NextFin News - When Anthropic filed its public IPO prospectus on October 26, 2026, the headlines wrote themselves: $16.3 billion of revenue in the first half of the year, a $965 billion private valuation, and a ticker — ANT — headed for the Nasdaq Global Select Market. But the filing carries a second story that matters almost as much, and it points to India. Anthropic now counts India as its second-largest market for Claude, and its most important Indian enterprise partner, Infosys, has spent 2026 building an entire practice around converting Anthropic's frontier models into billable client work. The IPO does more than put a price on an AI lab. It creates the public market's first real benchmark for frontier-model economics, and in doing so it forces a question that Indian IT services firms have been dodging for two years: is artificial intelligence a revenue engine for the companies that deploy it, or the force that eventually erodes the labor-arbitrage model they were built on?
The Filing: Growth at a Speed the Public Markets Have Never Priced
Anthropic, PBC filed its Form S-1 with the Securities and Exchange Commission on October 26, 2026, following a confidential submission in June that gave regulators a private look at the numbers before investors saw them. The company plans to list Class A shares on the Nasdaq Global Select Market under the symbol ANT, with Goldman Sachs and Morgan Stanley leading a 22-bank syndicate. The offering comes after a Series H round in May 2026 that valued the five-year-old company at $965 billion, with some market participants expecting a public valuation that could exceed $2 trillion.
The growth rates are what make this a landmark rather than merely a large IPO. Revenue reached $16.286 billion in the six months ended June 30, 2026, up from $1.163 billion a year earlier — roughly 1,300% growth. The second quarter alone produced $11.56 billion, more than Anthropic earned in its entire history through 2025 combined. Consumption-based revenue accounted for $13.1 billion, or 80% of the total, with the remaining $3.2 billion coming from subscriptions. That mix matters: consumption revenue is the most direct read on whether customers are actually using the models, not just signing contracts.
The retention figure is the one that stopped investors. Net revenue retention sat at approximately 900% as of September 30, 2026, meaning the qualified customer cohort is spending about nine times what it spent a year earlier. For comparison, the median public cloud software company reported 108% net revenue retention as of January, according to data compiled by Altimeter Capital partner Jamin Ball. The customer base is scaling with the revenue: organizations paying at least $100,000 on a trailing-twelve-month basis grew from more than 950 in September 2025 to more than 11,900 by September 2026, with roughly 1,900 customers at $1 million or more and about 250 at $10 million or more.
Growth this fast does not come without a cost line to match. Anthropic booked an $80.7 billion net loss in the first half of 2026, though $72.4 billion of that was non-cash fair-value changes on convertible notes, derivative liabilities and SAFEs — accounting volatility rather than cash burn. On an operating basis the picture is tighter: the company reported a $693 million operating loss for the half, down from a $2.135 billion operating loss a year earlier, and it posted $974 million of operating income in the second quarter, its first profitable operating quarter in the six quarters shown in the filing. Cash and short-term investments stood at $131.7 billion, backed by disclosed compute contracts with Google, Amazon, Microsoft, SpaceXAI, Broadcom, Nscale and Lambda that add up to roughly $637 billion.
That infrastructure stack is the bridge to the India story. Anthropic's net property and equipment reached $1.19 billion at June 30, 2026, up from $19 million at year-end 2024 — a more than 60-fold increase in eighteen months. Someone has to build, integrate and maintain the systems that sit on top of that capacity. In the United States, Anthropic does much of that itself. In India, it is doing something different: it is letting enterprise partners do it.
The India Thread Inside an American Prospectus
India is not an afterthought in Anthropic's expansion; it is the company's second-largest market for Claude, behind only the United States, and it ranks second globally in consumer usage of the assistant. The nature of that usage is what makes it strategically interesting. Nearly half of Claude usage in India involves building applications, modernizing systems and shipping production software — the most technically intensive category of use the company tracks. Educational and instructional tasks make up another 12% of Claude.ai usage in India.
Anthropic moved on that signal quickly. In February 2026 it opened an office in Bengaluru, its second in Asia Pacific after Tokyo, led by Irina Ghose as managing director for India. The company has been explicit about why it chose the city.
"India is compelling because of the scale of its technical talent and the commitment from the Indian government to ensure the benefits of artificial intelligence reach all areas of society, not just concentrated pockets,"CEO and co-founder Dario Amodei said during a visit to the country for the AI Impact Summit 2026.
"India's AI ecosystem will play a central role in how AI develops globally and democratically."
