AI was supposed to kill off consultants. It’s not happening, Capgemini’s strategy chief says

OpenAI and Anthropic are partnering with major consultancies (OpenAI’s Frontier Alliance with McKinsey, BCG, Capgemini, Accenture; Anthropic deals with Deloitte, Accenture, Cognizant) and each AI vendor has only about ~70 forward-deployed engineers, highlighting reliance on systems integrators. Capgemini is shifting to outcome-based contracts—charging for KPIs like issue resolution and NPS—and using AI to lower staffing costs to serve midmarket clients, while emphasizing domain expertise, cybersecurity, legacy integration, and governance. The main near-term risk is retraining staff to work with AI agents; consultants remain well-positioned as providers of industry knowledge and implementation services despite AI’s capabilities.

Analysis

Consulting firms shifting to ‘outcome-seller’ models create a two-part revenue upgrade: (1) larger, multi-year managed services contracts that convert lumpy project fees into annuity-like streams and (2) embedded hardware/software procurement clauses that steer client capex. Together these raise lifetime value of clients and increase stickiness, which can justify 200–400bps of operating margin expansion for top-tier integrators over 12–24 months if execution succeeds.

A concentrated infrastructure effect follows: systems integrators will standardize on a handful of inference stacks to lower integration cost and prove SLAs, accelerating enterprise GPU and datacenter orders in pulses tied to deployment waves (pilot → scale). That creates shorter, higher-amplitude demand cycles for component suppliers and gives procurement leverage to large integrators (volume commitments, preferred-vendor discounts) — winners will be suppliers who can convert pilot wins into guaranteed delivery schedules within 3–9 months.

Key tail risks and catalysts are governance, model trust, and reskilling velocity. A high-profile data-leak or a board-level rejection of agent-driven decisioning could freeze outcome contracts and reverse the annuity re-rating within quarters; conversely, published milestone KPIs from large rollouts (NPS, cost per ticket, churn reduction) or a single multi-hundred-million-dollar managed-services deal could re-rate winners quickly. Trackable near-term signals: backlog composition (outcome vs. time-and-materials), enterprise GPU orders/guidance, and counterparty contract language about indemnities and SLAs.

Key Decisions for Investors

• Long ACN (Accenture) — Buy shares or 12-month call (size 2% portfolio). Thesis: outcome contracts re-rate multiple; target +25% in 12 months. Risk: execution/reskilling failure; set protective stop at -12% or hedge with a 12-month 10% OTM put. Expected R/R ≈ 2:1.
• Directional NVDA exposure via 6–9 month call-debit spread — buy ATM call, sell a ~+30% OTM call (size 1.5% portfolio). Thesis: concentrated integrator-driven capex creates near-term inventory cadence and upside into quarterly guidance beats. Max loss = premium; potential payoff 2–4x if chips demand re-accelerates. Close on guidance miss or if enterprise order cadence lags by two consecutive quarters.
• Hedge/short META risk via 9–12 month put-spread — buy one near-ITM put, sell a deeper OTM put (size 1% portfolio). Thesis: product delays and enterprise trust issues compress monetization window; downside >30% possible on ad shock or model rollout failures. Put-spread caps capital at premium while offering asymmetric payoff if a catalyst hits.
• Selective long BX (Blackstone) — small position in shares or 18–24 month call options (size 1% portfolio). Thesis: private-equity portfolios that adopt outcome-based AI can drive EBITDA multiple expansion across holdings; expect gradual realization over 12–24 months. Risk: deal activity slowdown; target +20% with stop-loss at -10%.