- Share.Market
- 11 min read
- 20 Aug 2026
Every quarter, corporate slide decks bombard us with staggering numbers: hundreds of thousands of engineers trained in GenAI, record-breaking deal wins, and lofty promises of AI-driven transformation.
Yet, if you check your investment portfolio, the story looks very different. Seven out of the nine top IT services giants across India and the US saw their share prices drop over the past six months.
What is actually going on behind closed doors?
To separate executive PR from cold, hard math, we unpacked the audited financial disclosures, investor calls, and SEC/SEBI filings across nine major IT companies—Tata Consultancy Services Ltd., Infosys Ltd., Wipro Ltd., HCL Technologies Ltd., Accenture, Cognizant, IBM, Coforge Ltd., and Persistent Systems Ltd.
The AI Paradox (Or, Where Did All the Productivity Go?)
If you listen to management commentary, every IT company has transformed into an AI powerhouse overnight:
- TCS built a force of 270,000 employees with advanced AI proficiency by FY26 year-end (roughly 3x YoY), reaching 312,000+ by June 2026.
- Wipro certified 93,994 AI Masters and claims 240,000+ “AI-ready” workers.
- Infosys claims 84% of its workforce is “AI-aware” and bumped average annual training hours from 71 to 113.
- HCLTech trained 144,223 people in GenAI (63% of staff) and certified 11,800 AI Builders.
The Twist: Training vs. Reality
Here is the catch: Not a single company has shown AI-driven savings in its cost line.
While training numbers look impressive on paper, operational reality tells a different story:
- Wipro’s disclosures reveal that out of ~242,000 employees, only ~60,000 (roughly 25%) actively use AI tools in client delivery.
- Coforge reported spectacular engagement-level productivity gains of 30%–80%, yet revenue per billable employee barely moved, up just ~4% to $70,557.
- TCS saw employee costs hit 58.3% of revenue in Q1 FY27, a marginal annual high.
- Persistent saw revenue-per-employee growth slow down dramatically from +18% to +11%.
The AI Gap: What Companies Claim vs. What the Metrics Show
| Company | AI Capability Built | Actual Operational Outcome |
| TCS | 270,000 employees with advanced AI proficiency at FY26 year-end, roughly 3x the prior year; 312,000+ by June 2026 | Employee cost 58.3% of revenue in Q1 FY27, a marginal annual high, though the same quarter a year earlier was higher at 59.5% |
| Wipro | 93,994 AI Master-certified; 240,000+ “AI-ready” | Over 60,000 of ~242,000 employees actively use AI tools in delivery, roughly a quarter (ratio is our arithmetic; Wipro discloses the two figures, not the ratio) |
| Infosys | 84% of workforce “AI-aware”; training hours up from 71 to 113; 80,000+ using AI coding tools | Revenue per person grew 3% in both FY25 and FY26. No AI-specific productivity measure is disclosed |
| HCLTech | 144,223 GenAI-trained (63% of staff); 11,800 certified AI Builders | Revenue per employee up 4.2% to ~$64,548 |
| Coforge | Engagement-level productivity gains of 30–80% reported | Revenue per billable employee up ~4% to $70,557 |
| Persistent | 20–25% software-development lifecycle gains; 30–40% task-level | Revenue per employee growth slowing, +18% then +11% |
| Cognizant | 10,000+ certified AI architects | Cost of revenue sticky at 65–66% and rising |
The Productivity Giveaway
Why aren’t these impressive productivity gains boosting profits?
Because IT firms are being forced to hand those savings straight back to their clients just to win or keep contracts:
The Deflationary Trap: HCLTech explicitly quantifies this phenomenon: a contract previously worth $100 million now books at roughly $80 million (a ~20% price drop)—while requiring 25%–30% more effort per dollar of contract value booked!
- TCS confirms that project-level productivity gains of 10%–15% are passed directly to clients at contract renewal.
- Infosys explicitly factored productivity pass-through into its margin guidance, admitting that productivity “will get passed back to the client largely.”
- Wipro revealed that winning large AI deals now requires “forward productivity” commitments, giving up profit margins upfront at the point of winning the bid.
In short: AI is currently acting as a deflationary hammer. It helps win contracts by making them cheaper for clients, but it shrinks the total billable pie for IT providers.
