Some Additional Thoughts on IT Services

I've spent the the last few weeks diving deeper into IT services as well as Globant, an Argentina-based near-shore IT-services provider. I will be releasing a post on Globant shortly. While Globant has a unique operating model, happy clients, and happy employees, my conclusion is that Globant's quality of earnings is very low. I still believe Reply offers a very attractive risk-reward at today's valuation.

Please read my Deep Dive: IT Services Primer before reading my notes on Globant. Before diving into Globant, I want to set the stage.

Point 1: IT services spending is secularly growing and will rebound.

Historically, IT services spending grew 1% to 2% above GDP driven by the importance of technology to all businesses in all industries. This has been true through major technology cycles including the move from mainframes to client server in the 1980s, the rise of the internet, ERP and CRM systems in the 2000s, growth in social media, mobile, analytics, and cloud computing in the 2010's and more recently the emergence of AI. Prior trends like cloud which promised to eliminate the need for costly implementations only fueled demand for consulting and systems integration as businesses needed help with migrations and as new incremental use cases of software arose. While IT services spending, particularly shorter-cycle consulting projects, declines during recessions, spending bounces back quickly as IT is mission critical to most businesses. As the table below shows, IT spending is resilient.

We are currently in the worst downturn in the last 25 years. While IT services spend historically grew below GDP in 2002, 2009, and 2020, spending bounced back quickly. We are in a 3-year slump (2023, 2024 and potentially 2025) where IT-services spend is below GDP growth and this isn’t a huge surprise. From 2021-22, we saw a surge in IT services spending due to the COVID-fueled rush to digitize combined with stimulus and accommodative rates. The macro environment has cooled. Normally, I would expect 12 to 18 months of soft spend but this time is different as we are digesting COVID spend. Moreover, AI is, at least in the short-term, crowding out spending in traditional areas like cloud migrations as CIOs and CTOs figure out their AI strategy before continuing on with these plans. These are cyclical, not structural headwinds.

Discretionary programs are paused because of significant investments in GenAI, GPU and AI. 

Infosys CFO - Q1 F2026 Call

Once we digest the excess COVID spending and once companies formulate a preliminary AI strategy, we should see a reversion to growth, probably by 2026, and a likely acceleration in growth medium-term. However, the market has freaked out, especially with consultants focused on shorter-cycle consulting projects like Globant, EPAM and Endava. Incidentally, while these consultants are the most cyclical, they and are best positioned to benefit from AI. Over the last year, Globant is down 69%, EPAM 40%, Endava 59%, and Grid Dynamics 65%. They are now trading at low-teens and in some cases single digit P/E multiples whilst in many cases, these companies are poised to benefit from AI and can grow EPS double digits organically.

Point 2: AI will accelerate IT services demand longer-term

Clients need help infusing AI into their technology and business processes. Unless they are a sophisticated tech company, they don’t have this expertise in-house. Moreover, at the early stages of a disruptive technology like AI, a consultant’s expertise and capability are far more important to price. For example, if you are replacing a $40k/year customer service employee with $2k/year agent, you are far more concerned about the quality of that agent than saving 30% on your IT services bill. You are going to pay up for the best expertise. As a result, the IT services players that have focused on differentiated value propositions with true subject matter expertise are poised to benefit. As AI goes from the first to the second or third inning, AI will require significant technical lifts in terms of upgrading architecture, moving applications and data to the cloud and cleaning up data and there is a huge runway as 50%+ of enterprise workloads are still on-prem.  

Most of our clients started to prepare themselves for the AI adoption. In order to do AI adoption, they have to really touch upon their fundamentals. Fundamentals, meaning they have to take a close look at their legacy infrastructure, start doing modernization, going back shifting to the cloud and really addressing the backbone, which is data, which is in order to really adopt AI and actually roll out AI solutions in the enterprise, you need to make sure that your data environment, your core data assets are in good shape. This is very much playing to the sweet spot of EPAM. 

EPAM President of Global Business & CRO - Q2 F2025 Call

Point 3: AI is a boon for IT- services companies that design and build.