The expansion was paired with a push into local languages. Claude already supports major Indic languages and is adding enhanced performance in Hindi and nearly a dozen more, including Bengali, Marathi, Telugu, Tamil, Punjabi, Gujarati, Kannada, Malayalam and Urdu — a move aimed as much at public-sector adoption as at developers.
Enterprise adoption was already happening before Anthropic signed formal partners. CRED reported two times faster feature delivery and 10% better test coverage using Claude Code. Air India is using Claude Code to help developers ship custom software faster and at lower cost as part of a broader push into agentic AI. Cognizant is deploying Claude to 350,000 employees globally to modernize legacy systems. And on the social-impact side, education nonprofit Pratham is piloting an Anytime Testing Machine powered by Claude with 1,500 students across 20 schools, with plans to expand to 100 schools by the end of 2026.
Infosys: The Bridge From Model to Billable Outcome
The centerpiece of Anthropic's India strategy is its collaboration with Infosys, announced on February 16, 2026. The deal pairs Anthropic's Claude models, including Claude Code, with Infosys Topaz, the Indian IT firm's AI-first services platform, to build enterprise AI solutions across telecommunications, financial services, manufacturing and software development. The collaboration launches in telecommunications with a dedicated Anthropic Center of Excellence and is designed to expand across the other industries.
The focus is agentic AI — systems that go beyond answering questions to independently handling multi-step tasks such as processing insurance claims, generating and testing code, or managing compliance reviews. The two companies will also help clients modernize legacy systems, combining Topaz and Claude to make migration faster and reduce the cost of updating aging infrastructure.
The logic is clean on both sides. Frontier models are sold as API consumption, but regulated industries do not buy API access; they buy outcomes — a claims process automated, a compliance review completed, a network operation simplified. Infosys owns the domain expertise, the client relationships and the delivery workforce required to turn model access into those outcomes. Anthropic owns the model. The IPO prospectus prices the model; the Infosys deal is the distribution channel that determines how much of the model's economic value Anthropic captures and how much accrues to the services layer.
Infosys framed the partnership in exactly those terms.
"AI is not just transforming business, it is redefining the way industries operate and innovate,"said Salil Parekh, chief executive officer of Infosys.
"Our collaboration with Anthropic marks a strategic leap toward advancing enterprise AI, enabling organizations to unlock value and become more intelligent, resilient, and responsible."Amodei, for his part, was blunt about why a model company needs a services partner:
"There's a big gap between an AI model that works in a demo and one that works in a regulated industry — and if you want to close that gap, you need domain expertise. Infosys has exactly that kind of expertise across important industries."
The market read the announcement as offensive positioning rather than defensive capitulation. Infosys shares rose as much as 4.8% on the day the partnership was unveiled, outperforming a broader Indian IT index that had just endured its worst stretch in more than ten months on fears that AI tools would disrupt the industry's billing model.
Cyclical Tailwind or Structural Threat? The Call Indian IT Cannot Dodge
Here is where the cyclical-versus-structural judgment matters, because getting it wrong flips the conclusion. Indian IT services firms have spent two years telling investors that AI is additive — that it will raise bill rates and expand margins even as it automates routine coding work. The Anthropic-Infosys deal is the clearest test of that claim, and the evidence points in two directions at once.
The structural threat is real and it is specific. Agentic AI systems that write, test and debug code erode the labor arbitrage on which Indian IT was built. If Claude Code lets a team of 10 do the work of 50, the revenue base tied to headcount shrinks regardless of how many AI practices a firm launches. Anthropic's own operating model illustrates the endpoint: trailing-twelve-month revenue per employee rose from $1.4 million in September 2025 to more than $6.6 million in September 2026, and the company reported that Claude wrote more than 90% of contributions to its own codebase in September 2026. Those are not just productivity statistics; they are a signal of a company whose small team is amplified by its own product. A services firm whose revenue scales with headcount cannot compete with that leverage on the same tasks.
But the cyclical counter-force is also real, and it operates on a shorter clock. A survey of 838 AI platform decision makers conducted by Futurum Group in the first half of 2026 found that talent scarcity, cited by 56% of organizations, had overtaken compute costs and data privacy as the number-one barrier to AI adoption. Deployment, not model access, is the bottleneck — and that bottleneck is precisely what services firms sell. Legacy modernization, cloud migration, data integration and AI governance will keep large systems integrators busy for years, and firms that partner early with model leaders can capture the implementation spend before the automation wave reaches their own delivery centers.
