CGI Inc. (NYSE: GIB / TSX: GIB.A) — A Build-and-Buy Cash Machine at Its Cheapest-Ever Multiple, the Organic Engine Idling

Independent fundamental equity research. As-of date: June 19, 2026 · Figures CAD unless stated; NYSE GIB price = USD, TSX GIB.A = CAD. CGI is a Canadian foreign private issuer (files 40-F/6-K, reports IFRS, fiscal year ends September 30).

⚡ Claude’s Take

This is the author’s own subjective opinion and general information, not investment advice. It is the single place in this article where a position and a directional valuation zone are taken; the analysis that follows takes no position and contains no price target — it discusses valuation only as embedded expectations and scenarios.

Verdict: ACCUMULATE ON WEAKNESS / HOLD — an owner-operated cash compounder de-rated to a structural-impairment multiple on a GenAI debate the numbers have not yet settled, where the downside looks largely priced and the burden of proof sits with the bull. Constructive in tranches below ~$60 USD; illustrative valuation zone ~$55 (bear) / ~$85 (base) / ~$125 (bull) USD. Conviction: medium-low.

The single most important fact about CGI today is the gap between its tape and its cash flow. The stock is down ~42% over the past year and sits at the 2nd percentile of its own ten-year valuation range on P/E, P/B and P/S simultaneously — its cheapest multiple in a quarter-century — while the business still threw off ~$2B of free cash flow (≈10% FCF yield), ran a ~16.4% adjusted-EBIT margin, and shrank its share count ~30% over a decade. Its most recent leg down (−7.3% on June 18) was not its own news at all: Accenture cut guidance and fell ~20%, dragging the entire IT-services complex with it. Empirically (the factor read), this is an abandoned, low-beta (0.61) quality compounder in the worst drawdown of its life — a falling knife, not a euphoric unwind. That is the bull’s whole case: you are being handed a disciplined, founder-aligned consolidator at a no-growth price, with a buyback compounding per-share value while you wait.

But I stop short of pounding the table, because the bear case is evidenced, not imagined. CGI’s organic, constant-currency growth is ~0–2% and decelerating (1.6% last quarter; Canada and U.S. Federal are shrinking), so the “build-and-buy” model now depends on the “buy” — and marginal ROIC is already falling (16.0%→13.6%→13.1%) as 2025 goodwill stacks onto a balance sheet whose tangible book is negative ~$2.4B. GenAI deflation of the labor-services model is observable in peers (TCS shed 23,400 jobs; Accenture’s bookings fell), while CGI’s DigiOps “double-digit efficiency” answer is asserted, not yet visible in its organic growth or margin. The market is pricing the bear case as the base case (a reverse-DCF implies ~0% real per-share growth in perpetuity), so today’s price ≈ the bear scenario and the base case sits ~40% higher — attractive asymmetry — but there is no near-term catalyst and the next two prints carry the proof. Framing: deep-value / abandoned-quality with a catalyst problem — buy the engine, not the quarter, and in pieces. Conviction is medium-low, dragged below the author’s Accenture call by worse organic growth, M&A dependence, falling ROIC, and dual-class entrenchment (founders hold ~56% of votes on ~11% of the economics), and lifted by a cheaper price and a more secure, backlog-protected cash stream.

One-line tag: “Idling, not broken — buy the engine, not the quarter.”

Conviction & triggers (medium-low). Flips bullish if constant-currency organic growth re-accelerates — specifically Canada and U.S. Federal stop shrinking — with the adjusted-EBIT margin holding ≥16%, which would prove GenAI is net-accretive (monetized via DigiOps) rather than deflationary. Flips bearish if organic growth and margin soften together, group book-to-bill breaks below 100%, or reported ROIC drifts toward WACC (~<12%) — the signature that GenAI is taxing both price and volume and that build-and-buy can no longer outrun it.

Changes since prior coverage: none — this is the author’s first dedicated report on CGI. It is, however, the close cousin of a June 11 Accenture note (same “quality compounder de-rated on an unsettled GenAI debate” thesis); CGI is the smaller, Canadian, serial-acquirer, government-heavier version — cheaper, slower-growing organically, and more M&A-dependent.

📈 Stock Price Action — Five-Year Event Map

CGI’s NYSE-listed Class A shares (GIB, USD) trace a five-year round-trip from roughly $90 (mid-2021) down to a cyclical trough near $74.6 (Sep 2022), then a long compounding climb to an all-time high of $121.44 on Feb 13, 2025, followed by a relentless ~16-month de-rating to $61.28 at the close on June 18, 2026 — which is also the lowest close in the trailing 60 months. The stock now sits ~49.5% below its February-2025 peak and near the floor of a 52-week range of ~$61.3 to ~$105.5 (FACT: AZI 5-year price CSV, NYSE GIB adjusted close, accessed 2026-06-19). The Montreal listing GIB.A trades the same arc in CAD; the USD path additionally absorbs CAD/USD translation. The drawdown is almost entirely a 2025-2026 phenomenon: the slide began at the record high and accelerated through a sequence of soft-organic-growth prints and a sector-wide GenAI/IT-services derating, not a company-specific blow-up.

Cycle narrative. (1) Through 2022, GIB behaved like the low-beta name its factor profile implies — it drifted down ~17% with the rate-driven market rather than on any CGI-specific news. (2-3) From the late-2022 trough the stock compounded ~60% to its February-2025 record, tracking a clean fundamental run: revenue reacceleration (FY2023 +11.1% reported on the post-pandemic demand snap-back; the BJSS UK acquisition came later, closing Feb 2025), FY2024 GAAP EPS +10% to $7.31, ~$2B annual FCF, and a ~3.8%/yr share-count reduction from buybacks (FACT: the multi-year financials; FY2025 PR, 2025-11-05). (4-5) The de-rate began at the high: a −6% day on Feb 28, 2025 marked the first leg, then a ten-month grind as constant-currency organic growth decelerated and a large government end-market (US federal/DOGE, UK austerity) became an overhang (INTERPRETATION; organic decel is FACT per Q2-FY26 PR). (6) Feb 3, 2026 was pure sector contagion — the GenAI “AI is eating consulting” derating that took Gartner down 30% the same day on a zero-2026-growth guide and was catalyzed by Anthropic’s legal-automation tool launch (Fortune, 2026-02-16; Morningstar); CGI did not report that day. (7) April 29, 2026 is the one large move with a CGI-specific cause: the Q2-FY2026 print showed reported revenue +3.3% but only +1.6% constant currency and ROIC slipping to 13.1% from 15.4%, and the stock fell −10.8% on ~1.92M shares versus a ~485k average (FACT: Q2-FY26 PR, 2026-04-29; AZI CSV). (8) The most recent leg, June 18, 2026, was again sector read-across: Accenture cut its FY2026 revenue-growth guide to 3-4% citing weak demand and a federal slowdown, fell ~16-20% on its worst day in years, and dragged the whole IT-services complex down with it (CGI −7.3%) — CGI did not report (CNBC, 2026-06-19; TechTimes, 2026-06-18). Five of the eight largest moves are macro/sector-driven, not company-specific — the signature of a low-beta name caught in a sector bear market.

1. Executive Summary

CGI Inc. is the world’s fifth- or sixth-largest independent IT and business-consulting firm — a Montreal-headquartered, founder-controlled services company that sells strategic consulting, systems integration, long-term managed IT/business-process outsourcing (roughly half its ~$16B of revenue and recurring), and proprietary IP/software solutions (~20% of revenue, its highest-margin layer). For five decades it has compounded shareholder value through a disciplined “build-and-buy” model: thin organic growth plus a continuous, metro-market acquisition roll-up, rapid integration to a common operating model, high cash conversion, and a relentless buyback. The arithmetic is striking — over FY2017–FY2024 revenue compounded ~4.4%/yr but diluted EPS compounded ~11.5%/yr ($3.41→$7.31), the gap manufactured by margin expansion and a ~24% reduction in the share count. FY2025 revenue was CAD $15.9B (+8.4% reported, +4.6% constant-currency), adjusted EPS $8.30, operating cash flow $2.23B, and free cash flow ~$1.96B. (Figures CAD; FYE September 30.)

The reason this memo exists is a violent dislocation between that record and the share price. CGI’s NYSE line (GIB) has fallen ~42% over the past year to ~$61 USD — its lowest level on record, ~49% below its February-2025 peak — and now trades at the 2nd percentile of its own ten-year valuation range on P/E, P/B, and P/S simultaneously (~11× GAAP / ~10× adjusted earnings, ~7× EV/EBITDA, ~10% FCF yield). The de-rating is overwhelmingly sector-driven: five of the eight largest one-day moves of the past five years were macro or peer read-across, not CGI news, and the most recent (−7.3% on June 18, 2026) was pure sympathy with Accenture’s guidance cut and ~20% collapse. Empirically the stock screens as a low-beta (0.61), negative-momentum, abandoned-quality name — a falling knife rather than a crowded long unwinding.

The investment debate is unusually clean because the bull and bear agree on the facts and disagree only on their durability. The bull: a 13–16% ROIC, ~$2B-FCF, owner-aligned consolidator at its cheapest-ever multiple, with $31.5B of contracted backlog (~2× revenue), ~$2.2B of M&A firepower deployable into a cheap private market, and a buyback shrinking the share count ~3.5%/yr into a trough multiple — where even zero organic growth still compounds per-share value, and any re-rating is upside. The bear: organic constant-currency growth has decelerated to ~1.6% with Canada and U.S. Federal shrinking, so the model now needs M&A to grow; GenAI is observably deflating the labor-arbitrage/time-and-materials core (TCS shed 23,400 jobs; Accenture’s bookings fell and it cut its FY2026 growth guide to 3–4%); U.S. federal “DOGE” austerity cut CGI’s U.S. Federal revenue ~11%; marginal ROIC is already falling (16.0%→13.1%); goodwill (~$11.7B) now exceeds equity, leaving tangible book at roughly −$2.4B; and founders entrench control with ~56% of votes on ~11% of the economics.

On the numbers, the market is pricing the bear case as the base case: a reverse-DCF at today’s price underwrites roughly 0% real per-share growth in perpetuity — below CGI’s own historical algorithm. Scenario math (explicitly not a price target) brackets a wide outcome range because the single swing variable — whether GenAI deflates or augments the managed-services model — is genuinely unresolved: a bear path roughly equal to today’s price, a base path materially above it on a mere re-rate to CGI’s own mid-cycle multiple, and a bull path that requires the GenAI fear to invert into a positive. The downside, in other words, appears largely discounted; the upside hinges on a debate that will be settled not by the tape but by CGI’s next four-to-six prints — specifically by whether organic growth and margin hold together (GenAI accretive) or erode together (GenAI deflationary). This report takes no position and sets no target; it maps those expectations and the evidence on each side. The opening Claude’s Take is the one place a view is expressed.

2. Business Overview

2.1 What CGI does

CGI Inc. is an IT and business-consulting services firm — the fifth- or sixth-largest independent in the world by revenue, and the largest headquartered in Canada. It sells four end-to-end “levers” to large enterprises and governments (FY2025 AIF, filed Dec 17 2025): (1) business and strategic IT consulting, (2) systems integration (SI), (3) managed IT and business-process services (outsourcing), and (4) intellectual-property (IP) based business solutions — proprietary software platforms sold as “business-platforms-as-a-service.” FY2025 revenue was CAD $15,912.7M (+8.4% reported, +4.6% constant currency; FY2025 MD&A §3.4), generated by ~94,000 “consultants and professionals” the company calls CGI Partners, of whom 87.5% are shareholders (FY2025 AIF, “Human Resources”). Delivery is split between client-proximate onshore teams and an offshore/nearshore network centred on India and the Philippines (the Asia Pacific Global Delivery segment) and Eastern Europe.

The portfolio is deliberately full-stack: CGI wants to advise on the strategy, build the system, then run it for a decade. The genuinely differentiated piece is the IP layer — proprietary platforms such as Momentum (a federal ERP suite “trusted by more than 180 organizations across the three branches of the U.S. federal government, including intelligence and defense,” FY2025 AIF) and CGI Advantage (a government ERP for states/municipalities). Management discloses IP at roughly 20–22% of revenue and at structurally higher margin than labour-based services (company filings; consistent with prior-year MD&A mix charts). Interpretation: this is the one part of CGI that looks like software rather than bodies-for-hire, and it is the most defensible. Open question: CGI presents the revenue-by-type split (managed services vs. SI&C vs. IP) only as image charts in the MD&A, not as a text table — the precise FY2025 split could not be reconciled to a filed number and is carried as a disclosed-but-unverified ~20–22% IP figure.

2.2 How it makes money — contract structure and recurring mix

Revenue divides into two economic types. Systems Integration & Consulting (SI&C) is largely project work — time-and-materials (T&M) or fixed-price builds — which is higher-growth but non-recurring and re-wins constantly. Managed services / outsourcing is multi-year recurring revenue: a client hands CGI its application maintenance, infrastructure, or business-process operations under contracts that routinely run 5–10 years. CGI historically runs roughly half its book as managed services (the higher-stability half) and half as SI&C (company filings; prior-year MD&A mix charts — open question, exact FY2025 split is chart-only). Interpretation: the managed-services half is the ballast — sticky, renewable, and the source of CGI’s contractual visibility; the SI&C half is the growth-and-cyclicality lever that moves with discretionary IT budgets.

The visibility shows up in backlog: CAD $31.50B at Q2 FY2026 (Mar 31 2026) = ~1.9× annual revenue (Q2 FY2026 press release, Apr 29 2026). That is genuine forward cover — most of next year’s revenue is already contracted — and it is the single best quantitative argument that CGI is more durable than a pure staff-augmentation shop. Order intake remains positive: trailing-twelve-month book-to-bill of 108.4% at Q2 FY2026 (bookings $17.7B TTM; Q2 FY2026 MD&A §3.1), and 110.4% for FY2025 ($17.57B bookings). Caveat (management commentary is hypothesis): bookings include management estimates of demand-driven/volume usage and option years (AIF “Key Performance Measures” definitions), so book-to-bill >100% is a directional health signal, not a contracted-revenue guarantee.

2.3 Geographic operating segments

CGI is managed as nine geographic operating segments (realigned effective Oct 1 2025, moving Luxembourg into the Scandinavia/NW-Central-East unit; Q2 FY2026 MD&A §“Reporting Segments”). Segments are defined by where the work is delivered, not where the client sits. FY2025 revenue and the latest (Q2 FY2026) adjusted-EBIT margins:

Segment revenues marked * are inferred from Q2 FY2026 run-rates and prior-period MD&A; the four largest (W&SE, US Commercial/State, US Federal, Canada) are reconciled to FY2025 MD&A §3.4. Total reconciles to CAD $15,912.7M before eliminations (~$0.15B). Source: FY2025 MD&A §3.4; Q2 FY2026 MD&A §3.7.

The margin dispersion is the story. Canada (24.3%) and the U.S. units (13.8–16.6%) carry the firm; Asia Pacific shows ~30% adjusted EBIT margin but that is a delivery-cost centre whose output is sold internally to the onshore segments, so its “margin” overstates economics and is partly an intercompany artefact (eliminations were $34M in Q2’26). Western & Southern Europe (~13.5%) and Germany (8.4%) are the structural drags — competitive French/German markets, integration of acquisitions, and (Germany) low utilization. Interpretation: CGI’s blended ~16.4% adjusted-EBIT margin is a weighted average of a very good Canadian/government business and a mediocre Continental-European one; the European segments are where the “build-and-buy” roll-up has bought revenue faster than margin.

2.4 End markets — government is the anchor

CGI does not report a clean revenue-by-vertical table, but the concentration is unmistakable. The U.S. federal government alone was 14.1% of FY2025 revenue (12.3% in Q2 FY2026 as DOGE-related efficiency cuts and a shutdown bit; FY2025 MD&A §3.3.1, Q2 FY2026 MD&A §3.3.1) — CGI’s single largest client under IFRS common-control rules. Add Canadian federal/provincial, U.K., and Continental-European public sector, and government is comfortably CGI’s largest end-market in aggregate (interpretation; CGI’s own AIF repeatedly frames government as core, and three of its top segments — Canada, U.S. Federal, U.S. State Government — are government-anchored). The commercial book spans banking & capital markets, health, utilities & energy, manufacturing, retail, telecom, and transportation (AIF executive-residence titles map to these verticals). Interpretation: the government tilt is a double edge — sticky, mission-critical, recession-resistant demand, but exposure to fiscal austerity (U.S. DOGE, U.K. spending restraint) that is visibly compressing the U.S. Federal segment right now (revenue −11.1% YoY in Q2 FY2026).

2.5 The “Build and Buy” strategy, metro-market model, and member culture

CGI’s operating identity rests on three pillars investors must understand:

• Build and Buy — management’s explicit four-pillar growth doctrine (FY2025 AIF, “Mission, Vision and Strategy”): Pillar 1 win/renew/extend contracts; Pillar 2 win new large managed-services deals (these two = organic “Build”); Pillar 3 metro-market acquisitions; Pillar 4 large transformational acquisitions (these two = “Buy”). CGI states outright it “will continue to be a consolidator in the IT and business consulting services industry.” Interpretation: this is a serial-acquirer flywheel — and, critically, the model needs M&A, because organic constant-currency growth was only ~+1.6% in Q2 FY2026 and ~+4.6% in FY2025 (a chunk of which was acquired). Growth quality is examined later; the structural point here is that the business is engineered around disciplined roll-up, not organic compounding.

• Metro-market / proximity model — “We live and work near our clients,” with local CGI Partners “who speak our clients’ language,” complemented by the offshore global-delivery network (AIF, “Executing Our Strategy”). Acquisitions are chosen to deepen a metro market and “mirror the IT spend of each metro market over time.” Interpretation: proximity is a real, if modest, source of switching cost and relationship lock-in (developed in the Competitive Position section).

• Member/ownership culture — 87.5% of ~94,000 Partners own CGI stock via the Share Purchase Plan, alongside a Profit Participation Plan (FY2025 AIF). Founders retain control through a dual-class structure (Class B multiple-voting shares; founder Serge Godin and family). Interpretation: high employee ownership plausibly aids retention and delivery discipline in a people-business where attrition is the enemy; the dual-class control is an alignment-and-governance question handled in the Capital Allocation section. Note: leadership has turned over rapidly — George Schindler retired (Sep 30, 2024) → François Boulanger (Oct 2024) → Tim Hurlebaus (President & CEO, May 12, 2026); Serge Godin remains Founder & Executive Chairman (governance detail in the Capital Allocation section).

