The Indian IT firms watch their cost-arbitrage edge fade, and build fastest.

Every consulting firm calls itself AI-first, and the phrase means something different depending on who says it. In earnings calls, every firm is AI-first; in press releases, every firm has a named platform; in analyst briefings, every firm is a technology company that happens to deliver through people. None of that is false, and none of it is useful. The honest segmentation is not by firm size or tier, it is by what AI actually threatens in each business model, and therefore by how motivated each category really is to build something real. There are four different fires here, and the response to each is different.

IT services is in survival mode. Wipro, Infosys, and Cognizant were built on a single structural advantage: Western enterprises needed software at scale, offshore talent delivered it at a fraction of the cost, and the firm captured the spread. That cost-advantage model ran for thirty years, and GitHub Copilot, Cursor, and Claude Code are collapsing the spread in real time. You do not need ten thousand offshore developers when AI does the boilerplate. The core value of the model is not weakening, it is evaporating. Wipro Foundry IQ, Infosys Topaz, and Cognizant’s Project Leap restructuring are genuine attempts to find a reason to exist. Wipro absorbed a $3.1 million restructuring charge and an 11 percent first-quarter revenue decline to make the platform-as-delivery pivot, and Cognizant paired a $600 million acquisition of Astreya with a broader-and-shorter pyramid restructure that compresses the very labor model generating its current revenue. Those are expensive, painful moves, which is exactly what separates them from firms issuing press releases. Survival mode produces the most motivated builders in the space, and the signal that an internal build has failed will be in the language: when Wipro stops calling Foundry IQ a platform and starts calling it a managed service built on third-party AI infrastructure, the channel-partner window, roughly 12 to 18 months out, opens.

The Big Four are in urgent adaptation, not survival. Deloitte, EY, PwC, and KPMG have a protected revenue base the IT services firms lack, because the audit regulator still requires a human to sign the opinion and the compliance regime does not disappear. The audit franchise is a regulatory moat, not permanent but real, and it gives them room to experiment without existential pressure. Deloitte Zora AI and PwC Agent OS are real products and also insurance policies, because having a named AI product is now table stakes to stay in the room for advisory work. The protection is partial: advisory and consulting is often 50 to 60 percent of Big Four revenue at the firms with the largest AI practices, so they have one protected leg and one threatened leg. The right diagnostic is the runs-in-production test. EY’s client-zero rollout to 130,000 auditors on live engagement workflows is a production deployment, not a pitch tool, and the firms that get to production first on agentic AI will own the governance advisory narrative, because a governance claim is only as credible as the governance practice behind it.

Top-tier strategy faces a subtler erosion. McKinsey, BCG, and Bain are not in survival mode; their margins are still extraordinary, their relationships still run at the board level, and their alumni networks still move capital and careers. What is under pressure is more dangerous over a longer horizon: the opacity of their method was the product. A $15 million AI readiness engagement is harder to justify when a client’s CTO can get a competent analytical answer from Claude in an afternoon. The question is not whether these firms produce better analysis, because they often do and their benchmarking data at scale is not replicable from a prompt. The question is whether the premium survives once clients can separate the analytical component from the political one. The analysis is being commoditized; the political access, the partner who has sat in 400 similar boardrooms and manufactures the buy-in to act, is not. McKinsey Agentic Mesh and BCG X are not really analytical products, they are attempts to rebuild that hard-to-copy advantage at a new layer before the client realizes the old one is gone.

The finance boutiques are quietly the most in danger. The Hackett Group, ISG, and Riveron are not in the headlines, not printing dramatic AI pivots, not acquiring, not restructuring, and that is the problem. The Hackett benchmark was worth something because of the friction: months of data collection, human analysts running interviews, proprietary methodology built over years. That friction was the product, and AI eliminates it. ISG’s market-research model, collecting and synthesizing vendor and buyer data to produce sourcing advisory and contract benchmarking, is a direct casualty of AI at scale. These are not firms facing an adjacency threat, they are firms whose core deliverable is being automated. The IT services firms know they are in trouble and are spending money to respond; the boutiques are hoping the transition is slow enough to adapt client by client. The Hackett platform pivot is a genuine repositioning and Fernandez’s workflow-intelligence framing is analytically correct, but these firms have neither the brand nor the balance sheet to survive an accelerating transition. They are the most exposed category with the least runway.

Accenture does not fit the segmentation cleanly, which is itself the signal. It is too delivery-focused to be top-tier strategy, too engineering-heavy to be IT services in the traditional cost-advantage sense, and too large for the boutique category. With 700,000 engineers and enterprise C-suite relationships, Accenture is the one firm that could most credibly compete with DeployCo on day one, because it has the delivery muscle, the client access, and enough engineers to build the specification layer the middleman thesis assumed would stay inside the firm. It was never purely a cost-advantage business; forward-deployed engineering, center-of-excellence build-outs, and managed services were always part of the model, so the AI transition reprices some of the offering and potentially expands the rest. The diagnostic that resolves production versus press release is the restructuring charge. Wipro, Hackett, and Cognizant have all paid one. McKinsey and BCG have not, which is consistent with their posture: they are not pivoting under pressure, they are extending the premium under erosion.