Solid demand hiding critical changes in commercial mix. Global demand looks supportive: digitalization, interoperability mandates and the migration of AI pilots into production keep spend flowing, but headline growth is modest and bifurcated. Global IT consulting and services grew roughly +3.6% in 2025, with a similar pace expected into 2026 – a low-single-digit market, not a boom. Within software, the split is sharp: smaller subscription cohorts are re-accelerating (the sub-USD50mn revenue tier bounced back to ~18% growth in 2025, near 2023 levels) while large-caps continue to moderate. Amid other early weak signals, retention business has drifted down to a median near 101%, versus ~105% earlier in the cycle, signaling that expansion within the installed base is thinning. The critical mix shift is commercial: the industry is being pushed from seat-based SaaS toward outcome- and consumption-priced "service-as-software," where the vendor is paid for work delivered rather than licenses sold – a leading strategy house already books roughly a quarter of fees on outcome-based contracts. This decouples revenue from headcount and seat counts alike, rewarding those who can quantify delivered value and penalizing those still selling access.
The "AI deflation" vector is revamping market competition. The large delivery-led services majors have themselves flagged that AI-native execution will dilute future revenue: rising adoption is intensifying pricing pressure, triggering deal renegotiations and slowing execution as clients reassess spend. Where value historically accrued to billable hours, agentic coding and automated delivery compress the fee pool directly. Compounding this, competitive walls are falling: hyperscalers are absorbing more of the platform layer, open-weight infrastructure lets new entrants scale agile and differentiated capital-light tools and frontier-model providers are moving directly into the advisory and delivery market. The defensible ground narrows to sector depth, proprietary data, regulated-industry mandates and integration complexity that a generic model cannot yet own. Companies with low compressible digital development and high staffing ratio work are the most exposed. The competitive question for 2026/2027 is less about who has the best AI than who can convert AI from a margin threat into a pricing-power moat before the fee pool reprices.
Profitability premium erosion under scrutiny. Core software still boasts exceptional gross margins – a median near 80% on software revenue (~76% blended) – and scaled players are extracting real operating leverage, with the USD500mn–USD1bn revenue cohort lifting EBITDA margins from ~17% to ~29% year-on-year through expense discipline. Services margins remain healthy but pressured, holding around 22–23% at the operating level, with providers warning that cushion narrows as talent costs rise and AI investment continues. The new drag is inference: AI-native products carry structurally lower gross margins because compute and inference costs are genuine, recurring COGS. The winners of 2026/2027 will be those who pass inference costs through pricing and cap usage, and improve unit economics as they scale – turning compute from a margin leak into a moat. The risk case is cost arbitrage on the client side: as inference becomes a visible IT budget line, CIOs may reallocate away from third-party software and services toward in-house or model-native alternatives, squeezing the sector's addressable spend even where its own margins hold.
AI is dramatically shortening product lifespan; premium is stored in security and sovereign solutions. GenAI compresses the build cycle, which cuts both ways: faster time-to-market for incumbents, but also faster obsolescence and lower switching costs, eroding the recurring-revenue stickiness the whole valuation model rests on. Two structural demand pockets stand out against this. First, cybersecurity, where the threat surface is expanding mechanically with model capability – AI-enabled attacks have risen sharply (on the order of +70% year-on-year by some breach-tracking measures) – and defensive spend is one of the few line items still growing while budgets consolidate elsewhere: enterprise security budgets rose ~5% into 2026, with AI-specific security now exceeding 11% of the total. Second, data-sovereignty and industrial-specific applications: distrust of frontier models' data handling is pushing regulated and industrial clients toward proprietary, on-premise or verticalized ecosystems – a demand stream that favors providers with domain depth and controlled deployment. Both pockets reward specialization over horizontal reach and are the likeliest sources of durable, defensible growth over the outlook window.
Investors need to be reconquered. Investor sentiment has deteriorated materially in 2026. Leading pure-play names fell on the order of 25–30% even as they reported solid results, a sell-off driven not by earnings but by a collapse in confidence in the business model itself. Under a "SaaS-Apocalypse" mood, public investors concluded they lacked clarity on how to price AI risk – the fear that frontier-model capabilities and falling entry barriers erode software's growth, stickiness and pricing power. Valuations de-rated to distressed levels: a bellwether integrator shed ~40% of its value across 2025 and now trades below the broader IT-sector earnings multiple (roughly 17x), stripped of its long-standing quality premium. Such sentiment deterioration and flight of capital out of software industry matters because capital is the sector's lifeblood for building out structure and funding future growth: a higher cost of equity, colder IPO and M&A windows and cautious sponsors hit the sub-scale, growth-funded tier hardest, where cheap capital was the model's oxygen. Crucially, solvency is not the concern – sentiment is. Asset-light economics, high free-cash-flow conversion and predominantly net-cash balance sheets let profitable incumbents absorb a valuation reset without liquidity stress. Expect a "rich-get-richer" bifurcation: cash-generative leaders consolidating and self-funding their next phase, while capital-starved, weak-retention challengers face an existential funding squeeze.
