The Cult, Sorcery & Over-confidence of Indian IT - Part 1
- Normal Guy
- May 8
- 34 min read
Updated: Jul 13
Disclaimer: This is not a buy or sell recommendation. I’m not a SEBI-registered investment advisor, and nothing shared here should be taken as financial advice. This content is purely for educational and informational purposes. I’m not affiliated with SEBI or any brokerage, nor am I trying to promote or sell anything. Always do your own research before making any investment decisions.
Note: You can get 15% off on StockScans by using the code "NORMALGUY". It gives access to a lot of useful data, in-built scans, sector and index rotation, price data, announcements, and many other insights in one place. Personally, I use it and like it a lot. You can also renew your subscription using the same code. For anyone who is actively researching stocks and tracking market trends, I feel it is worth considering.
I was going to drop a comment on this, but then I became selfish while reading it. I felt, let's go deep dive into this and explore the dots. What is really going on this quarter? I also wanted to understand this industry in depth.
Well, my exposure to this sector is in single digits :)
Before reading this, read Ameya article and his thoughts on the IT sector domain. He is also an expert who looks deep into the IT domain. Follow him on X too. This will help Yu to understand the whole story.
2023 View of TCS CEO - Following the release of ChatGPT and the subsequent rise of Generative AI, TCS CEO K. Krithivasan (who took over in 2023) has maintained a forward-looking, proactive stance, viewing AI as a significant opportunity rather than an existential threat.
Reality Check of Quarter - 4 - 2026
Every single company, whether IT services giant (TCS, Infosys, Wipro), mid-cap (Persistent, Coforge, Mphasis), niche product (Intellect, KPIT, Aurionpro), or specialized (Indegene, IKS Health, Tata Elxsi) - is talking about AI.
What's actually happening:
AI is creating a two-sided force simultaneously:
Compression side: Existing work is getting done with fewer people. Code generation is 25-40% AI-assisted already. Testing, maintenance, documentation, shrinking per-unit cost.
Growth side: New work is opening up, agentic AI, legacy modernization (suddenly affordable), data readiness, physical AI, cybersecurity.
The net effect right now is growth is barely keeping pace with compression for most companies. That's why you see TCS at -2.4% CC growth, Infosys at 3.1%, Wipro at -1.6%, HCL at 3.9%. The compression is real and happening NOW. The growth side is still early days and pilots moving to production.
WHO IS WINNING AND WHY
Tier 1 Winners (clearly gaining share):
Company | Growth | Why Winning |
Persistent | 17.4% YoY (USD) | SASVA platform, AI-native from day one, BFSI focus, mid-cap agility |
Coforge | 29.2% YoY (USD) | Execution intensity, travel/BFS depth, Cigniti integration, aggressive deal machine |
Mphasis | 6.7% CC, accelerating | NeoIP platform, BFSI focus, decision intelligence acquisition, outcome-based pricing |
Cognizant | 3.9% CC | Strongest bookings in years, AI Builder positioning, TriZetto healthcare moat |
Accenture | 4% LC | $22B record bookings, balance sheet power ($5B acquisitions), full-stack AI |
Tier 2 - Stable but not exciting:
Company | Growth | Situation |
TCS | -2.4% CC | Largest but slowest. Client-specific issues. Strong order book but conversion slow |
Infosys | 3.1% CC | Guiding 1.5-3.5% for FY27. European manufacturing client ramp-down. Productivity pass-throughs |
HCLTech | 3.9% CC | Two telecom clients cut discretionary spend. Software business declining. 2-3% AI deflation acknowledged |
LTIMindtree | 5.3% CC | Top client concentration risk. Good Europe growth. Doubling revenue in 5 years target |
Tier 3 - Struggling or transitioning:
Company | Growth | Problem |
Wipro | -1.6% CC | Americas 2 weak. BFSI delayed ramp-ups. Still rebuilding |
Tech Mahindra | 1.3% CC YoY | Turnaround story. Margins improving. Comms mega deal won. But base still weak |
Tata Elxsi | 0.9% QoQ CC | Healthcare deal delays. Auto flat. Media volatile. Small company, concentrated bets |
KPIT | Negative organic | Auto industry spending down 20-25%. Pivoting to solutions. Next 12-18 months transition |
Cyient | -0.7% CC (DET) | Paused large M&A. Semiconductor fundraise. Buyback signals undervaluation |
LTTS | Flat organic | Divested SWC. Portfolio cleanup done. 13-15% CAGR aspiration over 5 years |
Niche Winners (different game entirely):
Company | Growth | Edge |
Indegene | 23.6% INR | Life sciences operating partner. 27 years domain. Revenue per employee $75K. Tectonic AI platform |
IKS Health | 24% YoY INR | US healthcare task elimination. 1.5% headcount growth for 24% revenue growth. Platform moat |
Intellect Design | 23% LTM | Banking software. eMACH.ai platform. Purple Fabric AI. INR 3000 crore LTM crossed |
Aurionpro | 26% 9M YoY | Banking fintech + transit + data centers. AurionAI launch. Massive pipeline growth 65% |
RateGain | 94% YoY (with Sojern) | Travel tech. Sojern acquisition. $12M synergies in 100 days. AI Concierge traction |
HCLTech CVK was the most honest - 3-5% deflation in AI-disrupted services, translating to 2-3% for our portfolio. That's the first time a CEO put a number on it.
Infosys said productivity pass-throughs will eat margins. TCS said nothing will change. Accenture said AI is a tailwind. But read the numbers - TCS revenue declined 2.4%, Infosys grew 3.1% on flat volumes. The deflation is already in the numbers. They just don't want to call it that because the stock market punishes honesty.
The truth is If you're growing below 5% in CC terms today, AI compression is probably eating 2-4% of your growth. The companies growing 15%+ (Persistent, Coforge) are growing DESPITE this compression - meaning their actual gross wins are even higher.
The Large Deal Game Has Changed
Almost every company is reporting record large deal Total Contract Value (TCV). But look deeper:
HCLTech: Deal TCV is flat, but technically required 25-30% more effort to convert and get to the same number. AI deflation is shrinking deal values.
Coforge: $1.75B executable order book, 16.4% higher YoY. But they're also dropping $20M/quarter of low-margin India business to protect margins.
TCS: $12B TCV in Q4, but revenue declined. Large deals take time to ramp, and older deals are completing.
Cognizant: 21% bookings growth, 7 large deals >$100M. But ACV was flat because deal duration increased.
The hidden dynamic is Companies are winning bigger, longer deals (3-5 years) with more AI productivity baked in. Year 1 is often cash-negative (transitions, rebadging). The real revenue comes in year 2-3. This creates a lag between bookings and revenue that investors don't fully appreciate.