The Infosys Clue & The Subcontracting Secret
The Infosys Math Puzzle
Infosys is the only company in the entire sector to disclose AI’s share of total revenue:
- 5.5% of revenue (disclosed at its Q3 FY26 AI investor day).
- 8.2% of revenue (published with Q1 FY27 results).
Management noted that AI revenue is growing at double-digit sequential rates. However, over that exact same window, Infosys’s total company growth was a modest 2.4% year-on-year in constant currency.
Doing the Math: If AI revenue is growing at double digits while overall growth sits at just 2.4%, the remaining ~92% of legacy business must be flat or shrinking. AI isn’t creating massive new revenue; it is simply replacing legacy IT work that is decaying.
The Margin Myth & The Subcontracting Spike
For a decade, the industry promised that new technology (Cloud, Automation, AI) would expand profit margins. The ten-year record from TCS proves this premise wrong:
TCS Ten-Year Cost Breakdown (FY2017 vs FY2026)
| Metric | FY2017 | FY2026 |
| People Costs (% of revenue) | 52.2% | 58.1% |
| Non-People Costs (% of revenue) | 22.1% | 16.9% |
| Operating Margin Band | Locked at 24.0% – 25.9% | Locked at 24.0% – 25.9% |
Ten years of aggressively slashing non-employee overhead was entirely consumed by rising talent costs. Profit margins didn’t budge by a single percentage point.
Furthermore, if AI were truly automating tasks and freeing up internal capacity, companies would hire fewer third-party contractors. Instead, outside-help expenses are surging everywhere:
- Infosys: Outside-help costs jumped from ~7.9% of revenue in FY25 to ~12.5% in FY26. Subcontractor expenses rose ~73% while revenue grew just ~9.6%.
- HCLTech: Outside contractor costs rose to 14.2% of revenue (reaching 14.9% in Q4 FY26), driving almost all of the company’s margin compression that year.
- Persistent: Professional fee expenses rose +25.7% in FY26 (after jumping +52.2% in FY25).
Order Books vs. Reality & The Acquisition Cushion
The “Record Order Book” Disconnect
We often see press releases celebrating “Record Total Contract Values (TCV)”. But order books are no longer translating into near-term revenue:
- Infosys signed ~$50 billion in contract value over a three-and-a-half-year period (FY24 to Q1 FY27), yet absolute revenue grew by only ~$1.6 billion in that same timeframe.
- Wipro’s order book rose to $16.4 billion (with large deals up 45% YoY), yet actual revenue fell 1.6%.
Why the gap?
Multi-Year Timelines: A $50M deal signed today is delivered over 3 to 5 years.
Ramp-Up Delays: It takes 12 to 24 months for signed work to turn into actual billings.
Consolidation Discounting: Winning a larger share of a client’s IT budget usually requires offering a rate cut across their existing business.
How Acquisitions Mask Organic Performance
When headline growth numbers are reported, M&A often does the heavy lifting:
- IBM (Software): Headline constant-currency growth of +5% masks flat organic growth (filled by acquiring HashiCorp, Confluent, and DataStax).
- Coforge: Headline growth of 29%–33% shrinks to 1.1% QoQ organic growth once Encora and Cigniti are stripped out.
- Cognizant: Headline growth of 3.9%–6.4% includes 1%–2% from acquisitions (Belcan, 3Cloud, Astreya).
- Infosys: Disclosed that ~70 bps of its FY26 growth came from acquisitions (Optimum Healthcare, Stratus).
Only Persistent Systems delivered growth (~17%–18%) that was substantively organic over FY25–FY26 (though it is now taking on debt to buy Nagarro).
A Sector Fractured: The Three Tiers
It is no longer useful to lump all IT stocks together. The market has split into three groups with fundamentally different financial dynamics:
The Three Tiers at a Glance
| Group | Key Characteristics | Core Financial Tension |
| India Large Caps (TCS, Infy, Wipro, HCL) | • Low single-digit growth (-1.6% to +3.9%)• Debt-free, defending margins | Paying out 82%–99% of cash while needing billions for AI data centers. |
| US Large Caps (Accenture, CTS, IBM) | • Low-to-mid single-digit growth (3%–6%)• Slowly expanding margins | Narrow growth theses; dependent on specific verticals or software mix. |
| India Mid Caps (Coforge, Persistent) | • Mid-teens to ~30% growth• Aggressive M&A, newly leveraged | Betting the company’s future on debt-funded mega-acquisitions. |
India Large Caps: The Cash Dividend Squeeze
India’s IT giants have built a reputation as cash-return machines, paying out almost all profits to shareholders:
- TCS paid out 99.2% of its earnings in FY26.
