Skip to main content
    All AI News
    Discovery — Partner / MDWednesday, September 16, 2026 9 min read
    AI

    The Consulting Firm Agent Practice Race 2026: Accenture, McKinsey, BCG & Capgemini

    OpenAI's consulting alliances signal that enterprise agent revenue flows through SIs, not API sales.

    Key takeaways
    • 01OpenAI's February 2026 'Frontier Alliances' with McKinsey, BCG, Accenture, and Capgemini formalize a two-tier model: strategy firms define agent operating models, systems integrators handle the messy implementation.
    • 02The arrangement mirrors ERP rollouts of the 1990s—which generated decades of recurring services revenue.
    • 03With the consulting AI market forecast to reach $72.8 billion by 2030 and agents already representing 17% of enterprise AI value, the race for dominant position is live, funded, and generating real margin data.
    Koko brief

    OpenAI's consulting alliances signal that enterprise agent revenue flows through SIs, not API sales.

    OpenAI's February 2026 'Frontier Alliances' with McKinsey, BCG, Accenture, and Capgemini formalize a two-tier model: strategy firms define agent operating models, systems integrators handle the messy implementation. The arrangement mirrors ERP rollouts of the 1990s—which generated decades of recurring services revenue. With the consulting AI market forecast to reach $72.8 billion by 2030 and agents already representing 17% of enterprise AI value, the race for dominant position is live, funded, and generating real margin data.

    Watch: Which tier—strategy or SI—captures higher margins at scale; early ERP history suggests implementation incumbency compounds faster than advisory relationships.

    In brief · from agentmarketcap.ai

    On February 23, 2026, OpenAI made a bet that the future of enterprise AI would not be won by selling API keys. The company announced "Frontier Alliances" — multi-year, certified partnerships with McKinsey & Co., Boston Consulting Group, Accenture, and Capgemini to jointly deploy Frontier, its new AI agent platform, inside the world's largest organizations. The message was unambiguous: to get autonomous agents running in a Fortune 500 company's core operations, you need a trusted system integrator standing between the technology and the enterprise.