The Pyramid Is Being Destroyed
Multiple companies are talking about reshaping the pyramid or moving from effort to outcome:
Cognizant launched Project Leap - $200-300M savings, 2/3 reinvested. This is essentially a restructuring disguised as transformation.
Accenture: We will hire MORE entry-level reinventors than last year - broadening the base while AI replaces the middle.
TCS: Announced salary increments for all grades but total headcount barely moved.
Persistent: 27,500 headcount, 88% utilization. Growing 17% with disciplined hiring.
Mphasis: We have not increased headcount significantly. Revenue per employee is up.
The traditional IT services model of hire 100 people, bill 100 people is dying. The new model is hire 50 people, deploy 20 AI agents, bill for outcomes. Companies that figure this out fastest will win. Companies stuck in the old model will see margin compression.The Real Competition Is Not Who You Think
For IT services: The real threat is not other IT companies - it's the client's own GCC (Global Capability Center). Every major bank, auto company, pharma company now has a GCC in India. These GCCs are hiring the same talent at the same cost.
For banking software: Intellect real competition is Palantir and C3.ai on the AI side, and the existing core banking vendors (TCS BaNCS, Temenos) on the product side. But neither has what Intellect has - a full-stack AI-first platform for banking.
For automotive ER&D: KPIT and Tata Elxsi are competing against the OEM own in-house teams who are now under cost pressure. The OEMs tried to build everything themselves (2019-2023), failed, and are now coming back to partners. This is a structural tailwind.
For healthcare outsourcing: IKS Health competition is literally the in-house staff of US hospitals. The $260B TAM is mostly work done internally. Only $34B is outsourced. The penetration opportunity is massive.
Europe Is the Battleground
Multiple companies reported Europe growing faster than Americas:
LTIMindtree: Europe 12.4% vs Americas 4%
Persistent: Europe 26.7% vs North America 17.2%
Coforge: UK public sector $150M deal
Accenture: EMEA growing
Tech Mahindra: Europe 11.2% YoY
Why - European companies are under massive cost pressure (especially German auto, manufacturing). They're consolidating vendors and moving work offshore to India much faster than before. The sovereign solutions theme (data residency, local compliance) is creating new demand. Also, European OEMs are losing market share to Chinese EVs, forcing them to cut costs and outsource more.Healthcare Is Bifurcating
Healthcare payers (insurance companies): Under pressure from regulation (Medicare Advantage rate changes, prior authorization rules). Driving demand for automation, cost reduction.
Healthcare providers (hospitals, clinics): Under extreme margin pressure. IKS Health story shows how they can double a hospital EBITDA by automating chore tasks.
Pharma/life sciences: Indegene is eating agency and CRO budgets. Three months to three days for creative briefs. This is genuinely disruptive.
MedTech: Slower AI adoption due to regulation. LTTS, Tata Elxsi working here but growth is modest.
The US healthcare system spends trillions on administrative waste. AI can eliminate massive chunks of this. Companies positioned here (IKS Health, Cognizant's TriZetto, Indegene) have a multi-decade runway.
RED FLAGS AND CONCERNS
Cash Flow Disconnect at Mphasis
Mphasis guided 80% OCF/PAT - below their historical 100%+. The reason is large deals require upfront working capital investment (contract acquisition costs, transition investments). This is a structural change. If growth slows, this invested cash may not come back quickly.
Cyient's Paused M&A
Cyient paused a transformative acquisition because of AI uncertainty and geopolitics. They spent INR 71 crore on due diligence costs alone. This suggests the deal was massive. The fact that they paused because AI changed the calculus tells you how fast the ground is shifting even for engineering services.
KPIT Organic Decline
KPIT had negative organic growth in Q4. The auto industry's R&D spend dropped 20-25%. KPIT is pivoting to solutions-based delivery, which is the right move, but the transition will take 12-18 months. During this period, growth will be bumpy.
Wipro Persistent Weakness
Wipro Americas 2 (BFSI-heavy) has been weak for multiple quarters. They keep saying delayed ramp-ups but the delays keep extending. The top client declined. The company is doing better in APMEA and Europe, but the US BFSI franchise needs serious attention.
HCLTech Software Business
HCLTech software segment declined 4.1% for the full year. ARR was essentially flat. The experience portfolio (older products) is dragging. This is a meaningful part of their business and needs investment to stabilize.
Infosys Lost a European Manufacturing Client
Infosys explicitly said they walked away from a deal at a European manufacturer because returns didn't meet expectations. They also said the client is ramping down due to tough macro. This 75-100 bps headwind to FY27 growth is significant. When the #2 player walks away from deals for return reasons, it suggests irrational competition is happening.
MOATS AND ENTRY BARRIERS
Company | Moat Type | Strength |
TCS | Scale + client relationships + talent | Wide but eroding. GCCs are a threat |
Accenture | Balance sheet + consulting + acquisitions | Widest moat. $5B acquisitions per year. Nobody can match this |
Indegene | 27 years life sciences domain + regulated workflows | Very deep. Agencies/CROs can't match tech. IT firms can't match domain |
IKS Health | 150,000 US providers + outcome-based model | Deep. 18 years of integration. System of record relationships |
Intellect Design | eMACH.ai platform + banking domain | Moderate. Core banking is sticky. 25-year customer relationships |
Coforge | Travel + BFS depth + execution culture | Moderate-high. 10-year track record of consistent delivery |
Persistent | SASVA platform + BFSI engineering DNA | Growing. AI-first approach resonating. Platform differentiation |
KPIT | Auto domain + 16 years OEM relationships | Deep in auto. But auto industry itself is disrupted |
Aurionpro | iCashpro+ (transaction banking) + transit IP | Narrow but growing. Make in India capability for transit hardware |
WHERE THE POWER AND LEVERAGE SITS
Power has shifted to the client in traditional IT services. Clients have options: GCCs, multiple vendors, AI tools. This is why pricing pressure exists.
Power is with the vendor in three specific scenarios:
Regulated domains where AI can't just be deployed without human oversight (healthcare, pharma, banking compliance)
Platform/IP plays where the vendor owns reusable technology (Intellect's eMACH.ai, IKS's platform, Indegene's Cortex)
Outcome-based models where the vendor takes operational risk (IKS Health's NEVA model, Mphasis's NeoIP deals, Indegene's Tectonic)
The most powerful position is when you combine all three: regulated domain + proprietary platform + outcome-based pricing. IKS Health and Indegene are closest to this. Traditional IT services companies are furthest away.
WHAT'S COMMON ACROSS ALL
Every company says AI is embedded in everything we do but most have <10% of revenue directly from AI services
Every company is talking about agentic AI - it's the buzzword of 2026
Every company mentions the new Labor Code - one-time gratuity provision hit Q3/Q4 across the board
Every company says pipeline is strong but conversion to revenue is slow
Every company is restructuring - Cognizant (Project Leap), Coforge (India business shutdown), LTTS (SWC divestment), Cyient (portfolio rationalization), TCS (undisclosed restructuring)
Every company talks about outcome-based pricing but most revenue is still effort-based
Rupee depreciation helped everyone on margins this quarter, but most reinvested the gains
WHERE IS HOPE (Positive)
Legacy modernization is unlocking: Mainframe-to-cloud cost dropped from $10/line to $1.50/line (KPIT example). This is a massive new market.