2.6 Verdict — business quality and durability

A genuinely good, but not great, services business — durable in the half that recurs, commoditized in the half that doesn’t. The evidence for durability is concrete: ~$31.5B backlog (~1.9× revenue), >100% book-to-bill, ~50% recurring managed-services revenue, mission-critical government IP (Momentum, Advantage) with high switching costs, 87.5% employee ownership, and a Canadian/government core earning ~16–24% segment margins. The evidence against treating it as a high-quality compounder is equally concrete: organic growth is structurally low-single-digit, so reported growth leans on serial M&A; the European third of the firm earns mediocre 8–14% margins; and the largest end-market (government) is exposed to fiscal austerity that is currently shrinking the U.S. Federal segment. The honest characterization is a disciplined, cash-generative roll-up of a structurally average industry, whose quality is a function of CGI’s execution and capital discipline far more than of any industry tailwind. The moat that underpins the durable half is dissected in the next section.

3. Industry Dynamics

3.1 Market structure, size, and growth

CGI competes in the global IT & business consulting services industry — a roughly $1.8–1.9 trillion market spanning strategy consulting, systems integration (SI), application and infrastructure managed services/outsourcing, and packaged IP/software. Gartner’s April 2026 forecast puts worldwide IT spending at US$6.31 trillion in 2026 (+13.5% y/y), with IT services the single largest segment at >US$1.87 trillion (~30% of total spend) (Gartner press release, 2026-04-22; CIO Dive, 2026-04). FACT. This is a vast, structurally growing addressable market — on its face a tailwind.

But the headline TAM is misleading for a firm like CGI, and reading it as a clean positive is the first analytical trap. INTERPRETATION: Gartner explicitly attributes the upgraded 2026 growth to “AI infrastructure, software, and IaaS,” describing a “multi-speed IT market” where hyperscaler purchases and AI-centric software outperform traditional categories — data-center systems alone are forecast +55.8% to >US$788B (Gartner, 2026-04-22). The growth is concentrated in hardware and cloud capacity, not in the labor-based consulting, systems-integration and staff-augmentation services that constitute the bulk of CGI’s revenue. The relevant sub-market for CGI — discretionary consulting and project-based SI — is growing low-single-digits at best, as the June 2026 sector-wide guidance resets confirm (discussed below). A trillion-dollar TAM growing double-digits coexists with a services book growing ~1% organically. The TAM is real; CGI’s slice of it is not riding the fast lane.

The industry is highly fragmented with no dominant share-holder. Accenture, the largest pure-play, generates ~US$70B revenue — under 4% of the IT-services market — and the top ten vendors collectively hold well under a third. The competitive field splits into recognizable cohorts: (i) global strategy/SI leaders (Accenture, IBM Consulting, Deloitte/the Big Four); (ii) Indian-heritage offshore-led players (TCS, Infosys, Wipro, HCLTech, Cognizant — built on labor-arbitrage delivery pyramids); (iii) Western/European roll-ups (Capgemini — the closest build-and-buy analog to CGI — Atos, DXC); (iv) CGI itself (~$16B revenue, Canadian, government-heavy, serial acquirer); and (v) US-government specialists (Booz Allen, Leidos, SAIC, CACI). FACT. Fragmentation, near-zero switching frictions at the project level, and the absence of a share leader are the signature of a structurally competitive — not oligopolistic — industry.

3.2 Value chain, profit pools, and labor-arbitrage economics

The services value chain runs strategy advisory → systems integration/implementation → managed services/outsourcing → proprietary IP/software, with profitability and stickiness rising left-to-right at the managed-services end but moat erosion at the advisory/staff-augmentation end:

• Strategy/advisory — high billing rates, low capital, but project-based, non-recurring, and the most exposed to client budget cuts and in-sourcing. Low barriers to entry: a credentialed partner and a deck.
• Systems integration & consulting (SI&C) — the project core; time-&-materials (T&M) and fixed-price work. Mildly recurring through follow-on work but fundamentally discretionary.
• Managed IT & business-process services (outsourcing) — multi-year contracts (often 5–10 years), embedded in client operations, the source of CGI’s revenue visibility (backlog $31.5B ≈ 2× annual revenue; book-to-bill 103% in Q2 FY2026 — CGI Q2 FY2026 MD&A, 2026-04-29). This is the genuine demand-side captivity in the model.
• IP/software solutions — ~20–22% of CGI revenue, higher-margin, the only piece with intangible-asset characteristics (Q2 FY2026 MD&A).

The dominant historical profit engine across the Indian-heritage and roll-up cohorts has been labor arbitrage: bill onshore rates, deliver from low-cost offshore/nearshore centers (CGI runs Asia Pacific “global delivery” centers in India, Philippines, Eastern Europe — company filings). The economics depend on a pyramid of many junior engineers under fewer seniors, and on headcount growing roughly in line with revenue. INTERPRETATION: this is precisely the structure GenAI threatens (the central debate below) — if code generation, testing and L1/L2 support automate, the pyramid’s base shrinks, T&M hours deflate, and the arbitrage spread compresses.

3.3 Barriers to entry — low at the top, real at the bottom

Applying the Greenwald taxonomy: the industry has no broad, durable barrier to entry. At the high-margin advisory/SI end, barriers are minimal — talent is mobile, there are no patents, no network effects, and minimal scale economies in delivering a project. Share is not stable at the project level; clients multi-source and re-bid routinely. By the Greenwald market-share-stability test, persistent share churn = weak/absent moat. VERDICT input: the industry as a whole is a no-moat, competitive-equilibrium business where excess returns accrue only to specific firms via execution, not to the industry structure.

Where barriers do exist, they are local and demand-side, not industry-wide:

• Managed-services/outsourcing captivity — once a vendor runs a client’s core systems, switching costs (re-transition risk, knowledge loss, re-badging staff) are high. This is genuine demand captivity (Greenwald) and underpins CGI’s backlog. FACT/INTERPRETATION.
• Regulated government work — security clearances, sovereign-data residency, agency-specific certifications, and incumbency on multi-year vehicles create real entry barriers in US Federal, Canadian, UK and EU public-sector IT. This is CGI’s most defensible turf — and its most policy-exposed (the government-demand discussion below).
• Scale + delivery footprint — at the very top (Accenture/TCS scale), there is a modest economies-of-scale + captivity advantage in winning mega-deals and amortizing global delivery infrastructure. CGI, at ~$16B, has sub-scale relative to Accenture/TCS but enough density in its chosen metro markets to compete; it is not the cost or scale leader.

Net: barriers are thin and segment-specific. CGI’s defensibility comes from the managed-services and government pieces, not from any industry-level moat.

3.4 The central debate — GenAI’s impact on IT services

This is the question that re-rated the entire sector in 2025–2026 and the single most important industry variable for CGI.

The bear case (deflation/disintermediation). GenAI directly attacks the labor-based revenue model. The mechanism: code generation, automated testing, and AI-assisted L1/L2 support let the same client outcome be delivered with fewer billable hours and fewer junior engineers, deflating T&M and staff-augmentation revenue and collapsing the pyramid that funds arbitrage margins. The supply-side evidence is now concrete, not theoretical: TCS cut >23,400 jobs in FY2026 (headcount 607,979 → 584,519), explicitly citing AI-driven “skill mismatch”; Indian IT majors collectively added a net 17 employees across nine months of FY2026 (Computerworld, 2026; Asia News Network, 2026). An industry analyst quoted by Asia News flags this as “a structural – not cyclical – correction driven by AI-led productivity compression, slower global discretionary tech spending, and a pivot away from legacy services.” FACT (headcount); INTERPRETATION (causation). Clients can also in-source more capability as AI lowers the skill threshold for building software, and code itself commoditizes. The demand-side read-through arrived on 2026-06-18: Accenture cut FY2026 revenue-growth guidance from ~4–5% to 3–4% and reported new bookings down ~2% y/y, citing weak discretionary demand and a US-federal slowdown; the stock fell ~16–20% (its worst day on record) and dragged the sector (Cognizant, Infosys ADR, Capgemini, CGI all −5% to −10%) (Reuters/CNBC/Sherwood, 2026-06-18/19). CGI itself reported Q2 FY2026 organic (constant-currency) revenue growth of only ~1% (Q2 FY2026 MD&A) — the model is already near stall-speed before any AI deflation fully lands.

The bull case (re-platforming super-cycle / productivity monetization). GenAI is also a demand driver and an efficiency tool the scaled incumbents capture. Three threads: (i) every enterprise must modernize legacy estates, clean and govern data, and re-architect applications to deploy AI — a multi-year modernization wave that favors trusted integrators with deep installed bases; (ii) vendor consolidation — as AI projects raise the stakes, clients concentrate spend on a few scaled, trusted partners, advantaging incumbents over boutiques; (iii) firms monetize their own productivity gains by embedding AI into fixed-price managed services and keeping the savings — CGI’s framing of DigiOps, “our AI-powered service delivery approach that can deliver sustainable double-digit efficiency improvements” (Q2 FY2026 MD&A), partnered with OpenAI and Google Cloud (company filings). In a managed-services contract, AI efficiency accrues to the vendor’s margin, not just the client’s bill. FACT (CGI positioning); HYPOTHESIS — management commentary, unvalidated by margin data yet.

Skeptical synthesis. The bull case is real for the managed-services/outcome-based book and genuinely cushions CGI relative to pure staff-aug peers — CGI’s ~20% IP and large outsourcing mix is a structural advantage here. But the bull case is also the convenient management narrative, and the burden of proof sits with it: there is no margin or organic-growth evidence yet that CGI is net monetizing AI rather than passing savings to clients under competitive pressure. The bear evidence (sector guidance cuts, structural headcount declines, ~1% organic) is observed; the bull evidence (DigiOps efficiency) is asserted. OPEN QUESTION for Variant/Valuation: does AI net-add to CGI’s revenue and margin, or net-deflate them? Watch organic growth and the managed-services margin trajectory over the next 4–6 quarters — that is the falsification test.

Marathon capital-cycle read. Supply-side, the industry is in the early phase of capital/labor leaving. For two decades, high IT-services returns attracted capital and headcount; the pyramids ballooned. Now the asset-growth (here, headcount-growth) anomaly is reversing: net hiring has gone to zero, TCS and peers are shedding staff, and the sector has de-rated hard — CGI sits at ~11× P/E, its cheapest-ever multiple (2nd percentile of its own 10-year history; AZI valuation_index, 2026-06-18), down ~42% over twelve months. INTERPRETATION: Marathon’s lens says capital exiting a fragmented, over-supplied industry is normally constructive for surviving low-cost operators — fewer junior bodies chasing the same work can lift returns on the remaining managed-services base. But this cycle has a confound: the capital is leaving because the unit of production itself (billable human hours) is being automated away, so reduced supply may simply track reduced demand for labor rather than tightening a stable market. The de-rating prices a structural impairment, not merely a cyclical trough. Which it is — terminal-decline or oversold-quality — is unresolved by the tape alone.

3.5 Government demand — structural factor and live headwind

Government is CGI’s largest single end-market — it is the top vertical across the Canada, US, UK/Australia and most European segments, alongside a dedicated U.S. Government segment (Q2 FY2026 MD&A). This cuts both ways structurally.

Tailwind (long-run): public-sector digitization is a durable, multi-decade demand pool; sovereignty/data-residency rules (EU, UK, Canada) advantage local-delivery incumbents over offshore-only Indian players and reward CGI’s onshore/nearshore footprint; clearances and incumbency raise entry barriers (as the barriers-to-entry discussion notes). FACT/INTERPRETATION.

Headwind (now): the demand pool is under fiscal and political pressure simultaneously across CGI’s three largest geographies. In the US, DOGE-driven federal cost-cutting cancelled ~$5.1B of consulting/IT contracts in 2025 (DoD found ~$4B of savings), hitting Accenture Federal (~$193M cancelled), Deloitte (~$473M), Booz Allen and Leidos (Washington Technology, 2025; Inc., 2025), with ~$65B of consulting fees flagged as at-risk and 2026 expected to bring more of the same (Washington Technology, 2025-12). CGI’s own Q2 FY2026 results show the damage: U.S. Government segment revenue fell to ~$618.8M from ~$671.7M (≈ −7%), explicitly attributed to “government efficiency initiatives and shutdown” (Q2 FY2026 MD&A, 2026-04-29). FACT. Accenture’s June 18 guidance cut was likewise pinned partly on US-federal weakness (≈1% drag) (Reuters, 2026-06-18). UK fiscal consolidation and uneven European public budgets compound the picture. INTERPRETATION: CGI’s defensive government tilt — usually a stabilizer through commercial cycles — is, in this specific window, a source of the organic-growth stall, not a buffer against it.

3.6 Cyclicality, FX, and macro sensitivity

IT services is moderately discretionary, not deeply cyclical: the managed-services/outsourcing base is contracted and counter-cyclical (clients outsource to cut cost in downturns), while SI&C/advisory/staff-aug is the discretionary swing factor that contracts when budgets tighten — exactly what is happening now. CGI’s low beta (~0.61) and 2× backlog coverage reflect the contracted base. FX is a material reported-growth distortion for a CAD reporter earning across USD/EUR/GBP/SEK: CGI’s Q2 FY2026 revenue rose +3.3% reported but only ~+1% constant-currency (Q2 FY2026 MD&A) — currency, not volume, did most of the reported work, and analysts must read constant-currency to see the true ~stall.

3.7 Verdict — structurally average-to-weak industry, mid-down-cycle, at an inflection

The industry is structurally fair-to-poor and currently in a down-cycle whose depth is the open question. Argued: it is large and growing in aggregate but fragmented, low-barrier, and competitive — by the Greenwald share-stability/ROIC tests it lacks an industry-wide moat; excess returns are firm-specific and execution-driven, not structural. The fast-growing part of the TAM (AI infrastructure/cloud) is not the labor-services part CGI sells. The defining risk — GenAI — most plausibly deflates the labor-arbitrage/staff-aug core (the bear evidence is observed: structural headcount cuts, sector guidance resets, ~1% organic growth), while the offsetting bull (AI re-platforming demand, vendor consolidation, DigiOps margin monetization) is plausible but still narrative, concentrated in the managed-services/IP layer where CGI is comparatively well-positioned. Government — CGI’s largest market — is a long-run tailwind but a present-tense headwind under DOGE/UK/EU fiscal pressure. Marathon read: capital and labor are leaving a crowded industry (constructive for survivors) but for the unusual reason that the product (human hours) is being automated (ambiguous, possibly impairing). The sector’s ~2nd-percentile de-rating prices structural impairment; whether that is an over-shoot on a still-cash-generative, managed-services-anchored incumbent or a correct read on terminal labor-deflation is the variant-perception crux this memo must resolve with CGI’s own numbers — not with the industry narrative.

4. Competitive Position

4.1 Framing the question in Greenwald terms

Greenwald & Kahn reduce competitive advantage to three genuine types: a supply-side cost advantage, demand-side customer captivity (switching costs, search costs, habit), and economies of scale combined with captivity — the most durable. Everything else (brand, “expertise,” good management) is either a manifestation of these or not a moat at all. The test that matters is operational: would CGI’s economics deteriorate if the advantage disappeared? And the empirical tells are market-share stability and ROIC persistently above cost of capital. Run honestly, CGI passes the ROIC test, partially passes the captivity test in its recurring/government/IP book, and fails the scale test — it is sub-scale versus the global leaders and has no supply-side cost edge. The net is a narrow, segment-specific moat (switching costs + relationships in managed services, government, and proprietary IP) wrapped around a commoditized, scale-disadvantaged core, where much of CGI’s realized advantage is not a structural moat at all but operational discipline: disciplined M&A and cost control. That is a defensible but candid verdict, developed below.

4.2 Scale economies — CGI fails this test

Scale is the most powerful moat in IT services because fixed costs (sales, delivery platforms, training, IP development, AI tooling) amortize over a larger revenue base and let the leader underprice on large deals. CGI is sub-scale. At CAD $15.9B (~US$11.4B) FY2025 revenue and ~94,000 staff, it is roughly:

Sources: company FY2025 releases (ACN 8-K FY2025; Capgemini FY2025 release; Infosys, TCS, Cognizant, DXC, Booz Allen FY2025 filings); CGI FY2025 MD&A/PR. CGI revenue converted at ~1.39 USD/CAD. Margin definitions vary by issuer (adjusted op vs. adj. EBIT vs. adj. EBITDA) — directional, not perfectly like-for-like.

The read: CGI cannot out-scale Accenture, TCS, or Infosys on the largest global deals — it lacks their delivery-platform reach, brand pull, and pricing room. Yet CGI’s adjusted-EBIT margin (~16.4%) and ROE (~19%, company filings) sit above Capgemini, Cognizant, and DXC, and only modestly below the Indian majors despite a far higher-cost onshore mix. Interpretation: CGI compensates for missing scale with disciplined cost management, a high-margin Canadian/government core, and a profitable IP layer — i.e., operational execution, not a structural scale moat. Greenwald would call this “running a commodity business well,” which is real and valuable but mean-reverts more easily than a true scale advantage.

4.3 Cost advantage and the global-delivery / labour-arbitrage question

CGI runs offshore delivery from India and the Philippines (the Asia Pacific segment, ~30% internal adjusted-EBIT margin), the classic labour-arbitrage cost lever. But CGI has no relative cost advantage: TCS, Infosys, Wipro, and Cognizant are far more offshore-weighted and structurally cheaper, which is exactly why TCS earns ~25% margins. CGI’s mix is more onshore/proximity by design, which lifts client intimacy but means it competes against the low-cost players, not with their cost base. GenAI sharpens this into a genuine threat (examined below): to the extent CGI’s revenue is staff-augmentation / T&M coding and testing, generative-AI tooling deflates the billable-hours base and lets clients in-source — eroding the part of the book that was never moated to begin with. Verdict on cost advantage: absent.

4.4 Customer captivity and switching costs — the real, but partial, moat

This is where CGI does have something. Three captivity mechanisms are visible in the financials:

1. Embedded managed-services / outsourcing contracts. Roughly half of revenue is multi-year (5–10 yr) recurring managed services. Once CGI runs a client’s core applications or back-office operations, the cost, risk, and disruption of ripping it out are high — the canonical Greenwald switching-cost captivity. The financial fingerprint is the $31.5B backlog (~1.9× revenue) and >100% book-to-bill (Q2 FY2026): clients renew and extend rather than re-tender. This is a genuine moat in the recurring half of the business.

2. Mission-critical government IP. Momentum (federal ERP used by 180+ U.S. federal organizations including defense/intelligence) and CGI Advantage (state/municipal ERP) are deeply embedded systems of record with multi-year switching costs, security accreditations, and regulatory entanglement. Replacing a financial-management ERP across federal agencies is a multi-year, high-risk program — captivity is high and re-compete cycles favour the incumbent. This IP layer (~20–22% of revenue) is CGI’s strongest and most software-like moat.

3. Proximity / relationship lock-in. The metro-market model — local Partners who “speak the client’s language” — builds search-cost and habit captivity: clients default to the known local provider for follow-on work. Modest but real, and hard to value because it does not show up as a discrete line.

