Anthropic Mythos Sparks Revenue Deflation Fears in Indian IT

Concern in India’s $250 billion information technology services sector is no longer theoretical. It now has a name, ‘Mythos’.

Anthropic’s latest frontier model, Mythos, still unreleased to the public, is forcing a fundamental rethink of how much human labour global tech services will require and how quickly that demand could shrink. The anxiety follows the company’s earlier market shock, when fears of a ‘SaaSpocalypse’ sent IT stocks sliding.

However, several IT leaders and industry bodies have called such concerns misplaced. Nasscom says Indian IT firms remain deeply embedded in complex enterprise environments with interconnected systems and fragmented data. 

“Creating real business value from AI requires careful coordination, with humans in the loop who understand business context, industry knowledge, and enterprise workflows,” it says, adding that AI is unlikely to be an out-of-the-box solution for large enterprises. 

As adoption scales beyond pilots, Nasscom argues, technology services companies will continue to play a critical role in enabling enterprise-wide AI deployment.

Yet, a recent Kotak Institutional Equities report accessed by AIM describes Mythos as a “step-jump in benchmark performance across software engineering tasks”, marking “a deviation from the trajectory of incremental or moderate improvements in the recent past”. That deviation is precisely what is unsettling the industry.

From Coding Assistant to Autonomous Engineer

Unlike earlier AI tools that augmented developers, Mythos signals a shift towards autonomy. The model can be given an engineering goal and handle all steps, investigation, implementation, testing, and reporting, allowing users to 'set and forget’ tasks for hours.

This is not just improved productivity, but it redefines the role itself.

The model’s ability to act like a senior engineer, catching subtle bugs and fixing root causes, suggests that large parts of traditional coding, testing, and maintenance workflows could be compressed or eliminated.

Biswajeet Mahapatra, Principal Analyst, Forrester, views Mythos as a radical "discontinuity" that automates vulnerability discovery, crushing manual L2/L3 support and testing. 

While high-end consulting remains safe, Indian IT firms face margin pressure, forcing a shift to outcome-based models to survive, he says.

Motilal Oswal notes that Mythos “further expands the list of things AI can do better than humans–coding, ERP, and now cybersecurity”.

Legacy Complexity Slows Enterprise Adoption

What makes Mythos uniquely disruptive is not just its coding capability, but its ability to break software. Anthropic says the model can autonomously identify and exploit vulnerabilities, including decades-old flaws. In one instance, it reportedly identified a 27-year-old vulnerability in OpenBSD.

The brokerage firm highlights that the model is extremely good at identifying and fixing cybersecurity vulnerabilities, outperforming human experts and existing tools, even spotting bugs that remained undetected for decades.

This dual capability, builder and breaker, has prompted caution even from its creators, who have withheld a full public release over concerns it could aid malicious actors.

For IT services firms, this raises a deeper concern. If AI can compress detection cycles from years to hours, large portions of testing, audit, and security services revenue could shrink.

The biggest exposure is in application services, the backbone of Indian IT. Kotak warns that risks are higher for firms heavily exposed to this segment, with uneven productivity gains.

This matters because application development and maintenance account for a substantial share of revenues for firms such as Infosys and TCS, as well as mid-tier players.

More critically, the report warns that AI-led efficiency gains could lead to pricing pressure. “Mythos could pressurise valuation multiples of IT services companies while “compounding near-term deflation risks for services,” it says. 

Kotak estimates a 3–3.5% annual growth headwind for the industry over the next three years due to GenAI adoption, adding that this could worsen if capability gains accelerate further.

At the heart of these concerns is what analysts call ‘effort compression’.

Billing Models Face Direct Disruption

Traditional IT services rely on large teams, long project cycles, and billing linked to human effort. Mythos challenges all three. Motilal Oswal says the model could reduce manual-heavy security services, compress time-to-detection, and address legacy code complexity far faster than human teams.

In practical terms, work that once required dozens of engineers over weeks could be executed by AI in hours with fewer people overseeing it. This shifts the industry from labour arbitrage to outcome-based delivery, an uncomfortable transition for incumbents built on scale.

The disruption, however, is not one-dimensional.

Kotak notes that AI could also unlock new demand, particularly in legacy modernisation, data transformation, and AI-native development. Motilal Oswal adds that enterprise adoption may lag, as deploying AI in brownfield environments with legacy systems built over 20–30 years remains complex.

AI capabilities are advancing quickly, but enterprise transformation in Indian IT markets will take time, as industry leaders mention in earnings calls. The industry’s future depends on how well it monetises this transition, balancing deflation pressures with new demand.