Physical AI is emerging: Data centers, manufacturing, robotics, automotive - AI moving from digital to physical world.
Mid-cap companies are taking share: Persistent, Coforge, Mphasis growing 2-3x industry rate. Scale advantage of Tier 1 is diminishing.
Healthcare/life sciences AI: Genuine transformation happening. IKS Health, Indegene proving the model.
India as a market: Transit (Aurionpro's MMRDA deal), defense, data centers, banking modernization - domestic demand is real.
WHERE IS CONCERN (Negative)
AI deflation will accelerate: As models improve (Anthropic Opus, GPT-5), compression in testing, coding, maintenance will increase. 2-3% today could become 5-7% in 2-3 years.
Geopolitical uncertainty: Middle East war, tariffs, US-China tensions - all creating decision paralysis in pockets.
Auto industry restructuring: European OEMs losing share to Chinese EVs. Japanese OEMs delaying programs. Auto ER&D spend down 20-25%.
Wage-growth disconnect: Companies are not hiring proportionally to growth. This is good for margins but concerning for the broader employment picture and future talent availability.
Concentration risk: Several mid-caps have heavy dependence on top 1-2 clients (Mphasis, KPIT, Tata Elxsi, LTIMindtree). One client issue can derail a quarter.
The IT services industry is at an inflection point similar to 2005-2008 (when cloud disrupted traditional outsourcing). The winners will be companies that:
Own platforms/IP (not just labor)
Price on outcomes (not effort)
Operate in regulated/complex domains (not commoditized coding)
Move fast on AI (not just talk about it)
Have financial strength to invest through the transition
Right now, the companies best positioned across these five criteria are: Accenture (global), Persistent and Coforge (Indian mid-cap IT), Indegene and IKS Health (domain-specific), and Intellect Design and Aurionpro (product/platform). The large-cap Indian IT companies (TCS, Infosys, Wipro) are navigating but not leading - their size is both their safety net and their anchor.
PART 2
The Article by Ameya makes a strong case for Coforge and Persistent. Which I given the link even in Intro :)
Most of what it says is factually correct. I will just add few point into this.
The growth gap is real. TCS at -2.4% CC vs Persistent at 17.4% vs Coforge at 29.2%, these numbers come straight from the concalls.
The large deal market share shift from 70% to 54% for top 5 - this matches what we heard in the concalls. Coforge Sudhir Singh literally said on the call 19 of our top 20 client wins came against large IT companies. He also said their executable order book is $1.75B, up 16.4% YoY.
The revenue per employee gap - Persistent at $88K vs TCS at $50K - also confirmed. In fact, Indegene is at $75K with only 5,000 people. The mid-tier and specialist companies genuinely extract more value per person.
Coforge organic growth is much lower than 29.2%
On the concall, an analyst directly asked: What was organic growth in constant currency in Q4?
The answer was 12% organic. For a company reporting 29.2% headline growth, only 12% is organic. The rest is Cigniti acquisition. That 12% organic is good - better than large caps - but it's not 29%.
Also, Coforge is shutting down $15-20M per quarter of low-margin India business. They said this on the call. They're cleaning up the base to make the organic number look better going forward. Smart, but it means Q1 FY27 will be flattish by their own admission Perhaps.
Persistent growth is decelerating
Controlled deceleration from 23% to 21% to 18.8%. Now in FY26 it's 17.4%. That's four straight years of slowing growth. Still excellent, but the trend is clearly downward, not accelerating. On the call, Sandeep Kalra was very careful - he said $2 billion run rate by Q4 FY27 but refused to give annual guidance. When pressed about FY27 growth, he only said we are marching towards it steadily, plus minus a quarter at worst.
That plus minus a quarter is an important hedge. It means they're not 100% sure they'll hit the $500M quarterly run rate by March 2027. Who knows the condition is like that investor have to accepts as of now. They not even showing off how much they now holding skin in the Game.
The margin math is not okay
Coforge at 17% margin with 28% growth will generate more absolute profit than TCS at 25% margin with 3.8% growth by year 3. This math is technically correct BUT it assumes Coforge maintains 28% growth for three years. That's extremely unlikely given their own organic growth is 12%. Once acquisition effects normalize, their growth will be somewhere in the 12-18% range, not 28%.
Encora integration risk
Coforge just closed the $2.35B Encora acquisition. On the concall, Saurabh said the Encora amortization will cause a 150 bps gap between standalone EBIT (16.5-17%) and consolidated EBIT (15.5%). That's $40M/year in amortization.
Also, every acquisition brings integration risk. Cigniti took two years to integrate and went from 12% to 19% EBITDA - that's good. But Encora is 3x bigger. The risk is proportionally larger.
Connecting All the Dots
The Large Caps Are Actually More Honest
On the concalls, TCS and Infosys were remarkably straightforward about their problems. Krithi at TCS said directly: FY26 revenue declined 2.4% in constant currency. No sugar-coating. Infosys Jayesh said plainly: we have 75-100 bps headwind from a European manufacturing client because we consciously walked away from a deal that didn't meet our return expectations.
Compare this to Coforge Sudhir Singh. His tone is extremely confident - execution intensity uniquely our own, hunger for growth remains unquenched, we have every intention of coming through on all outlook. He literally asked analysts to remember that we are the team that has delivered unfailingly for nine years. That level of self-confidence is either justified conviction or a warning sign. History says overly promotional CEO language correlates with future disappointments.
Persistent Sandeep Kalra was more measured. He was honest about healthcare delays: unfortunately, those deals have not closed. He was careful about forward guidance. This is healthier behavior.
Most honest CEO across all concalls: HCLTech CVK. He literally quantified AI deflation at 2-3% for their portfolio. He named the two telecom clients that cut spending. He explained the SAP program cancellations. He said I don't want to break out the champagne yet about US telecom recovery. That kind of specificity and restraint is rare.
Least honest / most promotional: Cognizant Ravi Kumar. His entire concall was buzzword soup - reforging first principles, AI Builder stack, tokenized rate cards, Project Leap. When asked direct questions about pricing pressure, he pivoted to marketing language. He launched Cognizant Innovation Network (a corporate VC), announced Astreya acquisition, talked about token metering. The substance-to-buzzword ratio was the worst of any concall I read.