Pressure test: are these durable? The managed-services and IP captivity would show up in deteriorating economics if removed — losing a Momentum re-compete or a large outsourcing renewal directly cuts high-margin recurring revenue, which is the deterioration test passing. But the captivity is narrow: it protects the recurring/government/IP ~50–60% of the book, not the SI&C/staff-aug remainder, which is fully contestable and re-won every project. Network effects: none — CGI’s services do not get more valuable to one client because another client uses them (no two-sided network), so the network-effects claim that sometimes attaches to “platforms” does not apply here.

4.5 Share stability and the ROIC test

ROIC test — passes, with a goodwill caveat. CGI earns ROE ~19% (FY2024) and ROIC ~16% (company filings), comfortably above any reasonable ~8–9% cost of capital, sustained across a decade — the Greenwald signature of some durable advantage. The caveat is that CGI’s invested capital is ~57% goodwill (~$9.5B ≈ 100% of equity) from the build-and-buy roll-up, so tangible ROIC is flattered by acquisition accounting; on a goodwill-inclusive basis the ~16% still clears the bar but signals that returns are partly bought (paying fair multiples for acquired books) rather than purely organic compounding — a Marathon “asset-growth” flag examined in the Financial Quality section.

Share-stability test — mixed. CGI has held or modestly gained share in its core geographies over a decade, but largely via acquisition (buying share in new metro markets) rather than organic share gains against Accenture/TCS. Organic constant-currency growth of ~+1.6% (Q2 FY2026) and ~+4.6% (FY2025, partly acquired) is roughly in line with — not ahead of — the market, which is the empirical signature of competitive parity, not dominance. In Greenwald’s framework, a firm that must keep acquiring to hold relative position has, at best, a local moat in each metro market and no aggregate share-taking advantage at the global level.

4.6 GenAI — the live threat to the un-moated core

The June 2026 sector sell-off (Accenture cutting FY2026 growth guidance to 3–4% and falling ~20%; CGI down sympathetically) crystallized the bear case: GenAI compresses the billable-hours model. Gartner’s 2025/2026 work frames IT outsourcing as “shifting beyond labour arbitrage,” with early AI-agent adopters reporting operational cost reductions “up to 38%” — productivity that, in a T&M world, deflates vendor revenue and lets sophisticated clients in-source routine build/maintain work (Gartner press releases, 2025–2026). Interpretation: this is most dangerous precisely where CGI is least moated (SI&C/staff-aug) and least dangerous where it is most moated (embedded managed services, government IP, regulated systems of record). CGI’s own counter — CGI DigiOps, an “AI-powered service delivery approach” claiming “sustainable double-digit efficiency improvements,” with OpenAI and Google Cloud partnerships — is the bull rebuttal: monetize AI productivity inside fixed-fee managed services and capture share as clients consolidate vendors. This is a hypothesis, not yet evidence — there is no disclosed metric proving DigiOps is net-accretive to revenue rather than merely defensive. The honest position: GenAI is a real, asymmetric risk to the commoditized half and an unproven opportunity in the moated half.

4.7 Verdict — a narrow moat plus operational discipline, in a crowded market

CGI has a real but narrow moat, and its broader edge is execution rather than structure. The durable advantage is demand-side captivity (Greenwald) in three places — embedded managed-services/outsourcing contracts, mission-critical government IP (Momentum/Advantage), and metro-market relationships — and it passes the deterioration test there: lose those renewals and high-margin recurring revenue falls directly. It also passes the ROIC test (~16–19% returns above cost of capital for a decade). But it decisively fails the scale test (sub-scale vs. Accenture/TCS/Infosys, no supply-side cost advantage, no network effects), and the share-stability evidence shows parity, bought through M&A, not dominance. Strip away the captive recurring/IP book and what remains is a crowded, undifferentiated, GenAI-threatened systems-integration market where CGI’s apparent superiority over Capgemini, Cognizant, and DXC comes mainly from disciplined acquisition pricing and relentless cost control — operational virtues that are admirable, replicable, and not a structural moat. The investable question for later sections: is ~16% ROIC and a half-moated book durable enough to justify a quality multiple, or is CGI a well-run consolidator in a commoditizing industry whose economics the market is right to be de-rating? The moat is real enough to defend the recurring base; it is not wide enough to make CGI a price-setter.

5. Growth History and Forward Opportunities

5.1 The headline CAGR flatters a low-organic engine

CGI’s top line compounded from $10,845M in FY2017 to $15,910M in FY2025 — a ~4.9%/yr revenue CAGR (CAD; FY end Sept 30; EDGAR ifrs-full XBRL, reconciled to the FY2025 press release, 2025-11-05). That number is the first place a skeptic should stop, because it bundles three structurally different growth sources that the market should not pay the same multiple for: (1) constant-currency organic growth (the only kind that reflects the durability of the business), (2) acquired revenue (bought, not earned — and CGI is a serial acquirer, as the Capital Allocation section examines), and (3) foreign-currency translation (a coin-flip that flatters the print when CAD weakens). CGI is unusually transparent here: the MD&A discloses constant-currency revenue growth, which strips FX but still includes acquisitions. Decompose the eight-year record and the picture sharpens (FACT, from each year’s CGI press release / MD&A):

Sources: CGI Q2 F2026 MD&A (2026-04-29); FY2025 PR (2025-11-05); FY2024/2023/2022 results PRs (cgi.com / SEC 6-K exhibits).

The structural conclusion (INTERPRETATION): once FX is removed, CGI’s underlying demand growth runs in a low-single-digit band — roughly 0–4% through the cycle, and the organic (ex-acquisition) component is thinner still. FY2024’s +0.9% CC is the cleanest data point because acquisitions were quiet that year — it implies organic growth essentially at zero. The strong-looking FY2022 (+10.5% CC) and FY2023 (+8.0% CC) were the post-pandemic spending snap-back plus smaller bolt-on deals, not a new structural growth rate; both have since faded. The FY2025 +4.6% CC and Q2’26 +1.6% CC are again acquisition-led: management attributes the increase in nearly every segment to “recent business acquisitions” (APSIDE, Online Business Systems, Comarch Polska — see Capital Allocation), “partially offset by” weak organic demand (Q2 F2026 MD&A, p.21–22). This is the Marathon capital-cycle fingerprint of a roll-up: the company buys the growth the end-market no longer hands it organically.

5.2 Bookings, book-to-bill and backlog — strong visibility, decelerating momentum

CGI’s forward indicators are genuinely a strength, and the bear case must engage with them honestly. Backlog stood at $31.50B at Q2 F2026 (~2.0× annual revenue), up from $31.45B (2.0×) at FY2025 and $28.72B (1.9×) at FY2024 (FACT; Q2 F2026 MD&A; FY2025/FY2024 PRs). Roughly $11.5B converts to revenue within twelve months, $10.7B in one-to-three years, and the rest beyond (Q2 F2026 MD&A, §3.1) — real multi-year visibility that few cyclicals enjoy, driven by long-dated managed-services/outsourcing contracts (the recurring-revenue core, discussed in the Competitive Position section). But two caveats temper it. First, backlog itself grows partly by acquisition — “backlog acquired through business acquisitions” is an explicit component of the definition (Q2 F2026 MD&A) — so the $28.7B→$31.5B climb is not all organically won work. Second, the book-to-bill trend is softening: FY2023 113.7% → FY2024 109.3% → FY2025 110.4% → Q2 F2026 103.8% (108.4% TTM) (FACT, segment table, Q2 F2026 MD&A §3.1). A book-to-bill barely above 100% means new signings are only just replacing consumed backlog — consistent with the low-single-digit organic run-rate, not with re-acceleration.

The segment-level TTM book-to-bill divergence is the most useful forward signal in the filing (FACT, Q2 F2026 MD&A):

Three segments are signing less than they consume (UK/Australia, Germany, Finland/Poland/Baltics). A sustained sub-100% book-to-bill is a leading indicator of future revenue decline in that geography — an honest red flag the bull narrative tends to skate past.

5.3 Segment growth divergence — where it grows, where it shrinks

The aggregate masks sharp dispersion (Q2 F2026, constant-currency, MD&A p.21–22): growth is acquisition-fed where it exists, and the organic core is flat-to-negative in CGI’s largest mature markets. U.K. & Australia (+16.5% CC) and Western & Southern Europe (+8.3% CC) led — but both are explicitly “mainly due to recent business acquisitions.” Asia Pacific Global Delivery (+7.2% CC) grew on internal offshoring demand (a margin story, not external revenue). Against that, Canada (−0.2% CC), U.S. Federal (−7.1% CC), and U.S. Commercial & State Government (−3.9% CC) shrank organically. U.S. Federal — a marquee government franchise (Stanley acquisition, 2010) — is being squeezed by the “U.S. federal government efficiency initiatives” (DOGE) and a government shutdown (MD&A); Canada, CGI’s home and highest-margin market (~24% adj. EBIT), is in low-grade organic decline on ended financial-services and government projects. The geographies that grow (Europe, UK) carry lower margins (~13–14% W&S Europe) and were bought; the geographies that are high-margin and organic (Canada, US Federal) are flat-to-shrinking. That is an unfavorable growth-quality mix.

5.4 Forward opportunities — credible levers, but each is double-edged

• AI-embedded managed services / DigiOps (INTERPRETATION, partly management hypothesis). CGI frames GenAI as embedded in delivery via “CGI DigiOps,” claiming “sustainable double-digit efficiency improvements,” with OpenAI and Google Cloud partnerships (Q2 F2026 MD&A; company filings). Management credits “AI-embedded managed services” for FY2025 double-digit EPS growth. The bull read: AI lets CGI win vendor-consolidation mandates as clients cut their supplier count, and monetize productivity. The bear read (and the more important one): GenAI deflates the labor-arbitrage / staff-augmentation core, lets clients in-source, and compresses the T&M billing base — the exact fear that drove Accenture to cut FY2026 guidance to 3–4% and the IT-services group to sell off ~5–20% on 2026-06-18 (company filings). DigiOps is unproven at scale as a revenue driver; treat the double-digit-efficiency claim as a hypothesis, not evidence.
• Large-deal / outsourcing pipeline & government digitization. CGI’s four growth pillars are organic wins, new large long-term managed-services contracts, metro-market M&A, and transformational M&A (Q2 F2026 MD&A). Government digitization across Canada/US/Europe is a large, sticky end-market — but it is simultaneously the source of the current headwind (US Federal/DOGE, UK/European austerity). Government is both the moat and the macro risk.
• M&A runway (FACT). CGI reported ~$2.24B of capital resources available at Mar-2026 against a credit facility increased to $2,500M (Q2 F2026 MD&A §4.2), plus ~$2B/yr FCF and ~1× leverage — ample “build-and-buy” firepower to keep buying growth. This is real and repeatable; it is also the admission that organic growth alone won’t move the needle.

5.5 Verdict — Growth quality: low

Low-quality growth, durably so. The disconfirming evidence is real: a $31.5B backlog (2.0× revenue), ~110% historical book-to-bill, and a genuinely diversified, recurring, government-anchored revenue base give CGI more visibility than almost any IT-services peer — this is not a company about to fall off a cliff. But on the question this section must answer — is the growth high-quality? — the answer is no. Strip FX and M&A and CGI’s organic growth runs ~0–2%, occasionally negative, in its largest and best markets (Canada, US Federal). The reported ~4.9% CAGR is manufactured roughly half from acquisitions and FX; the company structurally needs M&A to grow, which works only as long as the balance sheet and discipline hold (see Financial Quality and Capital Allocation). The decelerating book-to-bill (103.8% in Q2’26) and three sub-100% segments point to softening, not re-acceleration, into a GenAI demand shock that threatens the staff-augmentation core. Growth is defensible and visible, but slow, bought, and structurally challenged — high-quality in durability, low-quality in economics and organic momentum.

6. Financial Quality

All figures CAD unless stated. Sources: CGI FY2025 audited consolidated financial statements and MD&A (Form 40-F, filed 2025-12-17, EDGAR ex-99.2/99.3); FY2025 press release (2025-11-05); Q2-F2026 6-K/MD&A/press release (filed 2026-04-29); EDGAR ifrs-full XBRL (FY2017–FY2024). Every material figure below is reconciled to a primary filing; discrepancies flagged inline.

CGI’s financials are the clean part of the story. This is a high-return, low-capital-intensity, cash-generative compounder whose reported per-share economics — not its top line — do the work. The skeptic’s job here is to separate the durable cash machine from the accounting artifacts of a 49-year serial acquirer, and to test whether economics actually improve with scale or merely persist. The verdict, argued below: economics are genuinely good and the cash is real, but they have plateaued, not scaled, and FY2025 marked the first year the build-and-buy engine visibly diluted returns rather than compounding them.

6.1 The nine-year P&L: margin plateau, EPS engine running on financial gears

¹ FY2017–FY2022 adj. EBIT margin INTERPRETATION/ASSUMPTION — CGI’s current “adjusted EBIT” definition (EBT + net finance costs + restructuring/acquisition costs) was formalized later; the ~14.5–16% band is consistent with the FY2023–25 reconciled series and CGI’s historical “earnings from operations” margins, but is not a like-for-like reconciliation and should be treated as directional. FY2023–FY2025 figures are reconciled to the MD&A non-GAAP table (FACT). ² TTM adj. EPS approximated from FY2025 ($8.30) + H1-F26 adj. net earnings $944.4M vs. H1-F25 $929.7M on a falling share count; treat as ASSUMPTION for the Valuation agent.

The shape is unambiguous (FACT): FY2017→FY2024 revenue compounded at ~4.4%/yr while GAAP diluted EPS compounded at ~11.5%/yr (3.41→7.31). That ~7-point gap is not operating leverage in the normal sense — adjusted EBIT margin moved only from ~14.5% to 16.5% over eight years, roughly 25 bps/yr (INTERPRETATION: a margin plateau, not expansion). The EPS engine is instead financial: a −24% reduction in diluted share count (303M→232M, ~−3.8%/yr) plus modest net-margin lift. By Q2-F2026 the share count is down to 213.1M — a cumulative −30% since FY2017 (FACT, Q2-F26 PR). This is the single most important fact in the financial profile: a low-single-digit-growth business has manufactured low-double-digit EPS growth almost entirely through buybacks and EPS-accretive M&A. That is a legitimate, repeatable model — but it is a capital-structure story, not an organic-economics story, and it is mathematically self-limiting (you cannot shrink the share count forever).

The GAAP-vs-adjusted gap is the first quality-of-earnings flag. CGI adjusts out “restructuring, acquisition and related integration costs”: $73.2M (FY2024) → $213.2M (FY2025), net of tax — and on a pre-tax basis $96.9M → $285.0M (FACT, FY2025 PR non-GAAP reconciliation). FY2025 adjusted diluted EPS of $8.30 sits 13% above GAAP $7.35; the adjusted figure grew +8.9% while GAAP grew +0.5%. The skeptic’s point: these “one-time” add-backs recur every single year and are scaling with deal volume — $96.9M, then $285.0M. For a company whose strategy is perpetual acquisition and integration, restructuring and integration cost is an ordinary, structural cost of the business model, not a one-off. INTERPRETATION: adjusted EPS systematically over-states sustainable earnings power; the honest run-rate sits between GAAP and adjusted, closer to GAAP than CGI’s framing implies. The $196.8M FY2025 restructuring charge (a discrete completed program) is more defensibly excludable than the $88.2M of acquisition/integration cost, which is recurring.

6.2 Cash conversion: this is the genuine article

Where the P&L invites skepticism, the cash flow validates the business. FY2025 operating cash flow was $2,234.2M (FACT, ex-99.2), 14.0% of revenue and 1.0% above FY2024 — but flat OCF on +8.4% revenue is itself a yellow flag (working-capital drag; DSO rose to 45 days from 41, see the working-capital discussion below). Capital intensity is genuinely light: FY2025 PP&E purchases $116.6M + intangible-asset purchases $153.3M = ~$270M total capex (~1.7% of revenue) (FACT). That yields:

• Free cash flow (OCF − capex/intangibles) ≈ $1,964M (FACT, computed) — a ~12.3% FCF margin.
• OCF/net earnings = 1.35×; FCF/net earnings = 1.18× (FACT). Both above 1.0×, the signature of a business where reported earnings are conservatively stated relative to cash — the opposite of an earnings-management profile. Net income is NOT diverging unfavorably from cash flow (a key diligence test): cash exceeds accrual earnings, driven by D&A on acquired intangibles, deferred-revenue float on managed-services contracts, and stock-based compensation.

This is the strongest single argument that CGI is a real cash machine rather than an accounting roll-up: the goodwill on the balance sheet (discussed below) is the residue of cash deals, but the operating businesses bought with that cash throw off more cash than they report as profit. VERDICT on cash quality: genuine. The caveat is direction — OCF has been roughly flat at $2.1–2.3B for four years (2,112 → 2,205 → 2,234), so the cash machine is mature, not accelerating.

6.3 Returns on capital: high in absolute terms, but goodwill is now eating the incremental return

Profitability is unambiguously above cost of capital, but the multi-year trend is deteriorating, and this is the most important analytical finding in the section.

• ROE FY2025 ≈ 16.8% (net earnings $1,658.3M / avg. equity ~$9,855M), down from ~19.1% in FY2024 (FACT, computed; reconciles to net earnings and the FY2024/FY2025 equity of $9,428.0M → $10,282.3M). Note: the often-cited “~19%” ROE figure is the FY2024 number; FY2025 compressed to ~17% as equity ballooned faster than earnings.
• ROIC FY2025 = 13.6% as reported by CGI (FACT, FY2025 PR), down from 16.0% in both FY2023 and FY2024, and 13.1% on a Q2-F2026 TTM basis (FACT, Q2-F26 PR). CGI defines ROIC as last-twelve-months net earnings excluding after-tax net finance costs, over the four-quarter-average of invested capital (equity + net debt). My independent recomputation — NOPAT of ~$1,720M on average invested capital of ~$12,491M — yields ~13.8%, validating CGI’s 13.6% (FACT, computed; the ~0.2-pt gap is averaging convention).

The ROE > ROIC gap (16.8% vs 13.6%) is the goodwill fingerprint. ROE is levered by ~26% net-debt-to-cap; ROIC, which puts the full $11.7B goodwill and $3.45B net debt into the denominator, is the truer measure of the price paid for the earnings stream. The 240 bps ROIC decline from FY2024 to FY2025 is the single most important number in the financial section (INTERPRETATION): it is direct evidence, in CGI’s own preferred metric, that the FY2025 acquisition spree (detailed below) was bought at returns below the existing book’s, diluting the franchise’s return on capital even as it added to EPS. Under the Greenwald ROIC test, a 13.6% ROIC on a business with a low-double-digit WACC is a modest, shrinking excess return — consistent with a real but narrow moat (scale-based cost advantage and government switching costs) that is being progressively diluted by paying full price for incremental acquired revenue. Marathon capital-cycle read: capital is flowing into CGI’s roll-up (asset growth via M&A) precisely as incremental returns fall — the classic late-cycle pattern where high historic returns attract capital deployment that mean-reverts the return. Watch whether ROIC stabilizes ≥13% as deals integrate, or keeps sliding.

