Highlighted Holding: IT Services

DOES AI SHRINK THE REVENUE POOL?

The recent selloff in IT services stocks reflects a fear that AI will pressure an industry that has long billed for people’s time. If a project that once required ten engineers can now be done by seven, the client will expect to pay less, leading to lower revenue per contract. That is already visible in renewals, where annual price step-downs that once ran 3% to 4% are now closer to 6% to 7%.

There are two offsets. The first is a change in how contracts are priced. When a provider bills by the hour, every efficiency gain flows straight to the client as fewer billed hours. As work becomes AI-enabled, providers are instead signing fixed-price and outcome-based contracts, where the client pays for a result rather than hours. A provider that delivers that result at lower cost keeps part of the savings. The second offset is that cheaper work means more work, as the industry proved through prior deflationary waves from offshore delivery to cloud. Large companies still have long backlogs as they modernize aging systems and prepare for AI. As costs fall, project ROI improves, making more of that backlog worth doing and allowing volume to offset price. Consistent with that view, growth has slowed across our three holdings, but revenues have broadly held, despite steep drops in per-contract pricing (Exhibit 2).

Exhibit 2: Growth Slowed, But No Collapse

Quarterly Revenue, Indexed to Q1 2023

^{Source: Company reports, Pzena analysis
Constant currency, as reported. Accenture fiscal quarters mapped to calendar quarters.} 

DOES AI REPLACE THE INDUSTRY 

The second fear is that AI disintermediates the industry, shifting technology work away from IT services firms altogether. It could take three forms. Clients may bring more work in-house, reversing the long trend toward outsourcing. Software vendors and frontier labs may move downstream into implementation. AI-native startups may become more credible competitors as AI shrinks the advantage of employing thousands of low-cost engineers.

Some of this will happen, especially for routine, well-defined work, but each threat has limits. Most companies still do not want to manage non-core technology transformation themselves, and AI is making that work more complex, not less. Software vendors and frontier labs may offer more implementation help, but the cloud transition showed that platforms still need independent partners to integrate, customize, and manage technology across messy enterprise environments. AI-native startups can move quickly, but they lack the embedded relationships, trust, and business context needed to win complex work from large enterprises.

That does not mean every IT services firm will win. The most exposed firms are those that simply rent out cheap engineering labor. The winners will be firms that combine deep engineering expertise with trusted client relationships and real knowledge of how the client’s business operates. Every transition creates new winners and losers, and the task is to identify which firms are more likely than not to come out ahead. We believe that Accenture, Cognizant, and Globant each make that case in a different way.

ACCENTURE 

Accenture is the world’s largest IT services firm. Because it over-indexes to consulting-led transformation work, which pairs business expertise with engineering, Accenture is in the room when clients set direction. That position earns a premium, visible in revenue per employee of roughly $90,000, nearly twice the level of India’s largest outsourcing firms. The mix also carries little of the commoditized application development that AI automates first. The stock has been cut in half over the past year on fears that organic growth is impaired—fears that were deepened by a June guidance cut and stepped-up acquisition spending. But many companies are still in the early stages of adopting AI across the enterprise. Clients are asking Accenture what to do about AI but deferring the large programs that follow, because spending on AI itself is crowding out the rest of the technology budget. That work is deferred, not gone, and we believe that Accenture is well positioned to capture it when those programs begin. Meanwhile, investors collect roughly 12% of the market capitalization annually in dividends and buybacks.

COGNIZANT 

Cognizant is more exposed to the bear case than Accenture, with more of its revenue tied to delivery and operations work where AI can reduce human effort. However, roughly 60% of its revenue comes from healthcare and financial services, where claims, payments, and risk systems cannot fail. These are not industries where clients can simply hand a workflow to an AI tool and hope it works. Cognizant is embedded in them, with its proprietary TriZetto platform processing roughly two-thirds of U.S. healthcare claims. Cognizant has also been aggressive in evolving its pricing model, with fixed-price and outcome-based contracts crossing 50% of revenue for the first time this year, accepting the risk of cost overruns in exchange for keeping a share of the productivity AI creates. Demand is holding up, with trailing-12-month bookings up 11% from a year ago, as new work outpaces the price deflation on each contract (Exhibit 3).

Exhibit 3: Cognizant Bookings Keep Growing

^{Source: Company reports, Pzena analysis}

GLOBANT

Globant is the smallest of the three companies and the most digital-native. Founded in Argentina and staffed largely across Latin America, it builds the consumer-facing side of technology—the streaming apps, games, and fintech experiences of clients like Disney and Google—not legacy systems or back-office operations. Its expertise is pairing design talent with engineering, and design currently remains a challenge for AI. Deciding what a product should be and how people will interact with it, refined through iteration with the client, is valuable; consumers pay for novelty while AI models converge on the familiar. Engineering will become more AI-assisted, and Globant is adapting how it sells that work. Subscription teams called AI Pods combine its engineers with AI agents in an early attempt to decouple revenue from billable hours. Revenue per employee, roughly $85,000, rivals that of Accenture and has risen every year since 2022.

CONCLUSION

AI will reduce the labor required on many projects, pressuring pricing and forcing IT services firms to adapt. That is why Accenture, Cognizant, and Globant are cheap, all three trading at double-digit free cash flow yields for businesses that convert 90% or more of their earnings into cash. While we expect a difficult transition, their conservative balance sheets buy time to navigate it, with Accenture and Cognizant holding net cash and Globant carrying little debt. At today’s valuations, investors are being paid for the risk that the transition proves arduous, while getting little credit for the possibility that these firms remain central to enterprise AI adoption.