The SAP/Mercedes Story
Mercedes cutting SAP instances by 40% using AI And SAP blocking unauthorized AI agents This connects to what multiple concalls revealed:
Infosys said they walked away from a European manufacturing client deal. Could this be an auto/industrial company cutting IT spend because AI reduced their need for traditional IT services
HCLTech said two SAP programs were discontinued - clients deprioritized SAP modernization partly because there is a general understanding that the timelines for some of this is also going to get extended from SAP.
Accenture talked about how every SAP implementation now needs to be designed from day one to use processes that integrate AI."The old way of doing SAP is dead.
What's happening is SAP implementations used to be guaranteed multi-year, multi-hundred-million-dollar revenue streams for IT services companies. Now, AI is both reducing the need for some implementations (Mercedes cutting instances) AND changing how the remaining ones are done (AI-first from day one). This is a structural change that hits large caps harder because a bigger percentage of their revenue comes from SAP/ERP work.
The midcaps are less exposed because they do less traditional ERP work. Persistent and Coforge are more about custom engineering, AI platforms, and domain-specific solutions. This is actually a genuine advantage.
The Outcome-Based Pricing Claims Are Mostly Aspirational
Almost every company talked about moving to outcome-based or fixed-price models. But when you actually look at the numbers:
Persistent: Fixed price went from some lower percentage to current levels, but Sandeep was very careful not to quantify the exact mix shift
Mphasis: 69% of pipeline is AI-led but that's pipeline, not revenue. Current revenue is still mostly traditional
Cognizant: Ravi Kumar talked about tokenized rate cards with A-0, A-1, A-2, A-3 tiers (human to autonomous). But he admitted this is evolving and some clients are exploring meaning it's not yet happening at scale
TCS: 25% operating margin, highest in 4 years - achieved through cost discipline, not pricing innovation
The gap between the aspirational language about outcome-based pricing and the actual revenue models is one of the biggest blur areas across all the concalls. Everyone wants to move there, but the transition is slow because clients also don't want to change their procurement processes.
IKS Health is the exception. They actually get paid as a percentage of the healthcare provider's revenue. Their NEVA (Net Economic Value Add) model guarantees specific dollar savings to clients. They advanced $16.5M to Palomar Health and already recovered $3M in four months. That's real outcome-based pricing, not aspirational.
The Workforce Numbers Tell a Dark Story
Top 5 large caps cut 58,000 employees while Coforge and Persistent added 10,000. Here's what I noticed across the concalls:
TCS: 584,519 employees. Revenue per employee $50K. They announced salary increments but headcount barely moved. They're hiring 20,000+ freshers but also letting go of junior roles through natural attrition.
Infosys: 328,000 employees. Hiring 20,000 freshers. But utilization including trainees is only 81.1% - meaning they have significant bench.
Wipro: Net headcount additions are happening, but only because of acquisitions (DTS HARMAN). Organic headcount is flat to declining.
Coforge: 35,777 employees. Added 436 in Q4. Utilization at 82.5%. Attrition at 10.8% - one of the lowest.
Persistent: 27,502 employees. Added 791 in Q4. Utilization at 88%. Very lean.
This is the most imp. 2 para of entire Article
Large caps are becoming employment preservation programs to some extent. They can't fire 100,000 people overnight because of political and social pressure (especially TCS/Infosys which are national brands). But they also can't grow because the work that 600,000 people used to do can now be done by fewer people with AI. So they're stuck in a structural bind - too big to grow, too important to shrink.
Midcaps don't have this problem. Persistent with 27,500 people can be nimble. They hire who they need, when they need them. No bloat.
The Niche Players Are Actually Better Than Both Large Caps AND Midcaps
Coforge vs Persistent vs TCS/Infosys. But the concalls reveal that the truly differentiated players are the niche specialists:
Company | Revenue per Employee | Growth | Why Better |
Indegene | $75K | 23.6% INR | Owns the life sciences operating model. Not just tech — runs the actual commercial engine |
IKS Health | ~$60K+ | 24% INR | 1.5% headcount growth for 24% revenue growth. AI-native task elimination |
Intellect Design | N/A (product) | 23% LTM | Banking software product. eMACH.ai platform. Recurring revenue building |
Aurionpro | N/A | 26% 9M | Transaction banking + transit + data centers. Three growth engines |
These companies are not competing in the general IT services market at all. They've created their own categories:
Indegene replaced agencies and CROs in pharma. Their competition is WPP Health, IQVIA - not TCS.
IKS Health replaces in-house hospital staff. Their competition is the hospital's own employees - not any IT company.
Intellect competes with Temenos, Oracle Financial Services - not Infosys.
The future of Indian tech is not midcap IT services vs large-cap IT services. It's domain-specific platform companies vs everyone else. The real alpha is in companies that own IP, have outcome-based pricing, and operate in regulated domains where AI helps them but doesn't replace them.
The Vulnerability Map
Most Vulnerable to AI Deflation:
Traditional IT services doing app maintenance, testing, L1/L2 support - this is 40% of TCS/Infosys/Wipro revenue. HCLTech CVK said this segment will shrink 3-5% CAGR for years.
Most Vulnerable to Client Concentration:
Mphasis (top client is significant, grew 13.7% but one bad quarter could hurt)
KPIT (Honda is massive, program delays already hitting numbers)
LTIMindtree (two large clients, one in BFSI had productivity pass-throughs)
Tata Elxsi (healthcare deal delays caused 13% QoQ decline in one vertical)
Most Vulnerable to Macro/Geopolitics:
KPIT (auto industry spend down 20-25%, Japan delaying programs)
Cyient (Middle East projects pushed, paused large M&A)
HCLTech (two telecom clients cut discretionary spend for rest of CY2026)
Most Vulnerable to Competition from Frontier AI Models:
Anyone doing pure coding/testing services. Anthropic Claude, OpenAI Codex are directly eating into this.
BUT companies doing regulated work (Indegene in pharma, IKS in healthcare, Intellect in banking) are LEAST vulnerable because you can't just deploy an LLM in a regulated workflow without human oversight, domain context, and compliance validation.
The Unusual Behaviors I Noticed
1. Coforge shutting down profitable India business
2. Infosys walking away from deals
3. Cyient's buyback + semiconductor fundraise simultaneously
4. TCS's HyperVault data center play T
5. Wipro creating an AI-Native Business & Platforms Unit Wipro is essentially admitting their existing business model needs a parallel engine. They called it a "dual-engine model" — core services + AI-native platforms. This is a late start compared to Persistent (SASVA platform has been running for years) and Mphasis (NeoIP).
Who Controls the Power?
