Accenture is the scale play. It is the largest systems integrator on earth, roughly $70 billion in revenue and 800,000 people, and in 2025 it folded everything into a single "Reinvention Services" unit under Manish Sharma so it can sell strategy, build and run as one motion. Julie Sweet’s pitch is that no one else can do the whole job: "the only company that can cover, at scale, everything from the AI foundation to reinventing nearly every part of a business."

On the ground that is a platform (AI Refinery, built on NVIDIA) plus tools for software and process work, wrapped in an unusually deep set of vendor "business groups", dedicated teams of tens of thousands trained on NVIDIA, Anthropic, Microsoft and Google tools, with Microsoft’s Copilot now rolled out to all 743,000 employees. It is shifting from billing by the hour toward fixed-fee and outcome-based work, and buying 35 to 40 companies a year to keep the capability fresh.

The catch is size itself. The largest workforce in the industry is also the largest bet that AI will keep needing people to deliver it, and Accenture’s message, "AI is not a replacement for expertise, it’s an amplifier" (Sharma), is precisely the story that protects 800,000 billable staff. Its platform, meanwhile, is a switchboard: it orchestrates models it rents from its partners, not ones it owns.

For all that, no one is better at turning AI strategy into delivery at scale. But this week the counter-argument showed up in the share price: Accenture fell about 18% in a single session, its worst since 2016, on a cut to its guidance even as revenue grew. That is the tape pricing the very deflation the "amplifier" story is meant to wave away.