6.4 Balance sheet: conservatively levered, but tangible book is now negative

(FACT — all from ex-99.2 inline XBRL and the FY2025/Q2-F26 net-debt reconciliations.)

The leverage is conservative and the liquidity ample. Net debt of $3.57B at Q2-F2026 is ~1.1× EBITDA (adj. EBIT ~$2.6B + D&A ~$0.5B ≈ $3.1B EBITDA) and ~1.3× adjusted EBIT (FACT/INTERPRETATION). Net-debt-to-cap of 26.3% is up sharply from 16.2% a year earlier — entirely the funding of the FY2025–early-FY2026 deals (APSIDE, OBS, Comarch). CGI funded this with a $923.9M senior-unsecured-note issuance (FY2025) and an enlarged $2,500M credit facility, leaving “over $2.2B of capital resources available” (FACT, FY2025/Q2 MD&A). Net finance costs jumped to $83.7M (FY2025) from $27.9M (FY2024) as the company re-levered and cash fell (FACT). This is a deliberate, manageable re-leveraging of a chronically under-levered balance sheet, not financial stress — interest coverage remains >25×.

The quality flag is on the asset side, not the liability side. FY2025 goodwill of $11,744.8M now EXCEEDS total equity of $10,282.3M by ~$1.46B (FACT). Adding the $888M of other (largely acquired) intangibles, tangible book equity is approximately −$2.35B (FACT, computed). Goodwill is ~60% of total assets. This is the defining accounting characteristic of a build-and-buy roll-up: the entire equity of the company — and then some — is the unamortized purchase premium of 49 years of acquisitions. Two implications: (1) book value and P/B are nearly meaningless as value anchors here (the Valuation agent should lean on EV/EBIT, FCF yield and earnings, not P/B); (2) there is meaningful, untested impairment risk — a sustained GenAI-driven revenue or margin shock to a major cash-generating unit (US Federal, given DOGE pressure; or Western/Southern Europe, the weakest segment) could trigger a goodwill write-down that, while non-cash, would erase a large slice of reported equity. CGI has historically taken no material goodwill impairments (FACT/OPEN QUESTION — confirm via 40-F note), which both speaks to disciplined deal pricing and means the cushion has never been stress-tested in a structural-demand-decline scenario.

Off-balance-sheet and other items are not material distortions: lease liabilities are $693.5M (IFRS-16, already on balance sheet); the net defined-benefit pension obligation is small relative to the company ($556M gross DBO, largely funded; FACT, ex-99.2) and not a thesis risk.

6.5 Dilution/SBC, working capital, and FX — the small print that matters

Share-based compensation is modest and well-covered. The FY2025 SBC expense add-back in the cash flow statement was $68.6M (~0.4% of revenue), and the P&L SBC-related expense ~$68.6M — trivially small for a 94,000-person professional-services firm and a fraction of FCF (FACT, ex-99.2). CGI’s “Member”/employee-ownership culture runs largely through actual share purchase and a profit-participation/RSU plan rather than heavy option grants; option overhang is immaterial. Critically, net dilution is deeply negative: the ~$1.27B FY2025 NCIB (see Capital Allocation) swamps SBC many times over, so the share count falls ~3–6%/yr (FACT). Disclosure nuance worth flagging: the cash-flow line PaymentsToAcquireOrRedeemEntitysShares is only $13.3M (FY2025) — that is the RSU/treasury settlement, not the buyback; the ~$1,274.5M NCIB sits on a separate “purchase for cancellation of Class A shares and related tax” financing line. An analyst reading the XBRL alone would badly understate repurchases; reconcile to the MD&A capital-allocation narrative, not the single XBRL tag.

Working capital is a mild drag, not a red flag. DSO rose to 45 days (FY2025) from 41 (FY2024) before recovering to 40 days at Q2-F2026 (FACT). The FY2025 deterioration partly explains flat OCF on rising revenue; the Q2 recovery is reassuring. There is no evidence of channel-stuffing-style receivables build; for a managed-services/government-heavy book, 40–45-day DSO is healthy. CGI does not run a material client-funds/payroll-float business (unlike an ADP), so that lever is absent.

FX is the largest reported-growth distortion and the Valuation agent must normalize for it. FY2025 revenue grew +8.4% reported but only +4.6% constant-currency — i.e., ~3.8 points (≈$590M) of “growth” was CAD weakness against the USD/EUR/GBP, a translation effect with no economic substance (FACT, FY2025 PR). The pattern worsens at the margin: Q2-F2026 revenue grew +3.3% reported but only +1.6% constant-currency, with organic constant-currency growth effectively flat-to-low-single-digit once acquisitions are stripped (FACT). This is the quantitative spine of the central tension: CGI’s underlying organic engine is running near zero; reported growth is FX + M&A. FX also flatters the multi-year revenue CAGR.

6.6 Verdict — do economics improve with scale; cash machine or accounting roll-up?

It is a genuine cash machine that has stopped scaling — and an accounting roll-up whose returns the roll-up is now diluting. Both halves of the question resolve, and they pull in opposite directions:

• Cash machine: YES, and real. ~$1.96B FCF, FCF/net-earnings of 1.18×, OCF/net-earnings of 1.35×, ~1.7% capital intensity, ~1.1× net leverage, 13.6% ROIC and ~17% ROE all comfortably above cost of capital. Cash exceeds accrual earnings — the antithesis of an earnings-management profile. This is a high-quality business by any conventional financial screen.
• Economics improving with scale: NO — they have plateaued. Adjusted EBIT margin has been pinned at ~16–16.5% for three-plus years with ~25 bps/yr of historical drift; ROIC has fallen 240 bps in a year; OCF has been flat at $2.1–2.3B for four years. There is no operating-leverage flywheel here — the model does not get structurally more profitable as it grows; it gets bigger and returns capital.
• Accounting roll-up: YES, structurally — and FY2025 is the warning bar. Goodwill now exceeds equity, tangible book is ~−$2.35B, “one-time” integration add-backs recur and are scaling with deal volume, and the FY2025 deal wave demonstrably diluted ROIC. The EPS growth is overwhelmingly financial-engineering (buyback + accretive M&A), not organic economics. That is sustainable only so long as CGI keeps finding acquisitions at returns above its (falling) cost of capital and the share count keeps shrinking — both of which are getting harder, not easier.

The honest synthesis for the IC: a high-quality, cash-rich, conservatively-financed compounder whose reported per-share growth materially overstates its true organic earning power, and whose return on incremental capital is now visibly fading as the roll-up matures. The numbers do not support either the pure “quality compounder” bull narrative or a “fragile accounting fiction” bear narrative — they support a de-rating-to-quality read: a good business whose growth algorithm is more financial than operational and is reaching its mathematical limits.

7. Capital Allocation

Capital allocation is where CGI’s “Build-and-Buy” model lives or dies. The thesis the business sells — disciplined metro-market roll-up, rapid integration to a single operating model, double-digit cash returns on acquired capital, and a relentless buyback that shrinks the share count — is a capital-allocation thesis first and a services thesis second. We test each leg against the filings rather than the narrative.

7.1 The Build-and-Buy M&A engine — real returns, but a model that needs deals to grow

The model (FACT). CGI’s stated framework is to grow through a “double lever”: organic expansion plus acquisitions that (i) deepen metro-market scale or add a capability, (ii) are integrated rapidly onto CGI’s single operating/financial system, and (iii) target a “double-digit cash return on investment” with earnings accretion within roughly two years (FY2025 MD&A, 40-F Ex-99.3, Dec 17 2025). Two acquisition archetypes recur: small/mid “tuck-in” metro deals funded from free cash flow, and periodic “transformational” deals funded with debt.

The deal record (FACT, with a transparency caveat). Recent and landmark transactions:

Sources: PRNewswire/CGI.com deal releases (2010–2025); Q2 FY2026 MD&A §2.3, Apr 29 2026; Fasken/Cravath deal notes (Stanley); CGI Logica completion release Aug 20 2012.

Two corrections to the working facts (FACT). (1) BJSS closed in 2025, not 2023 — the SPA was signed Jan 29 2025 (CGI/PRNewswire). (2) The “Dec-2025 senior-notes” line is unconfirmed in the primary filings: the Q2 FY2026 MD&A references only the March 2025 $923.9M note issuance and the April 28 2026 facility increase. Long-term debt nonetheless rose from $3.40B (Dec 2024) to $4.29B at Dec 31 2025 “mainly driven by the issuance of senior unsecured notes” (Q1 FY2026 release, Jan 2026), so a FQ1-FY2026 note draw clearly occurred — we treat the amount/tranches of any specific December tranche as an OPEN QUESTION pending the F-10/SUPPL prospectus.

Is the M&A value-accretive, or growth-by-acquisition masking weak organic? (INTERPRETATION — the central capital-allocation question.) The accretion is real but increasingly carries the growth rather than supplementing it. FY2025 revenue rose +8.4% reported but only ~+5% constant-currency, and Q2 FY2026 was +3.3% reported / ~+1% organic (Q2 MD&A) — i.e., in the most recent quarter the majority of growth was acquired, not organic. That is the Marathon serial-acquirer fingerprint: when organic growth fades toward zero, the deal cadence must rise to keep reported growth and EPS moving. The corroborating balance-sheet tell is goodwill ≈ $9.5B ≈ 100% of equity and ~57% of assets, leaving tangible book near zero/negative — a business whose accounting equity is almost entirely purchased.

The return test (FACT/INTERPRETATION). Two offsetting readings. In CGI’s favor: consolidated ROIC was ~16% (FY2024) and ROE ~19%, both comfortably above any reasonable WACC (low-beta name, ~0.61 beta), and GAAP diluted EPS compounded ~11.5%/yr FY2017→FY2024 vs only ~4.4%/yr revenue — the gap delivered by margin expansion and the buyback, which is the model working. Against it: CGI’s own reported ROIC fell to 13.1% in Q2 FY2026 from 15.4% a year earlier (Q2 MD&A capital-management table) — exactly what you’d expect as a wave of goodwill-heavy 2025 deals enters invested capital before the synergies fully land. So the model still clears the cost of capital, but the marginal deal is diluting returns at the margin, and the disclosure does not let outside investors verify deal-level IRRs because CGI does not disclose purchase prices or EV/EBITDA multiples on most deals. That opacity is itself a governance/discipline flag (see the governance discussion below).

Greenwald lens: the durable edge here is economies of scale + a repeatable integration playbook (a process/cost advantage), not a structural moat in the acquired assets. The deals buy local-market scale and staff; the value-add is integration and margin lift, which is real but replicable by Accenture/Capgemini bidding for the same assets — so the “discipline on price” claim matters enormously and is unverifiable from outside.

Verdict (M&A): A genuine, above-WACC capital-deployment machine with a credible 15-year integration record — but one that is now dependent on M&A to grow at all, deploying into goodwill at a marginal-ROIC that is falling, with price discipline taken on faith because deal multiples are undisclosed.

7.2 Buybacks — the primary return vehicle, executed at scale and (recently) at low prices

The mechanism (FACT). Buyback, not dividend, has been the dominant cash-return tool. The diluted share count fell from 303.3M (FY2017) to ~213M (Q2 FY2026) — roughly −30%, ~−3.8%/yr — a direct, mechanical contributor to per-share compounding. Recent spend:

• FY2025: purchased 8,861,543 Class A shares for cancellation for $1,258.5M (FY2025 MD&A).
• H1 FY2026: 8,086,327 Class A shares for $958.8M; of which Q2 FY2026 alone was 3,511,574 shares for $391.9M (Q2 MD&A §2.2.2).
• New NCIB (FACT): authorized Jan 27 2026, effective Feb 6 2026 — up to 18,975,360 Class A shares = 10% of public float (Q2 MD&A). At the Q2 run-rate this is ~$2B of authorization.

Were they bought well? (INTERPRETATION — mostly yes, and the timing is improving.) Multi-year buybacks executed while the stock compounded were “fair-price” repurchases. The recent ones look better than fair: the H1 FY2026 / Q2 FY2026 purchases landed during a −42% trailing-year drawdown into the cheapest valuation in CGI’s own ~10-year history (~11× P/E, 2nd percentile; company filings/AZI). Buying ~$1B of stock at a 2nd-percentile multiple is the textbook-correct countercyclical use of the buyback. The honest caveat: management is simultaneously re-levering for M&A (see the debt-and-leverage discussion below), so the buyback is being partly debt-financed at these prices — defensible given the low multiple and ~1× leverage, but it is leverage-funded shareholder return, not purely FCF-funded.

Verdict (buyback): The single strongest piece of the capital-allocation case — large, consistent, share-count-shrinking, and recently deployed counter-cyclically at trough multiples.

7.3 Dividend — a token-but-growing newcomer, not the story

FACT. CGI ran a zero-dividend / all-buyback model for decades. It initiated a dividend in FY2025 — FY2025 paid $0.15/share = $135.1M (vs nil in FY2024; FY2025 MD&A). On Nov 4 2025 the Board raised the quarterly dividend to $0.17 (+13%), an “eligible dividend,” paid on both Class A and Class B shares; the Apr 28 2026 declaration continues $0.17 (Q2 MD&A §2.2.4). Yield is only ~0.5–0.8% and payout is <10% of earnings — deliberately small, a signaling/return-broadening tool layered on top of the buyback, not a competitor to it.

Total capital returned vs FCF (FACT/INTERPRETATION). FY2025 OCF was $2,234M; buyback ($1,258.5M) + dividend ($135.1M) ≈ $1,394M, ~62% of OCF, with the remainder plus new debt funding M&A. The balancing act is explicit: in light-deal years cash flows to buybacks; in deal years (H1 FY2026: ~$959M buyback plus re-levering) the firm leans on the balance sheet to do both. That is rational so long as leverage stays modest and the multiple stays low.

7.4 Debt & leverage — conservative, now re-levering deliberately for M&A

FACT. CGI runs a conservative balance sheet: net debt $3.57B, net-debt-to-cap 26.3%, ~1× EBITDA at Q2 FY2026 (Q2 MD&A) — though net-debt-to-cap rose from 24.1% a year earlier as deals were funded. Long-term debt + leases climbed to ~$4.29B at Dec 31 2025 from $3.40B (note issuance). On Apr 28 2026 the revolving credit facility was increased to $2,500M (3-yr $1,000M maturing 2029 + 5-yr $1,500M maturing 2031), giving the firm >$2.2B of available capital for Build-and-Buy (Q2 MD&A; FY2025 MD&A). CGI confirmed covenant compliance throughout. INTERPRETATION: this is a firm deliberately arming the M&A balance sheet at the bottom of its valuation — additive capacity for deals or buybacks — while keeping leverage well inside investment-grade norms. The risk is not the level (~1×) but the use: dry powder is only accretive if deployed at disciplined multiples we cannot independently verify.

7.5 Governance & incentives — founder voting control, modest economics, and minority entrenchment risk

This is the most important — and most adverse — part of the capital-allocation picture.

Dual-class control (FACT, computed from the AIF, Dec 9 2025). As at Dec 9 2025: 192,650,278 Class A subordinate-voting shares (1 vote; NYSE GIB / TSX GIB.A) and 24,122,758 Class B multiple-voting shares (10 votes each), held by the founders/insiders (Serge Godin, Founder & Executive Chairman; co-founder André Imbeau). The arithmetic:

• Class B = ~11.1% of the economics (24.1M of 216.8M total shares) …
• … but ~55.6% of the votes (241.2M of 433.9M total votes).

So the founder bloc holds an outright majority of the vote on ~11% of the equity — a ~5:1 wedge between control and economic interest. Minority Class A holders cannot win a contested vote, cannot force a board change, and have no take-out protection we could confirm (coattail/sunset provisions not located in the AIF — OPEN QUESTION).

Leadership — note the working-facts error (FACT). George Schindler is no longer CEO: he retired Sept 30 2024. François Boulanger was President & CEO from Oct 1 2024, and on May 12 2026 Tim Hurlebaus succeeded Boulanger as CEO (CGI/PRNewswire; StockTitan 6-K). Serge Godin remains Founder & Executive Chairman — the constant through three CEOs, which is the point: under a controlled structure the CEO is an operator, while capital-allocation control sits with the founder-chair. CEO-specific incentive metrics live in the management circular, a separate document not included in the 40-F and not in the local corpus — so the precise comp structure and metrics are an OPEN QUESTION for the proxy.

The two-sided read (INTERPRETATION). Pro-minority: founder skin-in-the-game has historically aligned CGI with long-horizon, return-on-capital behavior — the buyback discipline, the conservative balance sheet, and the absence of empire-building dilution all bear a founder-owner fingerprint, and the share count has fallen 30% rather than the serial-issuance pattern of worse roll-ups. Anti-minority: the same structure entrenches insiders, removes the market for corporate control, leaves minorities without a take-out backstop, and pairs with non-disclosure of deal prices — a combination that asks outside holders to trust, not verify, capital-allocation discipline. There is no evidence of related-party self-dealing in the reviewed filings, but the absence of a check is the risk.

Verdict (Capital Allocation): Capital has been allocated intelligently by outcome — above-WACC ROIC/ROE, a 30% share-count reduction via large, recently counter-cyclical buybacks, a conservative balance sheet armed for opportunistic M&A, and a 15-year integration record. But the verdict is qualified, not clean: growth is now M&A-dependent with organic growth near zero, marginal ROIC is falling as goodwill stacks, deal multiples are undisclosed, and a founder bloc controls ~56% of votes on ~11% of economics with no confirmable minority protection. The results are good; the governance check on future results is weak — minority holders are protected by founder track record and alignment, not by structure.

8. Changes and Headwinds — Last Two Years

The last two years have been a study in CGI doing what it has always done — buy growth, buy back stock, and protect margin — while the ground underneath the business model shifts. The changes are almost entirely management-controlled and incremental (M&A cadence, a modest re-levering, a maiden-then-raised dividend, a segment reshuffle, total board/management continuity). The headwinds are external, structural, and largely outside management’s control (a GenAI-driven de-rating of the entire IT-services group, U.S. federal “DOGE” austerity, near-zero organic growth, FX). The tension between a steady-as-she-goes operator and a market that has stopped believing the model is the whole story of the period.