The power dynamics have shifted fundamentally:
Old World (2010-2020):
Power sat with the CIO who chose the safe vendor (TCS/Infosys)
Large caps controlled pricing because switching costs were high
India's labor cost advantage was the moat
New World (2024+):
Power sits with whoever can deliver AI-embedded outcomes fastest
Clients have alternatives: GCCs, AI tools, specialized vendors
The moat is domain knowledge + proprietary platforms + speed of execution
Who has leverage today:
Accenture has the most leverage globally - $5B acquisition budget, deepest consulting bench, strongest ecosystem partnerships
Persistent/Coforge have leverage in their chosen verticals - BFSI, travel, healthcare - because they've built switching costs through deep integration
Indegene has leverage in pharma because they're embedded in the revenue-generating functions of clients - CMOs and brand heads depend on them
IKS Health has leverage because once their platform is integrated with a hospital's EHR (Epic, NextGen), ripping it out means 18-24 months of disruption
Who has lost leverage:
Pure-play labor arbitrage companies. If your only value proposition is Indian engineers at 50% cost, you're in trouble. AI makes that advantage less relevant because code generation itself is getting cheaper.
1. The bifurcation in Indian IT is real and permanent. Large caps will become dividend-yielding, cash-generating mature businesses. Think of them like utilities - reliable, boring, 5-8% returns. Midcaps that execute well will compound at 15-20% for several more years.
2. The truly interesting companies - Indegene doing three months to three days for pharma creative work. IKS Health achieving 24% revenue growth with 1.5% headcount growth. Intellect Design crossing ₹3,000 crore LTM with AI-first banking software. Aurionpro winning ₹250 crore transit deals with Make in India hardware. These companies have deeper moats than any IT services company because they own the domain, own the IP, and price on outcomes.
3. The biggest risk is "what if AI deflation accelerates faster than new AI revenue grows?" HCLTech put it at 2-3% deflation currently. What if it goes to 5-7% as models improve? That math breaks even for midcaps growing 15-17%. Only companies growing 20%+ with proprietary platforms survive that scenario comfortably. That narrows the list to: Indegene, IKS Health, Persistent (maybe), and the product companies like Intellect.
4. The SAP/Mercedes story is the canary in the coal mine. When the world's largest enterprise software company's biggest customer cuts instances by 40% using AI, it tells you that the entire enterprise software + services ecosystem will contract in absolute terms before it expands again in new forms. The companies that survive are the ones building the new forms, not maintaining the old ones.
Is Indian IT Dead or Transforming?
Neither. It's splitting into three completely different businesses that happen to share the same label IT services.
Think of Indian IT companies like this. For 25 years, they ran one business: Give us your boring tech work, we'll do it cheaper with smart Indian engineers. That was the whole game. Hire people, bill per hour, repeat.
Now THREE things are happening at the same time:
Thing 1: The old business is shrinking. AI tools like Claude, Codex, Copilot can now do what junior engineers used to do - write basic code, run tests, do documentation. HCLTech CEO put a number on it: 3-5% deflation per year in this old work. That's not a guess, that's what they're seeing in real deals.
Thing 2: New AI-related work is opening up. Legacy modernization (making old systems work with AI), data preparation (cleaning up messy data so AI can use it), agentic AI deployment (building AI agents for specific business tasks). This is growing fast but starting from a small base.
Thing 3: Completely new businesses are emerging. TCS building data centers. Accenture buying product companies. Cognizant acquiring healthcare software. These aren't IT services at all - they're infrastructure, products, and platforms.
The question is: Does Thing 2 + Thing 3 grow faster than Thing 1 shrinks?
Right now, for most large caps, the answer is barely. TCS: -2.4%. Infosys: +3.1%. The growth side is not yet big enough to overcome the shrinking side.
For midcaps like Persistent (+17.4%) and Coforge (+29.2%), the answer is yes, but partly because they were smaller and more focused, so they didn't have as much of the old shrinking business to begin with.
Will OpenAI and AI Labs Eat Indian IT's Lunch?
OpenAI's $10B Services as a Service venture.
Yes, in some areas. No, in most areas. And here's why.
The OpenAI $10B venture (with Blackstone, Goldman Sachs) is about going to big companies and saying: We'll install AI for you, customize it, train your people, and run it. This directly competes with what Accenture, TCS, Infosys sell.
BUT and Ameya makes this exact point in the podcast if yu watch it that the actual hard work of making AI work inside a large bank, hospital, or manufacturer is not about the AI model itself. It's about understanding that specific bank's 47 different legacy systems, their regulatory requirements, their internal politics, their data mess. OpenAI doesn't have 27 years of working inside JP Morgan's systems. TCS does.
Let me put it this way:
What OpenAI can do well: Sell the AI model, provide the base technology, do flashy demos, work with tech-savvy companies that have clean systems.
What OpenAI cannot do (yet): Navigate a European bank GDPR compliance across 15 countries. Understand why a specific mainframe system at a US insurer processes claims in a weird way that nobody documented. Handle the 80% of enterprise work that Ameya calls plumbing - the boring, messy, regulatory, relationship-driven stuff.
The real threat is different from what people think. It's not that OpenAI replaces TCS tomorrow. It's that OpenAI makes it possible for SMALLER, FASTER companies (like Persistent, Coforge, or even startups) to punch above their weight. Because if the AI model is commoditized and available to everyone for $20/month, then the only differentiator is: who understands the client's business better and who can deploy faster?
That's exactly why midcaps are winning. Not because they have better AI models - everyone uses the same Claude, GPT, Gemini. But because they're faster, more focused, and willing to take outcome-based risk.
The Dividend Question Why Are They Paying Out Cash Instead of Investing?
why the fuck are they giving dividends? Look at what each company is doing:
Company | Dividend/Buyback | What It Signals |
TCS | ₹110/share dividend (FY26) | "We generate so much cash we don't know what to do with it" |
Infosys | ₹48/share dividend | Same story |
Wipro | ₹15,000 crore buyback (LARGEST ever) | "Our stock is cheap, we'd rather buy it back than invest" |
Coforge | ₹2.25/share dividend (small) | Investing most cash in Encora acquisition |
Persistent | ₹40/share dividend | Moderate payout, reinvesting in growth |
Cyient | ₹720 crore buyback | Promoters NOT participating — signaling undervaluation |
Large caps are returning cash because they genuinely don't see enough high-return investment opportunities in their core business.
Compare this to what Accenture is doing: $5 billion in acquisitions this year. They bought Faculty (AI company), DLB Associates (data center engineering), CyberCX (cybersecurity), Ookla ($231M revenue subscription business with only 430 employees). Accenture is INVESTING that cash into new business models. TCS is RETURNING it.
The midcaps are the opposite. Coforge spent $2.35 billion on Encora. Persistent is investing in SASVA platform and AI capabilities. Mphasis acquired Theory and Practice for decision intelligence. They're using their cash to build new capabilities.
This is the single biggest divergence signal. When the large company returns cash and the small company invests cash, history shows the small company usually wins over the next 5-10 years. Same pattern played out with HCLTech 15 years ago, they invested when others returned cash, and became the #3 IT company.
TCS Going Into Data Centers: What Does This Mean?