    Read the full article at agentmarketcap.ai
    Show the full text · 9 min read

    On February 23, 2026, OpenAI made a bet that the future of enterprise AI would not be won by selling API keys. The company announced "Frontier Alliances" — multi-year, certified partnerships with McKinsey & Co., Boston Consulting Group, Accenture, and Capgemini to jointly deploy Frontier, its new AI agent platform, inside the world's largest organizations. The message was unambiguous: to get autonomous agents running in a Fortune 500 company's core operations, you need a trusted system integrator standing between the technology and the enterprise. And whoever locks up that position stands to capture the lion's share of a consulting market forecast to grow from $14 billion in 2024 to $72.8 billion by 2030. The consulting firm agent practice race is no longer a future prediction. It's a live revenue competition with billions already deployed, staffing battles underway, and the first genuine data on which practice models are generating margin. The OpenAI Frontier Alliances: A Watershed Moment OpenAI's Frontier platform is designed to function as a "semantic layer for the enterprise" — a unified control plane that lets AI agents navigate business software, execute workflows, and make decisions across an organization's full technology stack, from Salesforce CRM to SAP ERP to internal ticketing systems. It's ambitious, but also exactly what enterprises say they need: governance, auditability, and a single pane of glass for agent orchestration. The choice of Frontier Alliance partners was strategic and revealing. OpenAI didn't partner with boutique AI consultancies or cloud resellers. It went directly to the firms that have spent decades earning CIO and CEO trust on transformational IT programs. The four alliance partners divide naturally into two lanes: Strategy and operating model: McKinsey (with its ~5,000-person QuantumBlack AI unit) and BCG (with its ~3,000-engineer BCG X division) are chartered to help leadership teams define where and how to deploy agents — the "why" and "what" before a single line of code is written. End-to-end systems integration: Accenture and Capgemini go deeper into data architecture, cloud infrastructure, and the messy work of connecting Frontier to legacy enterprise systems. They stay in the engagement after strategy is signed off. This two-tier structure mirrors how enterprise ERP rollouts worked in the 1990s — strategy consultants designed the operating model, SIs delivered the implementation. The parallel is not lost on industry observers, because ERP created decades of recurring professional services revenue for the firms that dominated early implementation. What Each Firm Is Actually Building The investment numbers are striking in their scale and specificity. Accenture announced a $3 billion investment in its Data & AI practice spanning three years, with a goal to double its AI talent to 80,000 professionals. As of early 2026, Accenture has exceeded that target, employing more than 85,000 AI and data professionals. The firm has also launched "AI Refinery for Industry" — a framework for building reusable, production-grade industry agent solutions, with over 50 sector-specific agent blueprints already live across telecommunications, financial services, and insurance, with a target of 100 by year-end. In parallel, Accenture has certified a dedicated team of more than 25,000 Databricks-trained professionals to handle the data infrastructure layer that underpins most agent deployments. McKinsey's QuantumBlack operates as a standalone AI product and delivery unit embedded within McKinsey. With approximately 5,000 AI specialists and strategic acquisitions including Iguazio (data infrastructure), QuantumBlack has evolved from a data science advisory arm into a full-stack agent development and deployment capability. McKinsey is notably focused on the "operating model transformation" side — redesigning how workflows, incentives, and reporting structures need to change when AI agents become a permanent part of the workforce. BCG X — the firm's technology build-and-design unit — brings roughly 3,000 engineers to the table and has been expanding rapidly. BCG's focus has been on combining its proprietary strategic IP with technical delivery, aiming to avoid the pure-labor-arbitrage model that has historically commoditized IT services. Deloitte launched its Enterprise AI Navigator in September 2025, an orchestrated library of pre-built AI agents designed to help enterprises move from pilot to production. The firm has expanded its alliances with Google Cloud and ServiceNow specifically to address agentic AI demand, while its Deloitte AI Academy is upskilling tens of thousands of practitioners for agent-era engagements. Collectively, the Big Four and major strategy houses have poured over $10 billion into AI initiatives since 2023. The market context explains why: AI agents already account for approximately 17% of total enterprise AI value in 2025 and are projected to reach 29% by 2028. This is the fastest-growing slice of the AI pie, and consulting firms want the implementation fees that come with it. Three Practice Models Competing for Margin Not all agent practices are structured the same way. Across the industry, three dominant delivery models have emerged — and they have meaningfully different margin profiles. | Model | Description | Who's Leading | Margin Profile | | --- | --- | --- | --- | | Agent Factory | Centralized team builds standardized agent templates, deploys repeatedly across clients | Accenture AI Refinery, Capgemini | High — reusability drives efficiency | | Center of Excellence (CoE) | Embedded team sits inside client organization, builds custom agents over 12–24 months | Deloitte, McKinsey QuantumBlack | Medium — deep stickiness but high cost | | Federated Embedded | Small squads attached to existing transformation programs, agents as an add-on workstream | BCG X, EY wavespace | Variable — margin depends on deal structure | The Agent Factory model is gaining traction as the most scalable. Accenture's AI Refinery approach — building reusable "industry agent solutions" that can be re-deployed across multiple clients in the same vertical — allows a single development investment to generate fees from dozens of separate enterprise engagements. A telecoms agent template built for Verizon can be adapted for Deutsche Telekom in weeks rather than months. The CoE model generates the deepest client stickiness but requires the highest headcount commitment. Deloitte's Enterprise AI Navigator is specifically designed to accelerate CoE buildouts — giving clients a structured library of pre-built agents they can customize rather than starting from scratch. The tradeoff is that clients who successfully build internal CoEs eventually reduce their consulting dependency, which creates a natural tension between short-term engagement revenue and long-term relationship health. The Revenue Mechanics: Is Outcome-Based Pricing Actually Happening? One of the most-watched questions in the consulting industry has been whether firms would shift away from time-and-materials billing toward outcome-based pricing as AI agents make productivity gains measurable and attributable. The 2026 data is sobering. Despite years of rhetoric about "shared value" and "risk/reward partnerships," only approximately 25% of consulting fees globally are currently linked to outcomes at even the most progressive firms. The inertia comes from both sides: consulting partners have limited appetite for revenue at risk, and enterprise procurement teams default to familiar billing structures. The exceptions are instructive. Accenture has reported client results where advertising teams saw productivity rise by 50% while AI-powered campaign revenue increased by more than 20%. These outcomes are now being used as pricing anchors in proposals — not yet as pure outcome fees, but as benchmarks that justify premium fixed-fee structures over T&M baselines. Analysts project that Acce

    Don't miss tomorrow's

    The Daily Pulse in your inbox each morning — sourced and linked.

    How often
    Keep going — across the app