- HCLTech paid out 97.6% of its earnings.
- Infosys returned 82.1% of two-year Free Cash Flow via dividends and an ₹18,000 crore buyback.
The Tension: AI now demands massive physical capital expenditure. TCS is backing an ₹18,000 crore data-center project (its 51% share is ~₹9,180 crore). HCLTech is committing ₹3,500 crore to an AI data center (~20% of its annual free cash flow).
Our Reading: You cannot give 99% of your cash flow to shareholders and self-fund expensive, capital-intensive AI infrastructure without taking on debt or lowering payouts.
US Large Caps: Narrower Growth Motors
US-listed firms file deeper quarterly reports, revealing narrow growth engines:
- Cognizant: Driven entirely by Financial Services (+9%, +10%, and +12% growth over 3 quarters), while its other three business units averaged just ~1% growth.
- IBM: Driven by software mix shift. Software now makes up ~44% of revenue ($30.0bn). Software earns ~33% margins versus Consulting’s 11%–12%.
- Accenture: Takes the most optimistic stance in the industry, arguing AI expands its total market opportunity. It claims $80.6bn in FY25 bookings, a $34.6bn Managed Services floor, and 77,000+ AI practitioners. However, its gross margins slipped from 32.6% to 31.9%.
India Mid Caps: Debt-Fueled Transformation
Both Coforge and Persistent have shifted from net-cash companies to debt-financed acquirers:
- Coforge bought Encora (~$600M revenue) funded by a $550M term loan. Repayments spike to $209M in FY28 and $258M in FY29—which will consume roughly 80%–90% of its annual earnings.
- Persistent is acquiring Nagarro (~$1.14bn revenue) via a EUR 1.4bn facility. This expands Persistent to 1.7x its current revenue size ($2.9bn combined), but adds a lower-margin business (13.9% EBITDA vs Persistent’s 18.4%) that it cannot fully integrate for two years.
The Company-by-Company Tear Sheet
TCS (Tata Consultancy Services)
- The Moat: 670+ clients have been with TCS for over ten years. 66 clients spend >$100M annually (up from 60 in FY23).
- The Ceiling: A massive workforce of 593,798 staff makes top-line acceleration difficult.
- AI Signal: Discloses $2.6bn in annualized AI revenue (industry leader in absolute terms). However, quarterly additions are slowing: +$125M in Q4 FY26 down to +$75M in Q1 FY27.
Infosys
- Cost Execution: Its “Project Maximus” offset a 140 bps wage hike and 40 bps M&A hit in a single quarter through subcontractor and operational cuts.
- Growth Outlook: FY27 revenue growth guidance of 1.5%–3.0% sits at the low end of large caps.
- Leadership Transition: Long-time CEO Salil Parekh steps down on 31 March 2027. Company insider Ashiss Kumar Dash takes over in April 2027 (mentorship starts Oct 2026); Nandan Nilekani remains Chairman.
Wipro
- Conversion Gap: Total contract value rose to $16.4bn (+45% in large deals), yet revenue fell -1.6%.
- Client Erosion: Accounts spending >$100M slipped from 22 (FY24) to 17 (FY25) to 16 (FY26).
- Geographic Drag: Americas accounts for 60.6% of IT revenue; Americas 2 fell -7.3% YoY due to US healthcare budget cuts. Wipro is the only Indian large cap that provides no absolute AI revenue metric.
HCLTech
- Transparent Metrics: Discloses $684M in annualized AI revenue (+62.1% YoY) while explicitly quantifying deal deflation (~20% price drop per deal).
- Software Asset: HCLSoftware generated $1.063bn in ARR ($1,395M total revenue) at a 26.5% segment margin (vs. Services at 16.2%).
- Risk: Operating margin sits at 17.5%, the bottom of its 17.5%–18.5% guidance range.
Accenture
- The AI Bull Case: Frames AI as a tailwind expanding its addressable market into mid-market clients ($240bn TAM opportunity).
- Scale: $80.6bn annual bookings (FY25), $34.6bn managed services backlog.