TCS has never been in the infrastructure business. Building and operating data centers requires completely different skills - real estate, power engineering, cooling systems, physical security. This is a bet that will take 3-5 years to prove out. And Accenture mentioned the same theme - they acquired DLB Associates (data center engineering) and announced partnerships with AMD for AI infrastructure.
TCS is thinking Perhaps: "If clients need less of our old services, but they need MORE compute infrastructure to run AI, let me build the infrastructure." It's a pivot from selling the service to selling the pipe.
Where Is the Edge? Why Do I Think This Way?
The edge is NOT in having AI. Everyone has AI. The edge is in three things:
Domain depth - Understanding the client's specific business deeply enough to build AI that actually works in production. Not demos, production. Ameya story about three agents disagreeing and the system freezing - that's what happens when you deploy AI without domain context.
Speed of decision-making - Coforge deploying teams in week 1 vs TCS in week 8 is real. I saw it in the concalls: Coforge Sudhir Singh makes decisions fast, TCS governance processes are multi-layered.
Willingness to take outcome risk - When IKS Health advances $16.5M to a hospital guaranteeing they'll generate that value, when Indegene does fixed-outcome commercial models, when Persistent prices on outcomes not hours - these companies are taking real risk. That risk-taking ability comes from having domain confidence. You can't price on outcomes if you don't know what outcomes are achievable.
Indian IT is not dying. But Indian IT as one category no longer exists. It's splitting into dividend machines (large caps), growth engines (midcaps), domain champions (specialists), infrastructure builders (TCS data centers), and startup wildcards. The investor who still thinks of this as one sector is going to underperform. The investor who can tell the layers apart will find real opportunities.AI deflation could accelerate faster than anyone models. If Anthropic's next model makes coding 80% autonomous (not 40% as today), the compression hits even the midcaps hard. Persistent's 17.4% growth might become 8%. Coforge's organic 12% might become 5%. The only companies truly safe are the domain specialists (Layer 5) because their value is not in coding - it's in understanding pharma, banking, healthcare operations. Code is just the delivery mechanism.
The safest long-term bets are companies where AI HELPS their business but doesn't define it. Indegene, IKS Health, Intellect Design. Their business is eliminating tasks, managing operations, running platforms. AI makes them better at this. It doesn't threaten them. I can be wrong the way Model is Updating itself, how much it effect 1 year down the line will the showcase.
RECAP
Coforge Sudhir Singh said on his concall - 19 of the top 20 wins came against TCS, Infosys, Wipro, Accenture in contested pitches. Five years ago, this would have been impossible. Today it's normal.
When Persistent prices a deal on outcomes (fixed price for fixed result), they're betting they can deliver with AI in 11 months what TCS would do with people in 24 months. If they're right, they win. If wrong, they lose money. But mostly they're right.
Coforge derives 35%+ revenue from travel, transport, hospitality. They got the Sabre $1.56B deal because they understand travel tech better than TCS. Persistent dominates BFSI and healthcare. Specialization beats generalization.
Persistent growth has been DECELERATING - 23% → 21% → 18.8% → 17.4% over four years. So while they're winning a lot, the rate of new wins is slowing. As AI compression accelerates, even winners might find their gross wins can't keep up.
KPIT and Tata Elxsi: The OEM Insourcing-and-Coming-Back Story
Background (Pre-2019):
Auto companies like Volkswagen, Ford, GM, Toyota mostly built cars (hardware). For software, they used outside specialists - KPIT, Tata Elxsi, Bosch, Continental, etc. Indian companies like KPIT became experts in AUTOSAR (a standard for car software), embedded systems, infotainment, ADAS (driver-assistance).
Then Tesla happened. Tesla showed that cars are now computers on wheels. Software became 60-70% of a modern car value. The CEO of every old auto company suddenly thought: If software is now the core differentiator, why are we outsourcing it? We need to OWN the software.
The In-House Push (2019-2023):
Every major automaker started building huge internal software teams:
Volkswagen created CARIAD in 2020. They hired 6,000 software engineers in months. Spent €14 billion. The plan: one common software platform for VW, Audi, Porsche, Skoda - all brands.
General Motors built Ultifi platform. Ford created Ford+ software group. Toyota created Woven Planet (later Woven by Toyota). Mercedes built MB.OS. Hyundai invested in 42dot. Every OEM did this. The promise was: We'll be the next Tesla. We'll control our destiny. We'll own customer data.
The Disaster (2023-2024):
It failed spectacularly. Let me give you real numbers from the search results above:
CARIAD actual results:
2022: Lost €2.1 billion on €800 million revenue
2023: Lost €2.4 billion
2024: Lost €2.4 billion
Total losses: Over €7.5 billion across three years
Software delays caused Porsche Macan EV and Audi Q6 e-Tron launches to be DELAYED by a full year
What went wrong? An ex-CARIAD insider explained: We hired anyone who could carry a laptop. Sometimes hiring took just 24 hours. Many hires had zero automotive experience. Students became project managers.
The software was so buggy that early VW ID.3 cars literally had to sit in parking lots in Germany for months waiting for software fixes before being delivered to customers.
VW had to take a $5.8 billion stake in Rivian (the EV startup) just to get usable software. Think about that - Volkswagen, the world's #2 carmaker, gave up trying to build software in-house and bought minority stake in a startup just to use Rivian's software.
Similar stories:
GM closed down Cruise (their robotaxi unit) entirely after $10+ billion in losses
Ford cancelled the F-150 Lightning fully autonomous program
Toyota's Woven Planet had multiple resets and leadership changes
Stellantis has been cutting software jobs
Why did they fail? - domain knowledge takes 20-30 years to build. KPIT has 28 years of automotive software experience. CARIAD tried to replicate this in 3 years with 6,000 random hires. Impossible.
Also and this is important - building software inside an OEM means fighting with the brands. CARIAD insider said: Audi wanted this, Porsche wanted that. We developed the same feature six times because each brand wanted a different version.
The Return to Partners (2024-2026):
Now OEMs are quietly coming back. Not announcing it loudly because it's embarrassing. But the work is flowing back.
KPIT has been the biggest beneficiary. They:
Acquired Caresoft Group engineering business for $157 million in 2025
Won big AUTOSAR Adaptive deals
Their KPIT GenAI Platform is being adopted across multiple OEMs
Their KGPT compresses development timelines significantly
They are positioned as the integrator of choice for "software-defined vehicle" projects
Tata Elxsi has similar tailwinds in their automotive vertical, though they're more diversified (also in healthcare, broadcasting).