8.1 M&A cadence — the build-and-buy roll-up keeps rolling

CGI’s “build-and-buy” engine ran continuously. Recent closings: APSIDE (Apside-Advance SAS, France) on Aug 28, 2025; Online Business Systems (OBS, Canada/U.S.) on Dec 2, 2025; Comarch Polska SA (Poland) on Dec 22, 2025, with BJSS (UK, closed Feb 25, 2025; ~2,400 consultants, ~£300m revenue) the larger deal in the same recent wave [FACT — Q2 FY2026 MD&A, Apr 29 2026, §2.3 and §1.2; CGI press release, 2025-02-25]. The tell on scale: CGI itself discloses that APSIDE + OBS + Comarch combined contributed only ~2.6% of Q2 FY2026 revenue and ~3.1% of total assets [FACT — Q2 FY2026 MD&A §“Internal Control” scope-limitation note]. These are metro-market bolt-ons — OBS brought ~350 professionals, Comarch ~460 — not transformational deals on the order of Logica (2012) or Stanley (2010). INTERPRETATION: the cadence confirms the model is intact and disciplined, but it also confirms the model’s dependence: at this deal size, CGI must keep buying simply to offset organic stagnation (the headwinds discussed below). The Greenwald/Marathon lens matters here — a serial acquirer whose tangible book is ~zero (goodwill ~$9.5B ≈ 100% of equity, FY2024 40-F) is converting balance-sheet capacity into revenue, and the capital cycle only rewards that if multiples paid stay low and integration stays clean. OPEN QUESTION: exact multiples paid (EV/EBITDA, EV/Sales) on APSIDE/OBS/Comarch are not disclosed — a gap the Capital Allocation section must flag.

8.2 Re-levering for build-and-buy — notes and a $2.5B facility

Two financing moves expanded firepower. First, in March 2025 CGI issued US$650M of 4.950% senior unsecured notes due 2030 (~CAD$923.9M of proceeds) [FACT — Q2 FY2026 MD&A §4.1; Dec 18 2025 F-10/SUPPL prospectus]. Important correction to a common mis-reading: the Dec 18, 2025 F-10/prospectus-supplement was an A/B exchange offer to register those same March-2025 notes under the U.S. Securities Act — not a fresh cash raise [FACT — SUPPL filed 2025-12-18, “Offer to exchange all outstanding 4.950% Notes due 2030 issued on March 14, 2025”]. Second — and this is a subsequent event to Q2, dated April 28, 2026 — CGI increased its unsecured committed revolving facility to $2,500M (a $1,000M three-year tranche maturing 2029 plus a $1,500M five-year tranche maturing 2031), “to provide additional financial agility for future capital deployment” [FACT — Q2 FY2026 MD&A §4.2/subsequent events]. INTERPRETATION: this is deliberate dry-powder accumulation for the build-and-buy pipeline, executed from a position of strength — net debt $3.57B, net-debt-to-cap 26%, leverage ~1× EBITDA. The one near-term wrinkle: the 2021 U.S. Senior Notes of $837.3M mature in September 2026, which the MD&A itself flags as the reason Q2 working capital was negative $286.8M [FACT — MD&A §4]. Refinancing a sub-$1B maturity for a ~$2B-FCF generator is trivial, but it does mean leverage will not fall organically near-term.

8.3 Dividend initiation and the +13% raise; segment realignment; continuity

CGI completed its evolution from a zero-dividend, all-buyback capital-return profile: on Nov 4, 2025 the board raised the quarterly dividend 13% to $0.17/share (from $0.15) and reaffirmed it Apr 28, 2026 [FACT — FY2025 press release, Nov 5 2025; Q2 FY2026 PR]. The payout is still tiny — ~$73.2M paid in H1 FY2026, a yield of ~0.5–0.8% — so buyback + M&A remain the dominant uses of cash. Separately, effective Oct 1, 2025 CGI realigned its management structure, moving Luxembourg out of Western & Southern Europe into the renamed “Scandinavia, Northwest and Central-East Europe” segment, restating comparatives; the company now reports nine geographic operating segments [FACT — Q2 FY2026 MD&A §“Reporting Segments”]. Control continuity is total at the top, but the CEO seat has turned over twice: Serge Godin remains Founder & Executive Chairman (the constant), while the CEO role passed by internal promotion from George Schindler → François Boulanger (Oct 1, 2024) → Tim Hurlebaus (May 12, 2026), with founders retaining voting control (see the Risk Analysis and Capital Allocation sections). INTERPRETATION: none of these are strategic pivots — they are housekeeping and a confidence signal (a dividend raise into a falling stock). The signal cuts both ways: continuity is reassuring on execution but offers nothing new to a market worried the model is the problem.

8.4 The headwinds — where the thesis actually gets tested

• GenAI disruption fear (the dominant headwind). The market’s central worry is that GenAI deflates the labor-arbitrage / time-and-materials / staff-augmentation core of IT services and lets clients in-source [INTERPRETATION; external: IDC, Everest Group, Forrester 2025–26 — the ~$1.6T services model “will not carry the next one”]. CGI’s answer is DigiOps (AI-powered delivery claiming “sustainable double-digit efficiency improvements,” with OpenAI/Google Cloud partnerships) [FACT — MD&A §1]. This is management’s hypothesis, not proof: efficiency that accrues to clients as price cuts is a margin/revenue risk, not obviously a moat. The bull/bear hinges on whether DigiOps monetizes productivity or merely passes it through (discussed in Variant Perception below).
• The June 18, 2026 sector selloff (read-across, not company-specific). Accenture reported FQ3 FY2026, cut its FY2026 local-currency revenue-growth outlook to 3–4% (top end down from 5%), missed revenue (~$18.7B vs ~$18.78B consensus), saw bookings fall ~2% YoY, and fell ~16–20% — its worst day on record [FACT — Investing.com / TechTimes / Yahoo Finance, Jun 18 2026]. The selloff swept the group (Capgemini ~−8%, Indian IT up to −7–10%, IBM lower), and CGI fell ~7.1–7.3% in sympathy (GIB.A C$93.28→C$86.66) despite not reporting that day [FACT — Globe and Mail / FactorsToday, Jun 18 2026]. Accenture explicitly cited weak demand and a federal-business slowdown — the same theme below.
• U.S. federal / DOGE austerity (a measured, real hit). CGI’s MD&A repeatedly attributes weakness to “U.S. federal government efficiency initiatives and shutdown.” Concretely: U.S. Federal revenue fell 11.1% reported / 7.1% constant-currency in Q2 FY2026 (−11.8% / −9.7% H1), with adjusted EBIT margin slipping to 12.8% (H1) [FACT — MD&A §3.4.5/§3.7.5]. About $444M of the quarter’s U.S. Federal revenue was federal-civilian-based — the most DOGE-exposed slice. Government is a very large end-market across Canada, U.S. Federal, and Europe, so austerity (and parallel UK government pressure) compounds.
• Thin organic growth. Constant-currency growth decelerated hard: 7.0% (Q3’25) → 5.5% → 3.4% → 1.6% (Q2’26) [FACT — MD&A “Selected Measures” table]. Canada shrank (−0.2% cc in Q2) [FACT — MD&A §3.4.4]. Stripping acquisitions, organic growth is near zero — the structural core of the bear case.
• FX. With ~80% of revenue non-CAD, translation swings both ways; in H1 FY2026 a stronger CAD turned reported gains into smaller constant-currency ones, and reduced long-term debt optically to $4.30B [FACT — MD&A §3.4, §4].

8.5 Verdict — net weakening, but slowly, and self-inflicted only at the edges

On balance these developments modestly weaken the thesis, and the weakness is concentrated in the headwinds, not the changes. The management-controlled moves (disciplined bolt-ons, conservative re-levering into a $2.5B facility, a confidence-signaling dividend raise, total leadership continuity) are consistent, shareholder-friendly, and arguably strengthen the operating story at the margin. But they are dwarfed by two external forces that strike at the model’s two growth levers simultaneously: GenAI undermines the labor-arbitrage core, and government austerity (US federal/DOGE, UK) compresses the single largest end-market — while organic growth has already decelerated to ~1.6% cc and Canada is shrinking. The Accenture read-across is sentiment, not fundamentals, but it crystallized a real question the numbers cannot yet answer: can build-and-buy keep manufacturing EPS growth if the “build” half is structurally near-zero and the “buy” half depends on cheap deals? The changes say “steady operator.” The headwinds say “the model is on trial.” The thesis is not broken — backlog is ~$31.5B (~2× revenue), FCF ~$2B, leverage ~1× — but the burden of proof has shifted to management to show GenAI is a tailwind, not a deflationary tax.

9. Risk Analysis

CGI is a high-quality, cash-generative operator, but the risk profile has shifted from idiosyncratic (integration, key-person) toward structural-industry risk over the last two years. The two risks that can actually impair intrinsic value — not merely the multiple — are (1) GenAI deflating the labor-arbitrage/T&M delivery core and (2) sustained government austerity in CGI’s largest end-market. The remainder are mostly manageable (FX, integration, cyclicality) or governance overhangs that cap the multiple rather than destroy cash flow. The matrix below scores likelihood and impact over a 2–3 year horizon; the discussion then weighs each top risk and disconfirming evidence.

9.1 Risk Matrix

9.2 The two value-impairing risks

GenAI / technology disruption of the delivery model (H / H — the master risk). CGI’s economics rest substantially on billing skilled labor — strategic consulting, systems integration, and especially the large managed-IT/BPO base — much of it on time-and-materials or staff-augmentation terms that GenAI directly threatens. Industry analysts (IDC, Everest, Forrester) argue the ~$1.6T “scale-headcount, drive-utilization, price-by-T&M” model is being re-based by agentic AI; the June 18, 2026 Accenture guidance cut (FY26 to 3–4%, bookings −2% YoY) and the group-wide selloff are the first hard datapoint that the worry is leaking into prints, not just narratives. The disconfirming case: CGI’s DigiOps claims “double-digit efficiency improvements” it can keep rather than fully pass through, ~$31.5B backlog provides 2-year visibility, and vendor consolidation could route more work to scaled incumbents. But efficiency that lands as client price cuts is a deflationary tax on revenue — and CGI’s own organic deceleration to 1.6% cc is consistent with that tax already biting. This is the single risk most likely to impair intrinsic value, because it attacks both volume (in-sourcing) and price (rate deflation) at once.

Government austerity / customer concentration (H / M). Government is CGI’s largest aggregate end-market (Canada, U.S. Federal, Europe). The damage is already quantified, not hypothetical: U.S. Federal revenue fell 11.1% reported / 7.1% constant-currency in Q2 FY2026 (−11.8%/−9.7% H1), with the MD&A naming “U.S. federal government efficiency initiatives and shutdown” (i.e., DOGE) and UK government pressure compounding [FACT — MD&A §3.4.5]. Impact is scored M not H because U.S. Federal is one of nine segments (~$2.35B TTM bookings) and government work is also CGI’s most defensible, sticky, mission-critical revenue — so the hit is real but bounded, and could reverse as new projects ramp (the MD&A notes “ramp up of new projects” partly offsetting).

9.3 The manageable and overhang risks

Acquisition integration & goodwill impairment (M / H). With goodwill ~$9.5B ≈ 100% of equity and tangible book near zero, CGI’s balance sheet is its acquisition history. CGI has a long clean integration record and has never booked a goodwill impairment — but that record is untested through a genuine demand downturn, and a GenAI-driven structural impairment to acquired staff-aug businesses is exactly the scenario that would force a write-down. Likelihood is only M (CGI’s discipline and rapid CGI-operating-model integration are real), but impact is H given the goodwill load. Organic stagnation / M&A reliance (H / M) is the flip side: the model needs deals to grow, so any drying-up of cheap bolt-ons (or richer multiples) directly throttles EPS growth.

Governance / dual-class entrenchment (M / M). Founders hold ~56.3% of votes on ~11.4% of the economics (Class B, 10 votes/share) — minority shareholders cannot influence strategy, comp, or succession [FACT — MD&A capital-stock table, Apr 24 2026]. This has coincided with disciplined capital allocation historically (alignment via heavy insider ownership), so it is currently an overhang on the multiple rather than a realized harm — but key-person/succession risk around Godin — and a CEO seat that has changed hands twice in under two years (Schindler→Boulanger→Hurlebaus) — is real and untested.

FX (H / L), cyclicality (M / M), competitive pricing (M / M), wage (M / M), refinancing (L / L) round out the profile. FX is high-probability but non-cash and symmetric. Cyclicality is buffered by the managed-services/IP base and ~2× backlog. Refinancing the Sep-2026 $837.3M maturity is trivial against ~$2B FCF and a $2.5B facility.

9.4 Which risks actually impair value

Only two risks can durably impair intrinsic value rather than sentiment: GenAI delivery-model disruption (most likely, because it taxes both price and volume and would also be the trigger for goodwill impairment) and sustained government austerity (real, already in the numbers, but bounded). Everything else either caps the multiple (dual-class governance) or is operationally manageable (FX, integration, refinancing, cyclicality). The honest read: CGI’s downside is not balance-sheet fragility or a single bad quarter — it is the slow possibility that the labor-arbitrage business it has spent five decades compounding is being structurally re-priced, with build-and-buy no longer able to outrun it.

10. Valuation Discussion

All figures CAD unless stated; FX explicit. NYSE GIB price = USD; TSX GIB.A = CAD. Sources: NYSE GIB close $61.28 USD (2026-06-18, AZI price CSV); CGI Q2-F2026 press release/MD&A (filed 2026-04-29, 6-K); FY2025 press release (2025-11-05) and 40-F (filed 2025-12-17); AZI valuation_index (own-history percentiles, 2026-06-18); peer multiples per stockanalysis.com / Yahoo Finance / company filings (accessed 2026-06-19, cited inline). Every CGI figure is reconciled to a primary filing; the EV build is shown line-by-line so the reader can audit it. This section frames the price as embedded expectations — no price target, no BUY/SELL.

10.1 Where it trades — building the enterprise value by hand

CGI’s dual listing forces a currency reconciliation before any multiple is meaningful. The NYSE line (GIB, USD) closed at $61.28 on 2026-06-18, down −7.3% on the day in the Accenture-guidance-cut sympathy selloff (see the Changes and Headwinds section). The TSX line (GIB.A) is the same economic share priced in CAD. To build a clean enterprise value I work entirely in CAD, the reporting currency, and convert the market cap at the prevailing rate (USD/CAD ≈ 1.37 implied by the AZI snapshot; the rate has been volatile in the 1.37–1.41 band, an OPEN QUESTION/ASSUMPTION that moves the USD-denominated multiples by ±2–3%).

Share count (FACT, Q2-F2026 PR). Total shares are the sum of both classes: ~192.65M Class A (subordinate voting) + ~24.12M Class B (multiple voting) ≈ 216.8M total economic shares (per the Dec-2025 AIF; see the governance discussion). The Q2-F2026 diluted weighted-average count was 213.1M — the smaller figure reflects the in-period buyback. For market cap I use the full ~216.8M class-combined share base; both classes carry identical economic and dividend rights, so both belong in equity value. At $61.28 USD that is ~$13.3B USD market cap; the AZI snapshot of ~$13.8B USD (~$18.9B CAD) uses a slightly higher share count and is the figure I carry, noting the ~4% spread as share-count/FX noise.

The EV build (CAD, FACT where filing-sourced; computed where noted):

Net debt is CGI’s own non-GAAP figure: CAD 4,302.9M long-term debt + leases, minus $708.4M cash, $7.6M short-term and $24.5M long-term investments, and an FX-derivative adjustment, = CAD 3,573.4M (FACT, Q2-F2026 PR, line-item reconciled). That is ~1.1× EBITDA and 26.3% net-debt-to-cap — a lightly levered balance sheet (see Financial Quality), so EV sits only ~19% above equity value.

10.2 The multiple set — cheap on every lens, reconciled

Using TTM figures (CGI’s Q2-F2026 trailing-twelve-month disclosures where available):

¹ The “~7.4×” adjusted P/E is a CAD-on-CAD figure and must not be mixed with the USD price. On a consistent basis: GIB at $61.28 USD ≈ CAD ~$84/share (×1.37). Against FY2025 adjusted EPS of $8.30 CAD, that is ~10.1× adjusted P/E; against a Q2-F2026 TTM adjusted EPS of ~$8.45 CAD (an assumption, see Financial Quality) it is ~9.9×. A “~7.4×” figure arises only if one divides the USD price by the CAD adjusted EPS — a currency mismatch. The honest read is ~10× adjusted P/E / ~11× GAAP TTM P/E, CAD-consistent (FACT/INTERPRETATION). This is worth flagging because it is exactly the kind of FX error that makes a cheap stock look even cheaper than it is. Either way CGI is genuinely inexpensive — but ~10× adjusted, not ~7×.

The takeaway: ~11× GAAP / ~10× adjusted P/E, ~7× EV/EBITDA, ~8× EV/adj-EBIT, ~1.2× sales, and a ~10% CAD FCF yield for a 13.6%-ROIC, ~$2B-FCF business that has compounded GAAP EPS at ~11.5%/yr for eight years. On its face this is a high-quality compounder priced like a no-growth value name.

10.3 FCF yield — compute it carefully, because a casual “6–7%” understates it

A casual read flags a ~6–7% FCF yield; a careful reconciliation lands higher and the gap is instructive. FY2025 FCF (OCF $2,234.2M − PP&E $116.6M − intangibles $153.3M) ≈ $1,964M CAD (FACT, see the cash-conversion discussion). Against a ~$18,900M CAD market cap that is a ~10.4% FCF yield in CAD terms; in USD, ~$1.43B FCF / ~$13.8B cap ≈ ~10.4% (currency-neutral, as it should be). The only ways to get to 6–7% are (a) dividing by EV rather than equity (FCF/EV ≈ 1,964/22,470 ≈ 8.7%), or (b) netting out the ~$1.27B buyback first, which conflates a use of FCF with FCF itself. The defensible number is a ~9% FCF/EV yield and a ~10% FCF/equity yield (FACT/INTERPRETATION) — a high free-cash yield that is the strongest single argument the stock is statistically cheap. The caveat the bear must press: a slice of that “FCF” is supported by deferred-revenue float and D&A on acquired intangibles, and the model reinvests the cash into goodwill at a falling marginal ROIC — so a high FCF yield does not by itself prove the cash compounds at the historical rate.

10.4 Own-history context — cheapest multiple in a decade (own-history only, not cross-sectional)

The AZI valuation_index is unambiguous and is the single highest-signal valuation datum on this name: P/E, P/B, and P/S all sit at the 2nd percentile of CGI’s own ~10-year range (composite 0.02) — i.e., the cheapest multiple, on all three lenses simultaneously, in essentially CGI’s entire post-Logica history (FACT, AZI, 2026-06-18: P/E 11.1×, P/B 1.81×, P/S 1.16×). For a business whose fundamentals (margin, ROIC, FCF conversion) have deteriorated only modestly — adj. EBIT margin still ~16.5%, ROIC down 240 bps but still 13.6% — a move to the 2nd valuation percentile is a far larger de-rating than the fundamental erosion justifies on the historical relationship. That is the quantitative core of the “abandoned quality” framing (the factor-positioning read: negative momentum loading −0.32, −42% trailing year, all-time-low price).

