Companies still don’t know how to incorporate AI in a holistic way, says Wharton expert

Wharton vice dean Eric Bradlow argues AI’s main bottleneck is organizational change and keeping humans “in the loop,” not the underlying technology. He highlights that large language models increase the value of deep expertise and that top firms should “redistribute talent” and focus on revenue expansion via new business models, supported by AI training and governance to avoid getting stuck in pilot programs.

Analysis

The near-term monetization pool for AI is shifting away from model vendors and toward firms that can translate pilot projects into operating changes. That favors ACN and other systems-integrators/consultancies because the scarce input is not compute, it is workflow redesign, governance, and training; those budgets tend to be sticky and recur through multi-quarter transformation programs. The second-order effect is that enterprise customers may spend more on implementation services before they spend less on labor, so the first visible P&L uplift could be in billings and backlog rather than headcount reduction.

The market is still likely overpricing a fast software productivity payback. If management teams cannot prove AI-driven revenue lift, the narrative can stay trapped at the pilot stage for 1-3 quarters, which is a headwind for high-multiple application software that needs rapid adoption to justify valuation. Conversely, firms with deep domain expertise and change-management capabilities should see better pricing power as clients move from experimentation to controlled rollout; that makes the setup constructive for ACN relative to pure-play AI software and for legacy IT services that can sell reskilling plus governance.

The contrarian miss is that the biggest AI winners may not be the obvious “AI winners” everyone is buying, but the companies closest to the organizational bottleneck. The structural catalyst is 6-18 months out: once enterprises redesign processes, AI can drive new revenue products, not just cost cuts, which should expand TAM for services and potentially re-rate multi-line consultancies. Falsifier: if clients internalize transformation work and materially cut outside consulting spend, ACN’s upside shortens quickly; if not, this is a steady multi-quarter tailwind rather than a one-day event.

Key Decisions for Investors

• Long ACN on a 1-3 month horizon on any pullback: thesis is that AI budgets increasingly flow into integration, governance, and training rather than model purchases; risk/reward is favorable if bookings acceleration shows up in the next earnings cycle.
• Pair trade: long ACN / short a software basket proxy such as IGV over the next 1-2 quarters to express the view that AI adoption is slower to monetize in software than in services; stop if software vendors begin showing clear AI-driven net retention or ACV expansion.
• Set an alert on ACN for backlog and consulting margin commentary at the next print: if AI-related bookings do not convert into revenue within 1-2 quarters, the thesis shifts from bullish to watch-only.
• Avoid chasing broad AI beta here; the article argues for a delayed ROI curve, so the better expression is quality implementation names rather than expensive pure-play AI exposure.
• If enterprise clients start citing in-house AI operating-model teams on conference calls, reduce exposure to ACN and other services names; that would be the main falsifier of the outsourcing thesis.