This is what structural tailwind means. It's not a one-quarter or one-year boost. It's a multi-year shift where:
OEMs gave up trying to build software in-house (proven failure)
They face cost pressure (need to cut their own software teams)
They need partners who already have the expertise
The partners (KPIT, Tata Elxsi, Bosch, Capgemini) get the work back
BUT and this is the catch - KPIT's recent concall was actually weak. The auto industry is going through cyclical pain (tariff threats, EV slowdown, China competition). Honda delayed programs. European OEMs cut spending. Japanese OEMs pushed timelines. So even though the structural tailwind is there, near-term auto industry weakness is hurting KPIT's reported numbers.
Legacy modernization is a $9-13 billion market growing at 9-13% CAGR globally, and it's one of the most active battlegrounds in IT services. Let me explain.
What is "legacy modernization"?
Imagine a 1985 banking software written in COBOL (an old programming language from 1960s). Banks like Bank of America, Citibank, JP Morgan, plus governments, insurance companies - they all have these ancient systems still running their core operations. Why? Because they work. They process millions of transactions a day reliably. Replacing them is risky.
But these systems have problems:
Programmers who know COBOL are retiring (average COBOL programmer age is 60+)
Can't easily integrate with cloud, mobile apps, AI
Maintenance is expensive and slow
Adding new features takes 18-24 months
So legacy modernization is the process of taking these old systems and either:
Lifting them to cloud (rehosting)
Rewriting in modern languages like Java (refactoring)
Replacing with new systems (rearchitecting)
The cost numbers KPIT mentioned ($10/line to $1.50/line):
Old way: Manual modernization. Programmer reads each line of COBOL, understands it, rewrites in Java, tests it. Costs about $10 per line of code. A typical bank has 50 million lines of COBOL. So full modernization = $500 million project, takes 3-5 years.
New way: AI tools. Tools like AWS Transform, IBM watsonx Code Assistant, Capgemini CAALM platform read the old code, automatically convert it, document it, test it. Cost drops to $1.50 per line. Same 50M-line bank now costs $75M and takes 6-12 months.
Who is doing this beyond KPIT?
Pretty much every major IT services company has a modernization practice:
Company | Their Approach |
IBM | watsonx Code Assistant, mainframe-to-cloud bridge |
AWS | AWS Transform — agentic AI for code refactoring |
Capgemini | CAALM platform — generative AI for legacy migration |
TCS | "Modernization factory" approach — proprietary frameworks |
HCLTech | Mainframe specialty + AI-driven approach |
Cognizant | Pega GenAI Blueprint partnership for legacy |
Kyndryl | Partnered with Google Cloud on Mainframe Modernization Accelerator (March 2025) |
Accenture | Has been the biggest beneficiary — their consulting muscle wins these deals |
Persistent | SASVA platform for AI-driven modernization |
The market consolidation: Top 5 players (IBM, AWS, TCS, Capgemini, HCLTech) hold 70-75% of the mainframe modernization market.
Why is this market exploding NOW?
Three reasons converging:
Reason 1: AI makes it economically viable. A modernization project that used to cost $500M now costs $75M. Suddenly clients who couldn't afford it before can do it. Demand explodes.
Reason 2: COBOL programmers are retiring. There's a literal talent crisis. Banks are scared they'll have NO ONE who can fix their ancient systems in 5 years.
Reason 3: AI-readiness requires modern systems. You can't put AI on top of 1985 COBOL software. Banks want AI features → they need modern systems → they need to modernize legacy.
This was supposed to be a HUGE multi-year revenue boom for IT services. And it is but the cost compression (from $10 to $1.50/line) means revenue per project dropped 85%. So companies are winning more projects but each is much smaller. Volume up, but value per project way down.
What this means for IT companies:
Negative for traditional services: One $500M modernization project becomes a $75M project. Less revenue.
Positive for AI-native deliverers: Persistent, Coforge, who can deliver these projects with smaller teams and AI tools, can win more projects faster.
Best for product/platform owners: IBM (with watsonx), AWS (with Transform), Capgemini (with CAALM) own the tools. They earn license/platform fees ON TOP of services.
Midcaps are winning the new economy. They built AI-native platforms. They can deliver these compressed-cost projects profitably. Large caps with their pyramid models can't profit at $1.50/line because their cost base is too high.
The Core Point is - Midcaps are winning, but it's not because all midcaps are equal. It's specifically because they have:
Lean structures (no pyramid bloat)
AI-native delivery (built for the new world)
Vertical depth (real domain knowledge)
Speed of decision-making (week 1 deployment vs week 8 at large caps)
The OEM in-house failure (CARIAD, Cruise, Ford+) tells you something important about ALL enterprise AI today: clients underestimate how hard it is to build domain-specific software. They try, fail, and come back to partners. This will happen in banking (BFSI clients trying to build their own AI tools and failing), in healthcare, in retail. KPIT tailwind today will be replicated in other sectors over the next 3-5 years.
Legacy modernization is a massive market but with compressed economics. Volume up, value per project down. This favors AI-native deliverers (midcaps) and platform owners (IBM, AWS, Capgemini), not traditional pyramid-based services (TCS, Infosys, Wipro).
Coforge organic vs reported gap - Read concall transcripts. Companies will use acquisitions to mask organic slowdowns. The next 2-3 years will show which midcaps have real organic momentum vs which were just acquisition-driven illusions.
Connecting to the OpenAI Services-as-a-Service piece: OpenAI's $10B venture is real, but it threatens labor-arbitrage IT services more than domain-specialized partners. OpenAI will not learn 28 years of automotive software in 2 years. But they will eat the bottom of the pyramid - basic implementations, simple integrations, generic AI deployments.
Conclusion - The advantage has shifted from scale + cheap labor to "domain depth + AI-native delivery + speed. Companies that have all three will compound. Others will get compressed.The Pattern: Indian IT M&A Has Two Modes - Domain Fit (Works) or Buying Capability (Usually Fails)
LTTS-SWC failure (₹800 cr → ₹452 cr in 3 years)
Wipro-Capco struggle ($1.45B, cultural mismatch)
Cyient Citec/Celfinet issues
Coforge-Cigniti success
Mphasis-Silverline progress
HCL-IBM products acquisition success
LTIMindtree merger
Tech Mahindra acquisitions (Pininfarina, Comviva)
The Biggest Insight is Why M&A Failure Predicts AI Vulnerability
Companies that fail at M&A also tend to be the ones most vulnerable to AI deflation. Why?
Because both M&A and surviving AI require the SAME organizational capability: the ability to evolve and integrate new things into the core business.
If Wipro can't integrate Capco (a relatively simple transaction) successfully in 5 years, how will they integrate AI-native delivery models into their 250,000-person organization? They probably can't. Same organizational rigidity.
If TCS avoids M&A because they know they can't execute, they're also less likely to make bold AI bets. That's why TCS is doing the data center pivot (HyperVault), it's a single, focused infrastructure bet, not a complex business model transformation.
