork handles knowledge work.

Meanwhile, IT services firms Tata Consultancy Services (TCS), Infosys and LTM have signed on as Claude partners. These partnerships give Anthropic access to their engineering workforces and established client relationships. But the lab is also building capabilities that could eventually compete with them.

Why Anthropic Needs Systems Integrators

The filing is open about why Anthropic needs partners. It identifies a shortage of engineers who can take an AI pilot into production as a constraint on enterprise adoption.

Its answer is a network. The Claude Partner Network launched in March with $100 million (~₹9.5 billion) for training and certification. By September 30, over 150,000 practitioners across more than 4,000 organisations held Claude certifications.

In October, a second $100 million commitment, the Frontier Academy, followed. It aims to train 10,000 engineers by the end of 2027, beginning with cohorts from consulting firms and enterprises.

Indian IT firms are also building their own Claude-trained workforces. TCS joined as a Global Premier Partner in June and is rolling Claude out to 50,000 employees. Infosys partnered with Anthropic in February.

LTM followed in July, adding Claude, Claude Code and Claude Cowork to its BlueVerse platform and expanding its AI1000 programme to train Claude engineers. Chris Ciauri, Managing Director of International at Anthropic, said LTM brings “delivery expertise, trained people, and long-standing client relationships.”

Ray Wang, Founder, Chairman and CEO of research firm Constellation Research, tells AIM that frontier labs cannot yet implement and rely heavily on services firms to deliver outcomes. They should treat IT firms as “strategic channels for enterprise experience, front line feedback, and co-development,” he says.

Anthropic Builds Its Own Deployment Capabilities

Anthropic is also developing the capacity to serve enterprise customers directly.

It holds direct contracts with its customers, even where Claude is sold through Amazon Web Services (AWS), Google Cloud or Microsoft Azure. Its Applied AI teams, organised by geography and industry, design custom solutions for select customers. It also launched Ode with Anthropic, an AI services company backed by private equity firms to bring Claude to mid-sized companies.

Ode is a $1.5 billion (~₹142 billion) venture. OpenAI has a Deployment Company backed by more than $4 billion (~₹378 billion), and Microsoft has a $2.5 billion (~₹237 billion) Frontier Company staffed by 6,000 forward-deployed engineers (FDEs). Indian IT stocks fell after both announcements.

DD Mishra, VP Analyst at research firm Gartner, tells AIM that Ode signals a direct shift towards high-end consulting and implementation. Labs partner with firms such as TCS and Infosys today for global distribution, a strategy Mishra calls “coopetition.” They will inevitably clash with SIs over agentic AI transformations “in due course,” he says.

Wang sees these partnerships as a way for frontier labs to extend their reach into enterprises. Mishra expects the relationship to become more competitive as labs take on a larger share of consulting and implementation work.

Anthropic did not respond in time to AIM’s questions on how it defines its relationship with IT services providers, and how Ode fits alongside its SI partnerships. Its filing describes both its partner ecosystem and its direct implementation efforts, but does not clarify how the company expects these relationships to evolve.

AI Productivity Is Not Translating Into Revenue Yet

Anthropic's growth illustrates the economics of an AI-native business. It reported $11.6 billion (~₹1,097 billion) in revenue for the three months ended June 30, up from $787 million (~₹74 billion) a year earlier.

TCS reported $7.6 billion (~₹719 billion) for April to June. TCS, Infosys, HCLTech, Wipro and Tech Mahindra together reported $25.6 billion (~₹2,422 billion). Anthropic's quarterly revenue was roughly 45% of the combined total.

The difference is sharper on a per-employee basis. Anthropic's filing puts its trailing 12-month revenue per employee at over $6.6 million (~₹624 million). At TCS, Infosys and HCLTech, it ranges from roughly $51,000 to $65,000 (~₹4.8 million to ₹6.1 million) a year, more than 100 times lower. The comparison is not like-for-like: Anthropic is a compute-heavy AI lab, while Indian IT firms rely on large delivery workforces.

Wang says AI labs and AI-native firms are already starting to take market share. He says revenue per employee has made a huge difference.

For Indian IT firms, higher productivity has yet to translate into commensurate revenue growth. Infosys cut its FY27 growth guidance to 1% to 2% despite $5.2 billion (~₹492 billion) in large-deal wins, with AI accounting for 10.5% of revenue. Its largest new deal will contribute revenue only from the March quarter.

Analysts tracking the sector say AI-led price cuts still outweigh AI-driven revenue gains. Clients are seeking to capture part of the efficiency gains through lower contract prices, leaving services firms with less opportunity to convert productivity improvements into higher revenue.

Gaurav Vasu, Founder and CEO of research firm UnearthInsight, estimates that about 60% of the rise in revenue per employee at the big firms comes from workforce rationalisation, with the remaining 40% attributable to AI efficiency gains.

Enterprise Context Remains an Advantage for IT Services Firms

Indian IT firms have an advantage in their understanding of enterprise systems, business processes and legacy infrastructure. Frontier labs can build powerful models, but deploying them across complex enterprise environments requires domain knowledge and an understanding of how businesses operate.

“The owning of the IP itself is meaningless,” Ravi Vasantraj, Global Delivery Head at IT services firm Mphasis, said in an earlier conversation with AIM. Enterprises will buy outcomes that use AI, not AI services, a distinction he expects to land within 12 to 24 months.

Pricing is moving in the same direction. Fixed-price work, including outcome-based contracts, now makes up more than 65% of Accenture’s bookings. The shift predates AI, but AI sharpens it. Accenture’s own annual filing warns that AI could shrink demand for its services.

Building that capability also requires a different kind of workforce. TCS plans to put 1% of its workforce, nearly 6,000 engineers, in forward-deployed roles. Infosys targets 6,000 frontier engineers. Talent firm Xpheno puts India’s potential FDE pool at around 2,800, against targets in the tens of thousands.

Anthropic’s filing also describes how quickly adoption can accelerate once a model becomes reliable enough for production use. Coding has already reached that point, while cybersecurity is emerging as another area of application.

Mishra says which provider keeps the work depends on the client. Regulated industries often go in-house or straight to the lab to enforce data rules. Complex legacy estates favour integrators, who also serve as the “human-in-the-loop” governance layer. Over the long run, he adds, responsibility keeps shifting towards agents.

His test for whether labs are taking business away is a consistent pattern of clients shifting spend and workflow ownership. Until then, he classes the impact as productivity gains and competitive risk, not confirmed diversion. He says the shift from labour arbitrage to intelligence arbitrage is underway, and “time is running out.”

Tech Mahindra will disclose AI revenue separately in the second half of FY27. Mphasis reports its September-quarter results on November 5. Their disclosures could offer further evidence of how Indian IT firms are translating AI investments into revenue.

For Indian IT firms, the question is how much enterprise implementation work they can retain as frontier labs build their own delivery capabilities.