The right read is that both forces are present and they operate on different horizons. In the short term, AI implementation is a genuine revenue tailwind for firms with deep enterprise relationships and industry-specific knowledge — exactly what Infosys is selling through the Anthropic Center of Excellence. In the long term, the same tools compress the headcount-multiplier model. The partnership is rational for both sides precisely because the short-term incentive — capture the AI services wave now — dominates a long-term threat that arrives gradually and unevenly across clients and use cases.
The Second-Order Trade: A Public Comparator for the Services Layer
The first-order effect of the IPO is obvious: Anthropic raises cash to fund compute. The second-order effect is less discussed and more important for India. Once ANT begins trading, it becomes the public market's reference price for frontier-model economics — revenue per employee, gross margin, the split between consumption and subscription revenue, and the loss profile investors are willing to tolerate. Anthropic reported cost of revenue of $2.38 billion against $16.29 billion in revenue for the first half of 2026, implying a gross margin of roughly 85% even while scaling infrastructure at pace.
For Indian services firms, that benchmark cuts the other way. Their revenue per employee is a fraction of Anthropic's, and their margins depend on the scale of headcount, not leverage from software. A richly valued Anthropic creates a public-market lens through which investors will increasingly ask Indian IT firms one uncomfortable question: if the model layer captures most of the value, what multiple does the deployment layer deserve?
This is the repricing risk hidden inside the India connection. It is not about how much revenue Infosys sends to Anthropic — the partnership carries no disclosed financial terms and may never move Anthropic's quarterly numbers in a visible way. It is about how much of Anthropic's enterprise adoption in India flows through Indian systems integrators, and whether that flow lifts the entire sector's multiple or merely concentrates value in the firms that own the client relationships while the work itself becomes cheaper to perform.
The Counter-Thesis
The strongest argument against reading too much into the India story is simple: it is immaterial to Anthropic's valuation. India is the second-largest market for Claude.ai, but consumer and developer usage is not the same as enterprise revenue. Anthropic's five largest customers represented about 25% of revenue in the nine months ended September 30, 2026, and those customers are overwhelmingly large technology and cloud companies, not Indian consultancies. The Infosys partnership, announced without disclosed terms, may never show up in Anthropic's results at all.
That objection is correct on Anthropic's own numbers, and it is the right discipline to apply to any hype around the deal. But it misses the point for the other side of the trade. The significance of the India connection is not the revenue Anthropic books from Infosys. It is what the partnership reveals about where value accrues in the AI stack as the technology moves from experimentation into regulated production systems — and what that means for the repricing of Indian IT. The falsifying signal is concrete and observable: if Infosys and its peers fail to convert AI partnerships into measurable revenue growth and margin expansion over the next four quarters — specifically, if AI-related revenue growth stalls while overall headcount continues to rise — then the "AI as tailwind" narrative is wrong and the labor-arbitrage erosion thesis wins. Until then, the market will price the transition wave, not the endpoint.
What to Watch
The IPO is being priced as an AI infrastructure story: how much compute, how many customers, how fast the loss narrows. The India-Infosys thread is the distribution story underneath it. Three signals will determine whether that thread matters:
- Anthropic's quarterly revenue per employee and net revenue retention once it begins public reporting — the metrics that define the model-layer benchmark.
- Infosys's disclosure of AI-related revenue and deal wins tied to the Anthropic Center of Excellence — the proof that the partnership converts into billable outcomes rather than announcements.
- The relative performance of Indian IT peers against the ANT ticker in the first year of trading — the market's verdict on who captures the value.
In the short term, firms with credible model partnerships — Infosys first among them, followed by Cognizant and others deploying frontier models into enterprise workflows — are likely to benefit from implementation demand and from the sentiment lift of association with the market's most watched technology asset. Over a medium-term horizon, the split widens: firms that build reusable AI delivery assets and industry-specific agents can defend margins, while firms that simply resell model access as their headcount-multiplier model erodes will face multiple compression. Over the long term, the structural question resolves against the labor-arbitrage model unless Indian IT reinvents what it sells. Agentic AI that writes and tests code does not need the same pyramid of junior engineers. The winners will be the firms that move up the stack into domain-specific outcomes — the exact strategy Infosys described when it said the collaboration would target telecom, financial services and manufacturing.
The IPO prices the model. The next act decides who captures the value chain beneath it — and India's services giants are betting they can own that ground before the code writes itself.
Explore more exclusive insights at nextfin.ai.