If Coforge can integrate Cigniti from 11% to 19% EBITDA in 6 quarters with cross-sell working, they probably can also integrate AI-native delivery faster than TCS. The same organizational muscle is at work.
This is why Persistent, Coforge, and the domain specialists will win the AI transition. They have execution capability that translates from M&A integration to AI integration. The companies that fail at M&A (Wipro, TM, LTTS) will also struggle with AI integration. They lack the organizational reflexes.
Indian IT M&A has a 30-40% success rate. Most acquisitions destroy value or break even.
The successes share a pattern: same customer base, adjacent capability, cultural fit, disciplined integration.
The failures share a pattern: trying to buy capability you don't have, cultural mismatch, parent dumping assets, following hype cycles.
LTTS-SWC is the most expensive recent failure. ₹348 crore destroyed in 3 years. Now they're pivoting to Engineering Intelligence - another label, same underlying issue: low organic growth needs window dressing.
Coforge-Cigniti is the most successful recent acquisition in Indian IT. Almost everything went right. But Encora is 12x bigger ($2.35B vs $190M). The next 18 months will determine if Coforge's execution capability scales.
Watch Wipro-HARMAN DTS in 2026-27 to see if Wipro learned anything from Capco. Early signs suggest no.
The companies that have refused big M&A (TCS, Persistent, the domain specialists) tend to outperform over long periods because they don't destroy capital on bad deals.
M&A track record predicts AI transition success. Companies that integrate acquisitions well will integrate AI well. Companies that fail at M&A will fail at AI transition. Same underlying capability.
L&T was the wise seller of SWC. The buyer (LTTS) and its minority shareholders got hurt. This pattern - parent companies dumping problem assets on listed subsidiaries - happens often in India. Always check if the listed company's growth acquisition came from its own parent. If yes, be very skeptical.
Conclusion
So model access is not the moat.
The moat is:
Who understands the workflow deeply enough to redesign it using AI?
The power is moving away from:
I have people.
towards:
I understand the process, I own the workflow, I can guarantee the outcome.
Group | Companies | What is happening | My read |
Scale consolidators | TCS, Infosys, HCLTech, Accenture, Cognizant | Winning large deals, vendor consolidation, AI infra, cloud, data, modernization | Safer, deeper client access, but lower growth because base is huge |
Midcap compounders | Coforge, Persistent, Mphasis, LTIMindtree, LTTS, KPIT, Tata Elxsi | Faster growth through vertical focus, AI-led deals, customer mining | Higher growth, but more execution risk |
Vertical/platform specialists | Indegene, RateGain, Intellect, Aurionpro, IKS Health | Not generic IT. They own specific industry workflows | Highest moat potential if productization works |
Turnaround/recovery plays | Wipro, Tech Mahindra, Cyient | Big changes, margin reset, new deals, portfolio cleanup | Can improve sharply, but proof still needed |
What Ameya is saying:
FY26 numbers that arrived recently. Same year, same global macro, same client uncertainty, but two completely different outcomes:
TCS shrank 2.4% in constant currency
Wipro shrank 0.3%
Infosys grew 3.1%
HCLTech grew 3.9%
Persistent grew 17.4%
Coforge grew 29.2% (about 12% organic, rest from Cigniti)
When the same environment produces such different outcomes, it's not luck. It's structural. Something has fundamentally changed in how this industry works.
What Has Actually Changed
Shift 1: Decision-makers have changed.
Shift 2: Big deals are now getable for midcaps.
Shift 3: AI changes the headcount math.
Shift 4: AI is being deployed differently at different scales.
Shift 5: The talent pool is flowing to midcaps, not large caps.
The growth frontier has moved. It sits with the vertical specialists who combine domain depth, AI capability, and organizational speed."
Stop thinking about Indian IT as one sector. Evaluate each company on three questions - what business are they in (legacy or new), do they have organizational speed, and is their value protected by domain depth that AI cannot easily replicate? The companies that score well on all three will compound. Those that don't will become bond proxies.
After reading this whole article, you'll understand AI deflation is here.
When you look at the whole domain with a macro perspective, the way of doing business has totally changed in just 4 years. Now it depends on AI model capabilities. Clients have options now. Low-end margin jobs are not worth it, and a lot of them have already been replaced by AI. Integration with AI in projects is mandatory.
Midcaps are winning and competing directly with large caps and winning the deals. This is like AI is directly benefiting midcaps, and probably small caps too. Corporate culture is now most important. Risk is being given onto the company by the client now, they have to be more productive, and outcome-level results are now the priority.
Currency depreciation can help, but legacy business does not hold much power. Either do like Accenture and TCS - what they are doing - and go after new domains like semiconductors, data centers, cybersecurity etc., or be a specialist with years of domain knowledge, like an expert of that field. Like you read above how the OEMs tried building software themselves in Europe and failed miserably, and now they are moving on to Indian players, and now the deals are coming back. Niche players will survive better for now. Not sure of the future, but the niche ones hold complex, complicated power and are harder to remove from integration.
The new modernization theme is here, but going to help much of Indian IT - the cost has gone down a lot of everything, okay. Each year companies are depreciating. Even if you look at Persistent, what if AI becomes much better - what happens then?
You see, as I told in the past Articles in Dec i think, if the European auto industry is in control and Chinese players sell more cars there, those Indian players who have exposure there will be hit in the long term, because it affects Europe's own car sales numbers, and over time they will be squeezed and go into Chinese hands.
Well, that's showing off already in Europe but somewhere else - the OEMs and auto giants are cutting cost due to Chinese pressure, sales are consolidating, and it is affecting Indian IT giants too. So those ancillary players of India who have good exposure in Europe will also be affected in the future perhaps. But at least OEMs are moving back to outsourcing and cutting cost & failed in house projects. & if yu conclude everything - That's why the rerating is happening. And margin plus deflating conditions are making things worse.
When you read this article, you will understand why large caps are not winning against midcaps. Even when you look at AI revenue, it is still in single digits, so you can understand the state of this sector.
With my view and understanding, there are only 2 ways from here - go after new domains, find new business, new ways, whatever is needed, go after new themes; or be a specialist niche domain expert. That's how one can survive, and that's why Indian IT will still be safe. Old business is not worth much and is low margin, so it does not hold much, and is deflating fast.
One should focus on niche experts and midcaps for now only.
Earnings is the most important factor to judge anything, especially the concall.
I am not having any high expectations from this domain, but when you read this article, you will find opportunity for sure.
Why I am taking Interest? Well, if this sector deflate Indian GDP the whole economy will face the hit...
#TCS #Infosys #Wipro #HCLTech #Persistent #Coforge #Mphasis #IntellectDesign #KPIT #Aurionpro #Indegene #IKSHealth #TataElxsi #LTIMindtree #TechMahindra #Cyient #LTTS #RateGain #Hexaware #Birlasoft #Zensar #Mastek #Firstsource
Great Writeup.