- Flaw: Gross margins fell 70 bps to 31.9% despite 7% top-line growth.
Cognizant
- Single Engine: Growth rests on Financial Services (+9% to +12%), while other verticals sit near ~1% growth.
- Strategy: Reinvesting $200M–$300M in savings directly into AI rather than banking it as margin expansion.
IBM
- Software Driven: Software makes up ~44% of revenue ($30.0bn) with ~33% margins.
- The Trap: Software CC growth of +5% was flat organically. Consulting growth has been stuck at 0%–2% CC for six consecutive quarters.
Coforge
- Operating Leverage: Slashed overheads from 19.5% (FY25) to 15.3% (FY26) to 13.9% (Q1 FY27). Absolute G&A staff costs dropped -10% while revenue grew +36%. Attrition is sector-lowest at 10.4%.
- Accounting Note: EBIT margins look ~160–170 bps lower than EBITDA due to ~$40M/year non-cash M&A amortization (this is an accounting effect, not operational margin decay).
Persistent Systems
- Growth Streak: 25 consecutive quarters of sequential revenue growth (Q1 FY21 to Q1 FY27).
- Slowing Momentum: YoY growth rate is easing: 18.8% ➔ 17.6% ➔ 17.3% ➔ 16.2% ➔ 16.1%.
- Nagarro Bet: Nagarro acquisition expands revenue to ~$2.9bn (1.7x size), but adds lower-margin revenue and debt service.
Market Reality & The Analyst Disconnect
Six-Month Stock Price Performance
The market has been adjusting to these operational realities. Over the six months leading to mid-August 2026, 7 out of 9 stocks declined:
Stock Performance Breakdown (6-Month Period to August 2026)
| Company | 6-Month Return | Max Drawdown | Recovery Off Low |
| Coforge | +34.0% | -41.2% | +69.9% |
| Persistent | +0.9% | -23.9% | +29.1% |
| HCLTech | -3.4% | -28.9% | +33.3% |
| IBM | -8.8% | -35.8% | +12.5% |
| Cognizant | -11.5% | -41.6% | +50.3% |
| TCS | -11.6% | -27.1% | +20.0% |
| Infosys | -12.8% | -27.7% | +18.6% |
| Wipro | -14.3% | -22.0% | +9.0% |
| Accenture | -18.0% | -44.0% | +46.5% |
Three Dates That Moved the Entire Sector
- 12 February 2026 (Sector Crash): All nine stocks fell together in a single session (Cognizant -7.2%, Coforge -6.5%, Infosys -5.8%, TCS -5.5%). Surging volume confirmed a broad sector de-rating rather than company-specific news.
- 18 June 2026 (US Breakaway): Accenture plunged -18.0% in one day (41.7M shares, 8x normal volume). Cognizant dropped -10.5%. Indian peers moved less the next day (Infosys -6.7%, TCS -3.6%).
- 27–31 July 2026 (Sharp Rebound): The sector recovered lost ground inside one week (Cognizant +11.3%, Coforge +10.3%).
The Analyst Disconnect
Broker consensus shows two clear divergences from underlying disclosures:
- HCLTech (Under-appreciated by consensus): Carries only 36% positive analyst ratings and trades 7.9% above its mean target price. Yet it delivers ~2.9% organic growth, transparent AI disclosures, and strong software cash flow.
- Infosys (Over-appreciated by consensus): Carries 57% positive ratings (with only 7% negative). Yet its disclosures show subcontractor costs up ~73%, weak FY27 guidance (1.5%–3.0%), and stagnant non-AI business growth.
The Retail Investor Bottom Line
For retail investors, the takeaway from these disclosures is clear:
AI is Deflationary for Now
AI is currently driving contract price deflation rather than net-new revenue growth. It helps win renewals, but it shrinks the billable value of engagements.
Watch the Cash Flow Shift
Large-cap Indian IT firms face a growing conflict between high dividend payout ratios (82%–99%) and the billions needed for AI data centers.
Mid-Caps Have Shifted Risk Profiles
Coforge and Persistent are no longer low-risk, net-cash plays—they are now debt-leveraged acquirers that must execute seamlessly to service debt.
Look Past TCV Headlines
Record order book announcements mean little if ramp-up delays and deal pricing discounts flatten actual revenue growth.