Three hard caveats on the percentile (binding): (1) It is own-history only — a 2nd-percentile P/E tells you the stock is cheap versus its own past, not that it is cheap versus a fairly-valued peer set; if the whole IT-services category is structurally re-rating down on GenAI (see Industry Dynamics), CGI’s own history is the wrong yardstick and the “cheap” signal is a value trap. (2) The P/B percentile is near-meaningless here — book is ~100%+ goodwill and tangible book is ~−$2.35B (see Financial Quality), so 1.81× P/B is a number about purchase accounting, not asset value; weight P/E, EV/EBIT and FCF yield instead. (3) A 2nd-percentile multiple in a sector bear market is not a contrarian “buy signal” by itself — Accenture (see the peer comps below) is at its own ~18th percentile and Cognizant at a 52-week low; the entire group is de-rating together. The percentile is context, never a target.

10.5 Peer comps — where CGI sits and why it is the cheapest of the Western roll-ups

The relevant comp set is Western IT-services consolidators and the Indian offshore majors. The June-18 Accenture print reset the anchor: ACN cut FY2026 revenue-growth guidance from 3–5% to 3–4% and fell up to −20% (its worst day on record), dragging the group (FACT, Accenture FQ3-FY2026 PR, 2026-06-18, Investing.com).

¹ ACN was ~11.9× fwd P/E pre-June-18 (per a separate Accenture analysis dated 2026-06-11); the −20% drop lowered the multiple while the guide cut lowered the E, leaving it broadly ~11–12× (INTERPRETATION). Peer multiples are third-party aggregator snapshots (Yahoo/stockanalysis/TIKR/Alpha Spread, 2026-06-19), not filing-reconciled — directional comps, not precise.

Where CGI sits and why (INTERPRETATION). CGI trades at the low end of the Western pack on P/E (~10×) and mid-pack on EV/EBITDA (~7×) — cheaper than Accenture, Capgemini and the Indian majors, roughly in line with Cognizant (the other “cheapest of the West”) and the distressed-adjacent DXC. It deserves some discount, and the discount is explainable, not anomalous:

• Lower organic growth. CGI’s ~0–2% organic (see the Growth section) is below ACN’s 3–4%, Capgemini’s low-single-digit, and well below the Indian majors’ mid-single-digit. A roll-up that needs M&A to grow should trade below a peer that grows organically.
• Serial-acquirer / goodwill quality. Tangible book ~−$2.35B, falling marginal ROIC, undisclosed deal multiples (see Capital Allocation) — the market rightly assigns a “trust-but-can’t-verify” discount versus cleaner balance sheets (Indian majors are net-cash).
• Canadian small/mid-cap + dual-class + IFRS. CGI is ~$14B USD, founder-controlled (~56% of votes on ~11% of economics, see Capital Allocation), IFRS-in-CAD, and NYSE-listed as an FPI — a structurally narrower investor base, a governance discount, and reporting friction that all compress the multiple versus a $250B+ S&P-100 Accenture.
• Defensive offset. Against those discounts, CGI’s ~110% historical book-to-bill, $31.5B backlog (1.9× revenue), 0.61 beta, and counter-cyclical buyback argue for a narrower discount than DXC/Atos. CGI is the highest-quality of the cheap Western names, not the cheapest because it is the worst.

The honest synthesis: CGI’s ~10× P/E is a defensible discount to ACN/Capgemini for lower organic growth, M&A-dependence, and governance — but it is no longer a small discount; it is a near-trough, Cognizant-level multiple on a structurally better-converting, founder-aligned book.

10.6 Embedded expectations / reverse-DCF — what is the price underwriting?

Strip the multiple back to the cash math. At a ~$18,900M CAD market cap with ~$1,964M FY2025 FCF and ~213M diluted shares, the equity FCF yield is ~10.4%. A simple perpetuity-growth decomposition (cost of equity ≈ FCF yield + growth): if the market demanded, say, a ~9% equity return on a stable business, a ~10.4% starting FCF yield implies the market is pricing roughly −1% to +1% real per-share FCF growth in perpetuity (INTERPRETATION; ASSUMPTION on cost of equity — a 0.61-beta name arguably warrants <9%, which would imply the price embeds even less growth, i.e. the stock is cheaper still on that lens).

Put differently, a reverse-DCF at a ~9% discount rate that solves for the per-share FCF growth justifying today’s price lands near ~2–3% nominal / ~0% real growth — below CGI’s own historical ~4–5% revenue / ~11% EPS algorithm. The market is underwriting that CGI’s per-share compounding effectively stops — that buybacks + accretive M&A no longer drive double-digit EPS growth, consistent with the GenAI-deflation / organic-stall bear case (see Industry Dynamics and Growth). The embedded-expectations question for the IC is therefore sharp: is near-zero real growth the right central case, or is the market extrapolating a sector-wide GenAI scare onto a book with $31.5B of contracted backlog and a still-running buyback? The price says “the engine is broken”; the backlog and FCF say “the engine is idling, not broken.” That tension is the whole investment debate.

10.7 Scenario analysis (bear / base / bull) — scenario math, NOT a target

The following is illustrative scenario math to bracket the range of outcomes embedded in the debate — explicitly not a price target. I hold the framework constant (revenue growth → adj-EBIT margin → buyback-driven share-count reduction → exit multiple on FY2028E adjusted EPS) and vary the assumptions. All in CAD; convert to USD at ~1.37. FY2025 adj. EPS = $8.30 CAD base.

Reading the scenarios (INTERPRETATION; every figure is an explicit ASSUMPTION above).

• Bear (~$51–58 USD) roughly brackets today’s $61.28 — i.e., the current price already discounts a scenario where organic growth turns negative, margins compress 150 bps, the buyback slows, and the multiple stays trough. The stock is priced close to the bear case, not the base.
• Base (~$84–91 USD) holds the current ~+1.5% organic / ~16.5% margin / ~−3.5% share-count algorithm and re-rates only to CGI’s own mid-cycle multiple (~11–12×) — it does not assume a heroic re-acceleration, just that the model keeps working at its recent run-rate. This is meaningfully above today’s price.
• Bull (~$124–135 USD) requires the GenAI fear to invert into a positive — DigiOps efficiency monetized into margin, vendor-consolidation share gains lifting organic to ~4%, and a quality re-rating to ~14–15× — i.e., CGI re-rates toward Accenture/Infosys quality multiples. Plausible but it needs the central GenAI debate to resolve in CGI’s favor.

The spread (~$51 to ~$135 USD) is wide because the single swing variable — whether GenAI deflates or augments the managed-services model — is genuinely unresolved (see Industry Dynamics and Variant Perception). The asymmetry the scenarios reveal: the downside from here is largely already in the price (bear ≈ spot), while the base case sits well above spot. That is the embedded-expectations observation; the judgment on whether to act on it belongs to Claude’s Take, not here.

10.8 What is the market pricing correctly vs. incorrectly?

Correctly (the bear’s valid points): (1) organic growth is genuinely ~0–2% and structurally challenged (see the Growth section) — paying an ACN/Infosys premium would be wrong; (2) the model is M&A-dependent with falling marginal ROIC and undisclosed deal multiples (see Capital Allocation) — a goodwill/discipline discount is warranted; (3) GenAI is a real threat to the staff-augmentation/T&M slice, and government austerity (US Federal/DOGE, UK) pressures a huge end-market (see Industry Dynamics); (4) dual-class entrenchment and Canadian-FPI friction justify a structural discount.

Potentially incorrectly (the bull’s valid points): (1) the move to a 2nd-percentile multiple is a far larger de-rating than the modest fundamental erosion supports on CGI’s own history; (2) the $31.5B backlog (1.9× revenue) and ~108% TTM book-to-bill give multi-year revenue visibility that the “engine is broken” price ignores; (3) the buyback is shrinking the share count ~3.5%/yr into the trough multiple — a mechanical per-share tailwind the no-growth price under-credits; (4) ~$2B FCF and ~1× leverage mean the model is self-funding and the dividend/buyback are secure. The market is extrapolating an Accenture-driven sector scare onto a book that did not report that day and whose contracted backlog argues against a cliff.

10.9 Valuation Verdict — expectations framing

At ~11× GAAP / ~10× adjusted P/E, ~7× EV/EBITDA and a ~10% FCF yield — its cheapest multiple in a decade on every lens — CGI is priced for the engine to stall. The embedded expectation is near-zero real per-share growth in perpetuity: the market is underwriting that GenAI deflation and organic stagnation neutralize the buyback-plus-accretive-M&A algorithm that drove ~11.5%/yr EPS for eight years. The scenario math shows today’s price brackets the bear outcome — negative organic growth, margin compression, a stalled buyback, and a permanently trough multiple — while CGI’s own current run-rate (≈+1.5% organic, flat ~16.5% margins, continued buyback) supports a value materially above spot if it merely re-rates to its own mid-cycle multiple.

The valuation does not resolve the thesis; it sharpens it. The discount to Accenture and Capgemini is defensible (lower organic growth, M&A-dependence, governance, small-cap/FPI friction) but is now near-trough and Cognizant-level on a better-converting, founder-aligned book. The market is pricing the bear case as the base case. Whether that is prescient (GenAI structurally impairs the labor-arbitrage model) or an over-extrapolated sector scare on a backlog-protected compounder is the single question the rest of the memo — and Claude’s Take — must answer. What valuation can say without a recommendation: the price embeds pessimism, the downside appears largely discounted, and the asymmetry tilts to whether the GenAI debate breaks for or against the managed-services model (see Variant Perception). No price target; this is the expectations map, not a call.

Price Action & Factor Positioning — Quantitative Overlay

This overlay sits between the Valuation and Variant Perception sections; it is the quantitative read of the tape and CGI’s factor loadings that feeds both. An overlay subordinate to the thesis — no price target, no buy/sell.

Source: FactorsToday risk/factor model (ElasticNet betas, 756-day window) and AZI price CSV, accessed 2026-06-19. These are third-party statistical estimates, not primary data; loadings and realized returns are FACTs, any “will continue / will revert” reading is INTERPRETATION and regime-caveated. No price call is made here.

The track record is unambiguously bad, and the pain is recent. On FactorsToday’s leaderboard, GIB’s trailing one-year return is −42.5% (matches the AZI CSV: $106.63 → $61.28, June 2025 → June 2026), with a −49.5% drawdown from the February-2025 peak — the maximum drawdown across every horizon out to lifetime, i.e., the stock is at its worst level on record (FACT). The shorter windows are reported annualized and must be de-annualized to avoid overstating them: the m6 figure of −55.6% annualized is a raw ~−33% over six months (CSV: −33.3%), and the m3 figure of −49.4% annualized is a raw ~−15% over three months (CSV: −14.1%). One-year Sharpe is −1.54 and Sortino −1.69 — deeply negative. The contrast with the long record is the whole tension: lifetime return is +12.0%/yr (Sharpe 0.37) and the rs_10_year linear trend is still positive (+0.85), so this is a structural compounder having its worst stretch in a quarter-century, not a perennial loser (FACT: leaderboard + stock-info, 2026-06-19).

The market prices CGI as a stable quality-software name, not a cyclical — which cuts both ways. FactorsToday classifies GIB under “Industry: Software” (All-Factors industry beta 0.23) and lists its factor-similar peers as SAP (0.94 similarity), Gartner (0.94), Bentley Systems (0.93), Tyler Technologies (0.91), OpenText (0.91), ExlService, Manhattan Associates — almost entirely high-quality, low-beta application-software and information-services names, with Capgemini (0.88) and Cognizant (0.87) the only true IT-services analogs in the top set (FACT: related-stocks, 2026-06-19). This matters for the variant-perception debate: the model “sees” CGI as a sticky, recurring-revenue software-adjacent compounder rather than a labor-arbitrage staff-augmentation cyclical. The bull reading is that the moat (backlog ~1.9× revenue, managed-services contracts, IP solutions) genuinely makes it more software-like than its Indian-IT peers. The bear reading is that the same peer basket — SAP, Gartner, Bentley, Tyler — is exactly where the GenAI “AI is eating consulting/software” derating has been most violent (Gartner −30% in a single day on Feb 3, 2026 on a zero-growth 2026 guide), so being grouped with them is no refuge.

Low beta, negative alpha, idiosyncratic pain. Market beta is ~0.61 (All-Factors model; FACT) — genuinely defensive, consistent with the name historically drifting with the market rather than leading it. But alpha is −0.28 and the stock carries ~24%/yr idiosyncratic (specific) volatility (R² ~0.33, so two-thirds of the variance is stock-specific), meaning the recent collapse is not explained by market exposure — a 0.61-beta stock should not be down 42% in a year when its beta-implied move is a fraction of that. The drawdown is being driven by CGI-and-sector-specific repricing (organic-growth deceleration, GenAI disruption fears, government-austerity exposure), which the specific-return series confirms: the largest single-day idiosyncratic moves are −10.2% (Apr 29, 2026, the soft Q2 print) and −6.5% (Feb 3 and Feb 11, 2026, the GenAI-derating window) (FACT: specific-vol series + loadings, 2026-06-19).

Regime read: abandoned momentum, not a momentum darling. The most telling loading is Momentum (12-1m) at −0.32 in the All-Factors model (negative and the single largest style beta) — GIB is now a negative-momentum stock, i.e., the model’s trend signal places it firmly in the out-of-favor, falling-trend cohort, the opposite of a crowded long. It screens cheap on every value lens (AZI own-history: P/E 11.1×, P/B 1.81×, P/S 1.16×, all at the ~2nd percentile of a decade), yet the negative momentum loading says the value is not yet being rewarded — the classic falling-knife signature where a quality name has de-rated faster than its fundamentals have deteriorated, but the tape has not turned.

The framing this hands the Lead. This is empirically a low-beta quality compounder mid-air in a sector-driven drawdown, not a high-flying momentum trade unwinding from euphoria — the negative momentum loading, the all-time-low price, the value-percentile extremity, and the idiosyncratic (not market) nature of the decline together describe an abandoned quality setup. Whether it is “abandoned quality at a generational price” or a “value trap whose fundamentals are still eroding” is not resolvable from the factor data — the factors confirm the stock is hated and out-of-favor, but cannot tell us whether organic growth re-accelerates or GenAI permanently deflates the model. The disconfirming evidence for the bull case lives in the fundamentals (constant-currency organic growth now ~+1.6% and the federal/government overhang), not the tape; the tape’s only message is that consensus has thrown the stock into the negative-momentum, cheapest-decile bin and is waiting for a fundamental catalyst (INTERPRETATION, regime-caveated; no price call).

11. Variant Perception

Variant perception requires three things: an honest statement of what the market currently believes, the strongest version of each opposing case, and — most importantly — the handful of assumptions whose resolution actually settles the debate, each paired with the specific evidence that would prove it wrong. CGI is an unusually clean test case because the bull and bear arguments are not about different facts but about the durability of the same facts: both sides agree the organic engine has stalled, that goodwill dominates the balance sheet, and that GenAI is reshaping the labor model. They disagree on whether those are terminal conditions or a trough. The factor read is unambiguous that consensus has thrown the stock into the cheapest-decile, negative-momentum bin; it cannot tell us whether consensus is right.

11.1 The consensus belief

Consensus today treats CGI as a slow-growth IT-services roll-up that is on the wrong side of the GenAI transition, and is therefore dead money — a value trap that deserves to be cheap. The evidence consensus points to is real, not imagined (FACT): the stock is −42.5% over the trailing year (FactorsToday; AZI price CSV), trades at the 2nd percentile of its own ~10-year valuation history on P/E (~11×), P/B (~1.8×) and P/S (~1.2×) simultaneously (AZI valuation_index, 2026-06-18), and just took a −7.3% sympathy hit when Accenture cut FY2026 revenue-growth guidance to 3–4% and fell ~20% on June 18 2026 (company filings). The sell-side framing — applied to the whole group — is that AI is structurally compressing the billable-hour model, government austerity (US Federal/DOGE) is gutting the most defensive end-market, and a serial acquirer with goodwill ≈ 100% of equity and tangible book near zero is a financially-engineered EPS story with little underneath. INTERPRETATION: the market is not pricing a collapse (the low beta ~0.61 and 2× backlog argue against that); it is pricing permanent low-single-digit-or-worse organic growth with no terminal re-rating — a business that compounds EPS in the high single digits via buybacks but never gets paid a quality multiple again. That is the “value trap / dead money” thesis, and it is the default reading the −42% tape encodes.

11.2 The strongest bull case

The bull case is that CGI is a disciplined cash machine being mispriced as a structurally-impaired one, and that the very features consensus reads as weaknesses are the source of the return. Five legs:

1. Cheapest-ever multiple on a still-compounding business. GAAP diluted EPS compounded ~11.5%/yr FY2017→FY2024 (3.41→7.31) and FY2025 adjusted EPS hit $8.30 (company filings) — yet the multiple is at its all-time-low 2nd percentile. The bull does not need re-acceleration; he needs the multiple to stop falling while ~$2B/yr of FCF compounds per-share value. That is the classic abandoned-quality setup the factor read describes.
2. GenAI as monetizable productivity, not just deflation. In fixed-fee managed services (~half the book) and outcome-based outsourcing, AI efficiency accrues to CGI’s margin, not the client’s bill — DigiOps claims “sustainable double-digit efficiency improvements,” with OpenAI/Google Cloud partnerships (Q2 FY2026 MD&A). As enterprises consolidate vendors around scaled, trusted integrators for higher-stakes AI work, CGI gains share by vendor consolidation even in a flat market (the central GenAI debate in Industry Dynamics).
3. CAD ~$2.24B of available capital into a cheap private market. With ~$2.24B available, a $2,500M facility, ~1× leverage and ~$2B FCF (Q2 FY2026 MD&A §4), CGI can buy metro-market books at disciplined multiples while the entire sector is de-rated — counter-cyclical M&A at the bottom of the cycle (the M&A engine in Capital Allocation).
4. Relentless buyback compounding per-share value at a trough multiple. The diluted share count fell ~30% (303M→213M) over a decade; H1 FY2026 alone retired ~8.1M shares for ~$959M, and the new NCIB authorizes another ~19M (10% of float) — bought at a 2nd-percentile multiple, the textbook countercyclical repurchase (the buyback discussion in Capital Allocation).
5. Founder alignment. Serge Godin’s bloc has overseen 30% share-count reduction, a conservative balance sheet, and no empire-building dilution — an owner’s fingerprint, not a promoter’s (the governance discussion in Capital Allocation).