To reiterate the prior point, IT services capabilities will matter far more than cost. While onshore consultants like Accenture will benefit, Accenture has massive BPO operations that may cannibalize. Over the last 15 years, Accenture have spent billions building out managed services and BPO operations with the goal of leveraging their consulting relationships to cross-sell BPO work. Accenture, CGI and Capgemini all generate 40% to 50% of revenue from managed services and BPO. Their approach was to maximize EBITDA/Client, not EBITDA/head, even if it meant taking on large BPO contracts with limited differentiation. Indian companies have pursued the same goal, albeit they’ve had the harder task of translating BPO work into more strategic consulting work. As a result of these legacy operations, these businesses at best face a credibility problem and at worst face an innovator's dilemma.

Boutiques like Reply, Globant and EPAM have, by necessity, competed by focusing on the fastest growing area of technology where they can offer authoritative expertise in areas like AI, mobile development, and cloud computing. Companies like Reply (and Globant, as I will talk about) have developed innovative organizational models ensuring that they always stay on the cutting edge, something that is difficult to do within the context of a rapidly changing tech landscape. While Reply, Globant or EPAM aren't viewed in the same light as McKinsey when it comes to their strategic expertise, they are positioned as a differentiated consultant with true AI and engineering expertise without any legacy business to defend, which in turn makes them a credible partner to provide unbiased AI advice.

While many technology players are jumping on the AI revolution, branding themselves as experts without having made any real investments in the field, Globant has been investing in AI for the past 10 years. This long-term commitment means we are not merely riding the wave of the latest trends. We have developed deep proven expertise in the field.

Globant Co-Founder, Chairman and CEO - Q4 F2024 call

Point 4: AI opens up a new TAM for “Design and Build” companies to also “Run”

Accenture's massive BPO operations were historically a barrier to entry for boutiques like Globant, EPAM and Reply. As a result, while these boutiques could go after small to medium-sized deals, they were never a credible player on a $100m IT-services contract. BPO operations often involve discrete manual processes in low cost locations, like data entry for insurance claims or handling account cancellations over the phone. AI has the potential to replicate this functionality far cheaper to even offshore labour. To the extent there is a large $25m or $50m contract involving AI-powered BPO, specialized boutiques like Globant and EPAM can leverage AI to go head-to-head with Accenture. In fact, they may actually have more credible references doing AI driven projects and they don't have a legacy business to protect. The entire BPO TAM is incremental to these boutiques allowing them to be aggressive against incumbents. As you may recall from the primer, BPO is a massive $260bn TAM, equivalent in size to the entire IT consulting TAM.

Point 5: There are legitimate concerns around some cannibalization, but concerns are overblown.

First, capabilities will be more important to costs. While this clearly benefits Accenture and high-end consultancies (big 4 accounting, big 3 strategy), the nearshore players are well positioned given their focus on faster-moving innovative technologies and their “clean” business model unburdened with lower margin BPO work. Second, to the extent AI helps drive consultant productivity, e.g., speeds up coding, these gains will be shared between the end-client and consultant driving margins higher in the process, as long as the services being offered are cutting edge. We already have several examples of this. Reply called out higher margins in H1 F2025 due to productivity gains from leveraging AI which were shared with end-customers.

What we can say is that we are keeping these [higher] margins, and we are growing less in terms of people than we are growing in terms of revenue. We are increasing the value that each person generates. So, if you want, that is the simplest math. We believe that [higher margins] is mostly due to that. Then it's not so clear because there are some improvements in bench management in some countries. But most of that effect is more productivity due to more efficient way of working.

Reply CEO - Tatiana Rizzante - H1 F2025 Call

The existential threat is that AI will eventually commoditize IT consultants. Rather than implement a new software, you can simply ask AI to instantly build and implement a custom system from scratch. While this is a long-term risk, these concerns vastly overstate AI’s current potential and the propensity for enterprises to change. For context, most banks, governments, and insurance companies still run on mainframes with software coded on COBOL. Cloud penetration is still just 50%, and the vast majority of data requires cleaning and organization.

Summary

IT Services is a secularly growing industry that has a long track record of bouncing back. Within this industry, boutiques like Globant have historically grown double digits organically and have massive medium-term tailwinds from AI and an opportunity to penetrate an entirely new TAM. However, the bounce back in service spend is taking longer than usual and as a result, the market has punished these stocks to trade at historically low valuations despite AI having the potential to accelerate growth. My preferred way to express this idea is still Reply SpA given their unique and scalable operating model and their massive long-term runway.

In a forthcoming post, I will do a deep dive on Globant (GLOB). While Globant's studio model is interesting, it is ultimately a pass for me due to the poor quality of earnings.