The bull’s synthesis: a 16% ROIC, ~$2B-FCF, owner-operated consolidator with 2× backlog visibility, bought at ~11× — where even zero organic growth plus the buyback and disciplined M&A delivers high-single-digit per-share compounding, and any re-rating is upside.

11.3 The strongest bear case

The bear case is that the cheap multiple is correct, and possibly still too high, because the model is structurally broken in three places:

1. Organic growth is structurally ~0; the model needs M&A to grow. Constant-currency organic ran +0.9% (FY2024) and ~+1% (Q2 FY2026), with Canada (−0.2%), U.S. Federal (−7.1%) and U.S. Commercial & State Gov (−3.9%) all shrinking organically (the Growth section). The reported ~4.9% CAGR is roughly half FX and acquisitions. A company that must buy its growth is one acquisition-pipeline-failure away from flat-to-down revenue.
2. GenAI deflates the un-moated core and enables in-sourcing. The deflation evidence is observed, not asserted: TCS cut >23,400 jobs in FY2026 citing AI “skill mismatch”; Indian IT added a net 17 employees across nine months; ACN cut guidance and saw bookings −2% (the GenAI debate in Industry Dynamics). To the extent CGI’s revenue is T&M/staff-augmentation, AI compresses the billable base and lets clients build in-house — exactly the half of the book that was never moated (the Competitive Position section).
3. Government austerity hits the biggest end-market. Government is CGI’s largest single vertical, and it is under simultaneous fiscal pressure across the US (DOGE: ~$5.1B contracts cancelled in 2025), UK, and EU — turning the historical stabilizer into the present source of the stall (government demand, Industry Dynamics).

Layered on top: dual-class entrenchment (founders control ~56% of votes on ~11% of economics, no confirmable coattail/take-out protection — the governance discussion in Capital Allocation), a goodwill-heavy roll-up with tangible book near zero where marginal ROIC is already falling (15.4%→13.1% y/y, Q2 MD&A), and undisclosed deal multiples that ask investors to take price discipline on faith. The bear’s synthesis: a value trap that stays cheap because the cheapness is earned — buybacks shrink the share count of a business whose economic value-per-share is no longer growing organically, and the market is right to refuse it a quality multiple into a structural demand shock.

11.4 The assumptions that actually matter — and what falsifies each

The thesis turns on five testable assumptions. For each, the falsification test is specific and observable within ~4–6 quarters.

Assumption #4 is the master variable. Items #1 and #3 are the two observable halves of the single unanswerable question — is GenAI net-accretive or net-deflationary to CGI’s model? If organic growth and margin both hold or improve, AI is being monetized and the bull is vindicated; if either erodes materially, the deflation thesis is winning. Critically, the bear evidence (sector headcount cuts, ACN guidance, ~1% organic) is already observed, while the bull evidence (DigiOps double-digit efficiency) is asserted and unquantified — so the burden of proof sits with the bull, and the next two prints carry it.

11.5 What the tape is pricing — and where consensus may be offsides

The factor read sharpens the variant question rather than answering it. CGI screens as a low-beta (~0.61), negative-alpha (−0.28), negative-momentum (12-1m loading −0.32) abandoned quality name — not a high-beta cyclical falling knife. FactorsToday classifies it under Software with factor-similar peers SAP, Gartner, Bentley, Tyler, OpenText — i.e., the market models CGI as a sticky, recurring-revenue software-adjacent compounder, yet has de-rated it to the 2nd valuation percentile while two-thirds of its variance is idiosyncratic (R² ~0.33). This is the heart of the variant perception (INTERPRETATION): the tape encodes a quality business priced for structural impairment. If consensus is wrong, it is wrong in treating an idiosyncratic, sector-sympathy drawdown of a 16%-ROIC, 2×-backlog cash machine as if the impairment were already confirmed — the negative-momentum signal says the value “is not yet being rewarded,” the falling-knife signature. But the same peer basket (SAP, Gartner, Tyler) is exactly where the GenAI derating has been most violent, so being grouped with quality software is no refuge. The factors confirm the stock is hated and out-of-favor; they cannot confirm whether organic growth re-accelerates or GenAI permanently deflates the model. That resolution lives in assumptions #1–#4 above — in CGI’s next four-to-six prints, not in the tape.

Verdict on consensus: consensus is demonstrably correct on the diagnosis (organic stall, M&A-dependence, GenAI threat to the staff-aug core, government headwind — all evidenced) and unproven on the prognosis (that these are terminal rather than trough conditions). The variant-perception opportunity, if one exists, is entirely in the prognosis — and it is settled by the organic-growth-plus-margin trajectory, the single test the next two MD&As will run for us.

12. Fact vs. Interpretation Table

13. Open Questions

1. What multiples did CGI pay for APSIDE, Online Business Systems, Comarch Polska, and BJSS? Purchase prices/EV-EBITDA are not disclosed on most deals — the price discipline central to the thesis is taken on faith, and falling marginal ROIC is the only external read on it.
2. Is GenAI net-accretive or net-deflationary to CGI’s model? The master variable. Resolved only by the joint trajectory of organic growth and adjusted-EBIT margin over the next 4–6 quarters; today the deflation evidence (peers) is observed and the accretion evidence (DigiOps) is asserted.
3. Will organic growth stop shrinking in Canada and U.S. Federal? These two high-margin segments turning positive (or not) is the single load-bearing fundamental question.
4. CEO (Tim Hurlebaus) compensation structure and incentive metrics — these live in the management proxy circular, which is not part of the 40-F corpus; we cannot yet verify whether incentives are tied to per-share value, ROIC, or merely revenue/EPS scale.
5. Class A coattail / take-out protection terms — not located; relevant to minority-holder downside in any future control transaction given the dual-class structure.
6. Exact revenue-by-type split (managed services vs. SI&C vs. IP) — disclosed by CGI largely in charts rather than text; the ~50% managed-services / ~20% IP figures are close approximations, not precise reconciliations.
7. Goodwill-impairment risk — CGI has never taken an impairment; the ~$11.7B balance has never been tested through a genuine demand downturn, and a structural GenAI impairment to acquired staff-augmentation books is exactly the scenario that would force one.
8. The exact size/tranching of the late-2025 note issuance (long-term debt rose from $3.40B to $4.29B over calendar 2025 “mainly driven by senior unsecured notes”) — pending the F-10/prospectus-supplement detail; the March-2025 US$650M 4.95% notes are confirmed.

14. What Must Be True

For the bull case (abandoned quality at a generational price) to be right:

• CGI’s managed-services + IP base and DigiOps must let it monetize GenAI efficiency as margin rather than passing it to clients as price cuts; organic growth must trough near current levels and stabilize (not keep falling); and disciplined M&A + the buyback must keep compounding per-share value while the multiple re-rates toward CGI’s own mid-cycle (~11–12×).
• Falsification test: the bull is wrong if, over the next 2–3 MD&As, constant-currency organic growth stays ~0 or negative in Canada and U.S. Federal, or adjusted-EBIT margin falls more than ~100 bps without a one-off cause, or group book-to-bill breaks below 100%, or reported ROIC drifts toward ~12%/WACC. Any of those says GenAI is taxing price and/or volume and that build-and-buy can no longer outrun the erosion.

For the bear case (value trap; structurally impaired roll-up) to be right:

• GenAI must structurally deflate the labor-arbitrage/T&M core and enable clients to in-source; organic growth must stay ~0/negative; government austerity must persist; and M&A must be unable to outrun the erosion at disciplined prices — leaving the multiple permanently trough while goodwill-heavy “EPS growth” masks a business whose per-share economic value no longer compounds.
• Falsification test: the bear is wrong if organic constant-currency growth re-accelerates toward mid-single-digits with the margin intact (GenAI proving net-accretive), or book-to-bill recovers above ~105% with the sub-100% segments turning, or ROIC re-rises toward 15–16% as the 2025 deals synergize, or CGI announces a clearly cheap, accretive acquisition that re-rates the market’s view of the M&A pipeline.

The decisive evidence for both sides is the same: CGI’s organic-growth-plus-margin trajectory over its next four-to-six prints. The tape cannot resolve it; the income statement will.

15. Source Appendix

See Appendix B — Source Appendix (in the combined report) for the full, dated source list. Primary sources relied upon:

• CGI Inc. FY2025 Form 40-F (filed 2025-12-17, SEC EDGAR CIK 0001061574) and FY2021–FY2024 40-Fs — annual financials, segment data, AIF, risk factors, governance.
• CGI Q1 FY2026 and Q2 FY2026 6-K exhibits — MD&A, press releases, condensed financial statements (filed 2026-01-28 and 2026-04-29); FY2025 Q4 press release/MD&A (2025-11-05).
• SEC EDGAR XBRL company facts (CIK 0001061574, ifrs-full taxonomy) — revenue, profit, EPS, equity, goodwill, borrowings, cash flows, weighted-average shares (FY2017–FY2024).
• AZI valuation-index feed (2026-06-18) — own-history P/E/P/B/P/S percentiles; AZI 5-year price CSV (NYSE GIB) — price, EMAs, beta, alpha, volume.
• FactorsToday risk/factor model (2026-06-19) — stock loadings, leaderboard, stock-info, related-stocks, specific-vol.
• Public news/trade press (June 2026) — Accenture FQ3 guidance cut and the IT-services selloff (Investing.com, TechTimes, CNBC, Globe and Mail); CGI BJSS-acquisition and CEO-transition releases (CGI.com / PRNewswire); Gartner/IDC/Everest IT-services market data.
• Comparable-company analyses — Accenture (2026-06-11), IBM (2026-06-10), EPAM (2026-06-06), Booz Allen (2026-06-14) — industry structure, GenAI swing-variable framing, peer comps.

APPENDIX A — Standard Diligence Questionnaire

CGI Inc. (NYSE: GIB / TSX: GIB.A). Supplemental to the memo body — not counted toward memo length. All figures CAD unless stated; FX explicit. Sources: CGI FY2025 40-F (filed 2025-12-17, EDGAR Ex-99.2/99.3), FY2025 press release (2025-11-05), Q2 FY2026 6-K/MD&A/press release (filed 2026-04-29), FY2025 AIF, and EDGAR ifrs-full XBRL (FY2017–FY2024). Labels: FACT / INTERPRETATION / ASSUMPTION / OPEN QUESTION. Management commentary is treated as hypothesis and validated against filings.

General — What thoughtful questions have other investors asked about CGI?

Five recurring questions dominate the buy-side debate, each engaged in the memo body:

• Is organic growth structurally dead? Constant-currency growth decelerated 7.0% → 5.5% → 3.4% → 1.6% (Q2 FY2026), with Canada (−0.2% cc) and U.S. Federal (−7.1% cc) shrinking organically [FACT — Q2 FY2026 MD&A]. Once FX and acquisitions are stripped, the organic engine runs ~0–2%. The question is whether this is cyclical (DOGE/macro) or structural (GenAI). (see the Growth section)
• Does GenAI deflate the labor-arbitrage / time-and-materials core, or does CGI monetize it via DigiOps? This is the master question — it drove the entire IT-services group’s de-rating, crystallized by Accenture’s June 18, 2026 guide cut. (see Competitive Position, Risk Analysis and Variant Perception)
• How long is the M&A runway, and is it disciplined? CGI needs deals to grow; investors ask whether bolt-on supply at low multiples persists and whether the ~$2.2B firepower will be deployed without overpaying — a concern sharpened because CGI does not disclose purchase multiples on most deals. (see Capital Allocation)
• Does the dual-class structure entrench founders at minority expense? Founders hold ~56% of votes on ~11% of economics [FACT — AIF, Dec 9 2025]. The track record is shareholder-friendly, but the structural check is absent. (see Capital Allocation)
• How exposed is CGI to government austerity? Government (Canada + U.S. Federal + Europe) is the largest aggregate end-market; investors probe the size of the DOGE/UK-austerity hit. (see Business Overview and Risk Analysis)

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? INTERPRETATION: Margins are near a cyclical/structural high; returns on capital are off their peak and falling. Adjusted EBIT margin (~16.4–16.5%) sits at the top of its decade band, and adjusted EPS reached a record $8.30 (FY2025). But ROIC has fallen for six straight quarters — 16.2% → 16.0% → 16.1% → 15.4% → 14.6% → 13.6% → 13.3% → 13.1% (Q2 FY2026 TTM) [FACT — Q2 FY2026 MD&A capital-management table] — so on return on capital the franchise is past peak. Organic growth is at a cyclical low (1.6% cc). Net: profitability high, growth and incremental returns low — a mid-to-late-cycle profile, not a trough.

Driven by external environment or internal actions? INTERPRETATION: Overwhelmingly internal. EPS growth came from margin discipline, a −30% share-count reduction (303.3M → 213M diluted, FY2017 → Q2 FY2026 [FACT]), and EPS-accretive M&A — not end-market tailwinds. The external environment (GenAI fear, DOGE austerity, FX) is currently a headwind that internal levers (buyback + M&A) are offsetting. This is the central quality-of-earnings point: per-share growth is a capital-structure story layered on a ~0–4% organic business.

How stable are revenues? FACT: High visibility via backlog and recurring mix. Backlog $31.50B at Q2 FY2026 ≈ 1.9–2.0× annual revenue; ~$11.5B converts within twelve months; TTM book-to-bill 108.4% (decelerating from ~110–114%). Roughly half of revenue is multi-year (5–10 yr) recurring managed services. Caveat: backlog grows partly by acquisition (an explicit component of CGI’s definition), and three segments (UK/Australia, Germany, Finland/Poland/Baltics) now show sub-100% TTM book-to-bill — a leading indicator of future revenue contraction there [FACT — Q2 FY2026 MD&A]. Revenue is stable, but the discretionary SI&C half (~50%) cycles with IT budgets.

Market size — growing/shrinking, domestic/international? FACT/INTERPRETATION: The global IT & business-consulting services market is large (multi-trillion CAD) and still growing low-to-mid-single-digits, but the growth rate and revenue model are under structural threat from GenAI (the active debate). It is overwhelmingly international: only ~13% of revenue is Canadian (home market); the rest is U.S. (~30% across Federal/Commercial/State), Western & Southern Europe (~17%), UK/Nordics/CEE. ~80% of revenue is non-CAD [FACT]. So CGI is a Canadian-domiciled, globally-diversified operator, not a domestic play.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? INTERPRETATION: More. GenAI lowers barriers to routine build/maintain work, enables client in-sourcing, and compresses T&M pricing; the offshore Indian majors (TCS/Infosys ~20–25% margins) remain structurally cheaper. Vendor consolidation could favor scaled incumbents, but the base case is intensifying price competition in the commoditized SI&C half.

Profitability (ROIC/ROE)? FACT: ROE ~16.8% (FY2025), down from ~19.1% (FY2024); ROIC 13.6% (FY2025) / 13.1% (Q2 FY2026 TTM), down from 16.0% (FY2023–24). Both clear an ~8–9% WACC (beta ~0.61), so returns are genuinely above cost of capital — but both are declining, and the ROE > ROIC gap reflects the goodwill load (invested capital is ~57% goodwill). INTERPRETATION: a real but narrowing excess return — the FY2025 deal wave was bought at returns below the existing book’s, diluting franchise ROIC (the Marathon late-cycle signature).

How profitable is the industry — competitors, barriers to entry? FACT/INTERPRETATION: Industry margins range widely — Indian majors ~20–25%, Accenture ~15.6% adj. op, Capgemini ~13%, Cognizant ~15%, DXC ~7%/distressed Atos negative. Barriers are moderate and mostly non-structural: scale (delivery platforms, brand), switching costs in embedded outsourcing/government IP, and security accreditations — but no hard entry barrier prevents a TCS or Capgemini from bidding for the same metro-market assets CGI buys. CGI is sub-scale (CAD $15.9B vs Accenture’s ~US$69.7B, ~6×) and has no supply-side cost advantage.

Can the business be easily understood? INTERPRETATION: Yes — sell advice, build systems, run them under multi-year contracts, plus a ~20–22% higher-margin proprietary-IP layer (Momentum federal ERP, CGI Advantage). The complexity is in capital allocation (serial M&A, undisclosed multiples, goodwill accounting), not the operating model.

Can it be undermined by foreign low-cost labor (and GenAI)? FACT/INTERPRETATION: Partly already is, on both fronts. CGI competes against (not with) the cheaper offshore base of TCS/Infosys/Wipro; its more onshore/proximity mix lifts intimacy but forfeits cost leadership. GenAI is the sharper threat — it deflates the billable-hours/staff-augmentation base and enables in-sourcing, attacking the un-moated SI&C half. The moated half (embedded managed services, government IP, regulated systems of record) is more defended.

Do brands matter? INTERPRETATION: Modestly. CGI’s brand carries weight in government/regulated procurement (Momentum’s 180+ federal organizations, security accreditations) and in metro-market relationships, but it lacks Accenture’s marquee brand pull on the largest global transformation deals. Brand is a relationship/credential asset here, not a consumer-style pricing moat.

Nature of competition? FACT: Compete-per-deal RFP/re-compete dynamics in SI&C (fully contestable); renewal/extension dynamics in managed services and government IP (incumbent-favored). Direct rivals: Accenture, Capgemini (closest build-and-buy analog), Cognizant, Infosys/TCS/Wipro, EPAM, IBM Consulting, DXC; US-gov: Booz Allen/Leidos/SAIC/CACI.

Customers’ switching costs? FACT/INTERPRETATION: High in the recurring/government/IP ~50–60% of the book (ripping out a decade-long outsourcing contract or a federal financial-management ERP is high-cost, high-risk, multi-year — the genuine Greenwald demand-side captivity), low-to-nil in the SI&C/staff-aug remainder (re-won each project). The moat is narrow and segment-specific, not firm-wide.

Financial Condition & Balance Sheet

Assets not fully recognized on the balance sheet (client relationships, IP, backlog)? FACT/INTERPRETATION: Yes — the $31.5B backlog and internally-developed IP (Momentum, CGI Advantage, DigiOps tooling) and long-standing client relationships are economically valuable but largely unbooked (internally-generated intangibles are expensed under IFRS; only acquired intangibles and goodwill sit on the balance sheet). Counterpoint: the balance sheet over-states one asset — goodwill of $11,744.8M (FY2025) now exceeds total equity of $10,282.3M, so tangible book is roughly −$2.35B [FACT, computed]. The unbooked operating assets are real; the booked goodwill is the unamortized purchase premium of 49 years of deals.

Off-balance-sheet liabilities (leases, pensions, client-funds)? FACT: Minimal and well-disclosed. (i) Leases are on balance sheet under IFRS-16 (lease liabilities ~$693.5M). (ii) The net defined-benefit pension obligation is small (~$556M gross DBO, largely funded) — not a thesis risk. (iii) Client-funds float exists but is modest: CGI held $788.8M of cash within “funds held for clients” at Q2 FY2026 [FACT — Q2 FY2026 MD&A] — a payments/funds-administration float (relevant in some government/payroll/health programs), with a matching client-obligations liability. This refines the Financial Quality section’s statement that CGI runs “no material client-funds business”; the float is real but small (~$0.8B) and not a major earnings lever like an ADP’s. No material take-or-pay or guarantee exposure surfaced in the filings [OPEN QUESTION — full contingent-liability note not separately verified here].

Accounting conservatism (GAAP vs adjusted add-backs, goodwill)? INTERPRETATION: Mixed, leaning aggressive on the non-GAAP framing, conservative on cash. The GAAP-vs-adjusted gap is the flag: CGI adds back “restructuring, acquisition and integration costs” of $96.9M (FY2024) → $285.0M (FY2025) pre-tax, lifting FY2025 adjusted EPS ($8.30) 13% above GAAP ($7.35). These “one-time” costs recur every year and scale with deal volume — for a perpetual acquirer they are an ordinary structural cost, so adjusted EPS over-states sustainable earnings power [INTERPRETATION]. Offsetting (conservative): cash exceeds accrual earnings (OCF/net earnings ~1.35×), CGI has never booked a goodwill impairment [FACT — though this means the goodwill cushion is untested in a structural-demand decline], and SBC is small (~$68.6M, ~0.4% of revenue). Net: read earnings closer to GAAP than to adjusted; the cash is real.

How CapEx-hungry is the business? FACT: Very light. FY2025 PP&E purchases $116.6M + intangible-asset purchases $153.3M ≈ ~$270M total capex (~1.7% of revenue). This is a people-and-IP business, not an asset-heavy one — the basis for ~$1.96B FCF (~12% FCF margin) and strong cash conversion.

Capital Allocation & Management

FCF generation, use, and philosophy? FACT: FCF ≈ $1.96B (FY2025) (OCF $2,234.2M − ~$270M capex). Philosophy is the explicit “Build and Buy”: (1) accretive M&A (metro-market bolt-ons + periodic transformational deals targeting “double-digit cash return on investment”), (2) a relentless share buyback, and (3) — newly — a small, growing dividend. FY2025 buyback ($1,258.5M) + dividend ($135.1M) ≈ $1.39B (~62% of OCF), with the balance plus new debt funding M&A. INTERPRETATION: disciplined, return-on-capital-oriented allocation — but increasingly M&A-dependent for growth and re-levering to fund both deals and buybacks.

Significant recent acquisitions? FACT: APSIDE (France digital/engineering, ~€250M turnover, Aug 28 2025); Online Business Systems (Canada/US IT & security, ~350 staff, Dec 2 2025); Comarch Polska (Poland, ~460 staff, Dec 22 2025) — all metro-market bolt-ons (combined only ~2.6% of Q2 FY2026 revenue / ~3.1% of assets [FACT — MD&A internal-control scope note]). Larger predecessor: BJSS (UK, closed 2025 not 2023, target rev ~£300.7M). Landmark transformational: Logica (2012, ~doubled CGI) and Stanley (2010, US Federal entry). OPEN QUESTION: purchase multiples (EV/EBITDA, EV/Sales) are not disclosed on most deals — price discipline is taken on faith.

Buying back shares? FACT: Yes, aggressively and counter-cyclically. FY2025 repurchased 8,861,543 Class A shares for $1,258.5M; H1 FY2026 8,086,327 shares for $958.8M (Q2 alone 3,511,574 for $391.9M). New NCIB (Feb 6 2026) authorizes up to 18,975,360 Class A shares (~10% of public float). Recent buybacks landed into a −42% drawdown at the cheapest valuation in CGI’s ~10-year history (~11× P/E, 2nd percentile) — textbook-correct timing, though partly debt-financed.

Issuing shares to insiders / SBC? FACT: No — net dilution is deeply negative. SBC is ~$68.6M/yr (~0.4% of revenue), swamped many times over by the ~$1.27B buyback. Employee ownership (87.5% of ~94,000 Partners own stock) runs largely through actual share purchase and a profit-participation plan, not heavy option grants. Disclosure nuance: the XBRL tag PaymentsToAcquireOrRedeemEntitysShares is only $13.3M (RSU/treasury settlement) — the ~$1.27B NCIB sits on a separate financing line; reconcile to the MD&A, not the single tag.

Director/management compensation? OPEN QUESTION: Precise comp structure and incentive metrics live in the management proxy circular, a separate document not included in the 40-F and not in the local corpus. Cannot verify CEO incentive metrics here. FACT: leadership turned over — George Schindler retired Sept 30 2024; François Boulanger President & CEO from Oct 1 2024; Tim Hurlebaus succeeded Boulanger May 12 2026; Serge Godin remains Founder & Executive Chairman (the constant through three CEOs — capital-allocation control sits with the founder-chair).

Management motivations (founder dual-class)? FACT/INTERPRETATION: Founder voting control: 192,650,278 Class A (1 vote) + 24,122,758 Class B (10 votes), so founders hold ~56% of votes on ~11% of economics [FACT — AIF, Dec 9 2025]. Pro-minority read: founder skin-in-the-game has historically driven long-horizon, return-on-capital behavior (the 30% share-count reduction, conservative leverage, no empire-building dilution all bear a founder-owner fingerprint). Anti-minority read: the structure entrenches insiders, removes the market for corporate control, and pairs with undisclosed deal prices — minorities must trust, not verify. No evidence of related-party self-dealing surfaced; the risk is the absence of a structural check. Coattail/take-out protection for Class A holders not located [OPEN QUESTION].

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? FACT: None of those. CGI is a Canadian foreign private issuer with a dual-listed ordinary share — Class A Subordinate Voting Shares trade directly on the NYSE (GIB, in USD) and the TSX (GIB.A, in CAD); it is not an ADR (no depositary receipt), not an MLP, and issues no K-1 (it is a corporation). It files 40-F + 6-K (IFRS in CAD), not 10-K/10-Q. Tax note for U.S. holders: dividends are designated “eligible dividends” for Canadian tax purposes and may carry Canadian non-resident withholding tax (typically reduced to 15% under the Canada–US treaty) — a relevant consideration for U.S. holders.

Dividend policy? FACT: Recently initiated, small, growing, paid on both share classes. CGI ran a zero-dividend / all-buyback model for decades; it initiated a dividend in FY2025 ($0.15/sh; $135.1M paid). On Nov 4 2025 the quarterly dividend was raised +13% to $0.17/share (paid on both Class A and Class B; the Q2 dividend of $0.17 totaled $36.2M for the quarter / $73.2M H1 [FACT — Q2 FY2026 MD&A]; designated an “eligible dividend”). Yield is only ~0.5–0.8%; payout <10% of earnings. The dividend is a signaling/return-broadening overlay, not a competitor to the buyback.

How profitable is the business? FACT: Net margin 10.4% (FY2025; ~11.5% in FY2022–24 before the FY2025 restructuring charge), adjusted EBIT margin ~16.4%, ROE ~16.8%, ROIC 13.6% — high-quality absolute profitability, but with a declining return-on-capital trend (see Business Quality above).

Is net income diverging from cash from operations? FACT: Yes — favorably. OCF/net earnings ≈ 1.35× and FCF/net earnings ≈ 1.18× (FY2025), i.e., cash exceeds accrual earnings — the opposite of an earnings-management red flag, driven by D&A on acquired intangibles, managed-services deferred-revenue float, and SBC add-back. The one yellow flag is direction: OCF has been roughly flat at $2.1–2.3B for four years (mature, not accelerating), and FY2025 OCF was flat on +8.4% revenue partly from a working-capital drag (DSO rose to 45 days from 41, recovering to 40 at Q2 FY2026).

Risks & Downside

What factors would cause the stock to decline? INTERPRETATION: (1) GenAI deflating the labor-arbitrage / T&M core — the master risk; taxes both volume (client in-sourcing) and price (rate deflation) and is the most likely trigger for an intrinsic-value impairment (and a goodwill write-down). (2) Sustained government austerity (U.S. Federal already −11.1% reported / −7.1% cc in Q2 FY2026 on DOGE/shutdown; UK pressure) compressing the largest aggregate end-market. (3) Organic-growth stagnation forcing reliance on ever-larger M&A at richer multiples. (4) Further sector de-rating / read-across (the June 18 2026 Accenture guide-cut sympathy selloff). (5) A failed or over-priced large acquisition, or multiple expansion failing to materialize despite the 2nd-percentile valuation (value-trap risk).

Risk of a catastrophic loss? INTERPRETATION: Low. No single client > ~14% of revenue (U.S. Federal, the largest); ~$31.5B diversified, recurring, multi-year backlog; conservative ~1.1× net leverage; ~$2B FCF; ample liquidity ($2.5B facility, >$2.2B available). The realistic catastrophic scenario is a goodwill impairment if a GenAI-driven structural-demand shock hits a major cash-generating unit (US Federal or Western/Southern Europe) — but that is non-cash and would not threaten solvency.

Chance of a total loss? INTERPRETATION: Negligible over any normal horizon. CGI is a profitable, cash-generative, conservatively-financed, investment-grade-profile business with positive equity (despite negative tangible book), diversified revenue, and strong interest coverage (>25×). Total-loss scenarios require a multi-year, simultaneous collapse of demand across all geographies and a financing-market freeze — not a base or even bear case. The realistic downside is de-rating and slow value-erosion, not impairment of the going concern.

Recent News & Events

Has the business environment changed recently? FACT/INTERPRETATION: Yes — materially, on sentiment; modestly, on fundamentals. The dominant change is the GenAI-driven de-rating of the entire IT-services group, crystallized on June 18, 2026 when Accenture cut its FY2026 revenue-growth guidance from 5% (top end) to 3–4%, saw bookings −2% YoY, and fell ~16–20% (its worst day on record); the group sold off (Capgemini ~−8%, Indian IT −7–10%) and CGI fell ~7.3% in sympathy despite not reporting that day [FACT — Jun 18 2026, multiple sources]. Fundamentally, U.S. Federal/DOGE austerity is already in CGI’s numbers (Q2 FY2026 U.S. Federal −11.1% reported / −7.1% cc), and organic growth decelerated to 1.6% cc.

Significant acquisitions? FACT: APSIDE (Aug 2025), Online Business Systems (Dec 2025), Comarch Polska (Dec 2025) — metro-market bolt-ons; BJSS (UK, 2025) the larger predecessor. (Detailed above under Capital Allocation.)

Change in accounting policies? FACT: No material accounting-policy change. The relevant structural change is non-accounting: effective Oct 1 2025 CGI realigned its operating segments (moved Luxembourg into the renamed “Scandinavia, Northwest and Central-East Europe” segment; restated comparatives; now nine geographic segments) [FACT — Q2 FY2026 MD&A]. CGI continues to report under IFRS as issued by the IASB.

Recent changes — new markets, facilities, management? FACT: (i) Management: Schindler retired (Sept 30 2024) → Boulanger (Oct 1 2024) → Hurlebaus (May 12 2026) as CEO; Godin remains Founder & Executive Chairman. (ii) Capital structure: maiden dividend (FY2025) then +13% raise (Nov 2025); credit facility increased to $2,500M (Apr 28 2026) arming the M&A balance sheet; March-2025 US$650M note issuance (~$923.9M). (iii) New markets/capabilities: the 2025 deals deepened France (APSIDE), Canada/US security (OBS), and Poland delivery (Comarch). INTERPRETATION: these are continuity-and-housekeeping moves and dry-powder accumulation — a steady operator reinforcing the model, into a market that has begun to question whether the model itself is structurally challenged.

APPENDIX B — Source Appendix

CGI Inc. (NYSE: GIB / TSX: GIB.A) · Research as of June 19, 2026. Sources are primary-first. CGI is a Canadian foreign private issuer: it files Form 40-F (annual) and Form 6-K (interim) with the SEC and reports under IFRS in Canadian dollars; fiscal year ends September 30. All figures CAD unless stated; NYSE GIB price is USD, TSX GIB.A is CAD.

Primary — CGI filings (SEC EDGAR, CIK 0001061574)

• FY2025 Form 40-F (filed 2025-12-17; accession 0001193125-25-322911) — annual wrapper incorporating the FY2025 Annual Report, audited IFRS financial statements, Annual Information Form (AIF), and risk factors.
• FY2021–FY2024 Form 40-Fs (filed 2021-12-17, 2022-12-16, 2023-12-15, 2024-12-18) — multi-year financial series, segment disclosures, governance.
• Q2 FY2026 6-K (filed 2026-04-29; accession 0001061574-26-000012) — exhibits: Q2 FY2026 MD&A (cgi-fy26_q2xmda.htm), press release (cgi-fy26_q2xpressrelease.htm), condensed consolidated financial statements. Source for: revenue $4,156M, GAAP diluted EPS $2.09, adjusted diluted EPS $2.27, bookings $4.31B, book-to-bill 103.8%, backlog $31.50B, net debt $3,573M, net-debt-to-cap 26.3%, segment revenue/EBIT, ~94,000 staff, organic +1.6% cc, U.S. Federal −11.1% reported / −7.1% cc, nine-segment realignment, NCIB authorization, credit-facility increase to $2,500M (April 28, 2026 subsequent event), capital-stock/dual-class table.
• Q1 FY2026 6-K (filed 2026-01-28; accession 0001061574-26-000006) — MD&A/press release; long-term debt at Dec 31, 2025 (~$4.29B), note-issuance commentary.
• FY2025 Q4 / full-year 6-K (filed 2025-11-05; accession 0001061574-25-000006) — FY2025 revenue $15,912.7M (+8.4% / +4.6% cc), GAAP net $1,658.3M / diluted EPS $7.35, adjusted net $1,871.5M / adjusted diluted EPS $8.30, OCF $2,234.2M, bookings $17.57B, book-to-bill 110%, dividend +13% to $0.17 (declared Nov 4, 2025).
• F-10 / F-X / prospectus supplement (SUPPL) (filed 2025-12-18) — A/B exchange offer registering the US$650M 4.950% senior unsecured notes due 2030 issued March 14, 2025.
• SEC EDGAR XBRL company facts (CIK 0001061574, ifrs-full taxonomy) — Revenue, ProfitLoss, Equity, Assets, Goodwill, Borrowings, FinanceCosts, CashFlows (operating/investing/financing), DilutedEarningsLossPerShare, WeightedAverageShares / AdjustedWeightedAverageShares, FY2017–FY2024. Multi-year series reconciled to the press releases above.

Primary — CGI corporate communications

• CGI press release, “CGI completes acquisition of UK-based BJSS” (2025-02-25, CGI.com / PRNewswire) — BJSS closed Feb 25, 2025; ~2,400 consultants.
• CGI press release, “CGI enters into an agreement for the acquisition of BJSS” (2025-01-29) — SPA signed.
• CGI announcement, François Boulanger appointed President & CEO effective Oct 1, 2024 (Schindler retired Sep 30, 2024) — PRNewswire / Globe and Mail.
• CGI 6-K / StockTitan, Tim Hurlebaus appointed President & CEO (May 12, 2026; Boulanger retired) — leadership transition.
• CGI press releases / MD&A — APSIDE (closed Aug 28, 2025), Online Business Systems (control Dec 2, 2025), Comarch Polska (Dec 22, 2025); CGI DigiOps and OpenAI / Google Cloud partnership commentary.

Quantitative data feeds

• AZI valuation-index feed (azi.sh fundamentals GIB, accessed 2026-06-18) — own-history percentiles: P/E 11.1× (2nd pctile), P/B 1.81× (2nd), P/S 1.16× (2nd), composite 0.02; TTM EPS, BVPS, sales/share.
• AZI 5-year price CSV (download-data.php?t=GIB, accessed 2026-06-19) — NYSE GIB adjusted/unadjusted OHLCV, 21/50/200-day EMAs, beta (0.61), alpha (−0.28), dividend/split columns. Source for the Five-Year Event Map.
• FactorsToday risk/factor model (factorstoday.com/api, accessed 2026-06-19) — /stock-loadings/GIB (Momentum −0.32, Market 0.61, Industry: Software 0.23; R² 0.33), /leaderboard/GIB (1yr −42.5%, 6m −33% raw, lifetime +12.0%/yr, drawdowns/Sharpe/Sortino), /stock-info/GIB (RS, beta, market cap ~$13.8B USD), /related-stocks/GIB (SAP, Gartner, Bentley, Tyler, OpenText), /stock-specific-vol/GIB (~24%/yr idiosyncratic).
• AZI news feed (azi.sh news GIB, 2026-06-19) — near-empty (count 1), consistent with the foreign-issuer pattern; recent-events timeline built from filings + trade press instead.

Industry, peer & news sources

• Accenture FQ3 FY2026 results / guidance cut (2026-06-18) — FY2026 revenue-growth guide trimmed to 3–4%, bookings −2% YoY, ~−20% (worst day on record); IT-services group selloff (Cognizant −5.8%, Infosys ADR ~−10%, Capgemini −8%, CGI −7.3%). Sources: Investing.com, TechTimes, CNBC, Globe and Mail, Outlook Money, The Federal.
• Gartner / IDC / Everest Group / Forrester (2025–2026) — global IT-services TAM (>$1.87T, +13.5%), managed-services and consulting forecasts, GenAI-disruption commentary; TCS FY2026 ~23,400 job reductions; Indian-IT net hiring near zero.
• Third-party valuation aggregators (stockanalysis.com, Yahoo Finance, TIKR, Alpha Spread, 2026-06-19) — peer multiples for Accenture, Capgemini, Cognizant, Infosys, TCS, IBM, Booz Allen, DXC/Atos (directional comps, not filing-reconciled).
• Comparable-company analyses — Accenture (ACN, 2026-06-11), IBM (2026-06-10), EPAM (2026-06-06), Booz Allen (BAH, 2026-06-14), ADP (2026-06-12) — industry structure, GenAI swing-variable framing, comparable multiples and capital-return context.

Analytical frameworks

• Greenwald & Kahn, Competition Demystified — moat taxonomy (supply/cost; demand/captivity; economies-of-scale + captivity), market-share-stability and ROIC tests, applied to CGI’s competitive position.
• Marathon / Chancellor, Capital Returns — supply-side capital-cycle and serial-acquirer/asset-growth analysis, applied to the build-and-buy model and the sector’s GenAI-driven repricing.

Reconciliation note: third-party aggregated data (AZI, FactorsToday, valuation aggregators) are cross-checks, not primary; every material CGI figure in this report is reconciled to a CGI filing. Where a quick aggregator number and a filing disagreed (e.g., the USD-price-÷-CAD-EPS multiple mismatch, and ROIC/ROE that belonged to FY2024 rather than FY2025), the filing governs and the discrepancy is flagged in-text.