IGI Logo

IT Sector: Time series analysis of the pre and post pandemic era

We need to comprehend and dissect the historical growth rates for the IT sector – pre and post pandemic, and the much-touted slowdown over the last two years. For simplicity, we are sticking to averages of the top three IT companies (TCS, Infosys, and HCL) for our study. There will, no doubt, be a few outliers within each sector. The IT sector's pre-pandemic (FY17-20) constant currency (CC) revenue CAGR averaged at 10% YoY, while the rupee revenue CAGR was marginally higher, at 11.3% YoY. Likewise, EBITDA growth mirrors rupee revenue growth, at 11.3% CAGR over the same period. After a brief stall during the early days of the pandemic, revenues surged later – however, CC dollar growth averaged at 10.5% over FY20-23, or was just about 50bps higher than the pre-pandemic average. However, the rupee revenue CAGR was materially higher at 14.5%, even as EBITDA growth averaged just about 100bps higher than the prepandemic average of 12.6% YoY. In a nutshell, the marginal outperformance in revenues and operating profit in the immediate post-pandemic years could be largely attributed to a weakening rupee. Now we can argue (with a hindsight bias) that the rupee surge in revenue in FY22/23 over-compensated for the stall in the growth rates that was to come later

Over the last two years—specifically FY24 and FY25—IT sector revenues have stalled significantly, with CC dollar growth averaging at only 3.2% YoY and 4.4% YoY, respectively. Over the same period, operating profit growth averaged at 5.8% YoY and 6.4% YoY. Such low revenue growth rates, averaging at only about half that of the pre-pandemic averages, need to be questioned and understood.

Impact of Macros: Convenient, but the half truth

Most analysts attribute the slowdown in the IT sector over the last couple of years to global macros and the uncertainty caused by higher geo-political risks and trade wars. While these issues are relevant, we can argue about the degree of impact these would have had on IT companies. Global GDP, after rising 3.6% in CY22, held up considerably well in CY23/CY24, rising 3.5% YoY and 3.3% YoY, respectively (corresponding to the IT sector’s FY24 and FY25 revenues). Interestingly, even this minor moderation in global GDP was primarily led by Europe – with real GDP growth collapsing from 3.7% in CY22 to 0.6% in CY23, before recovering modestly to 1.2% in CY24. Meanwhile, USA’s growth actually accelerated, from 2.5% in CY22 to 2.9% YoY and 2.8% YoY in CY23-24.

Surprisingly, it is the US geography that is causing the most pain. IT sector’s US revenues grew only 1.8% YoY and 1.1% YoY in FY24/FY25, while Europe’s revenues rose by a healthy 8.7% YoY and 6.1% YoY over the same period. Of-course, an optimist may argue that Europe had to step up its outsourcing due to the energy crisis – in our view, this still fails to explain the muted performance in the US geography despite the robust economic growth. Incidentally, IT bulls should be more worried now, given the decisive slowdown witnessed in leading indicators over the last few months. Without disregarding the macro uncertainty, we believe the problems faced by the IT sector are more structural than cyclical.

Impact of GCC: Friend or foe?

While most IT Services players will not admit it, GCCs are increasingly competing with incumbents for business, as they insource more technology functions like R&D, digital strategy, and AI work. Media reports indicate that about 65% of enterprises are relocating at least 10% of their workforce from their vendors to GCCs. It is hard to imagine that a near-40% growth in GCCs’ revenues in CY24 did not cause a rather sharp slowdown in IT Services revenue growth in the same year. Say JP Morgan decides to outsource X volume of work to a low-cost offshore destination in India. This X volume then gets divided between its own GCC and that of its vendor – depending on the criticality, compliance, strategic importance, etc. How difficult is that to understand?

Impact of AI: When the disruptor gets disrupted

I am not an expert on AI, but then I am not tasked with writing an algorithmic model for LLM or SLM. It is a well-known fact that the most common use case of Generative AI is automated production of code, music, and art. Indeed, the response to my question “How efficient is generative AI in writing software codes” on Perplexity was this: “Generative AI is highly efficient in writing software codes, with multiple studies indicating productivity improvements ranging from 10% to 50%, faster developmental cycles, and enhanced code quality when integrated effectively into developmental workflows”. It then, of course, is caveated with the fact that human oversight is required for best results and to mitigate risks – which I completely appreciate. While there are upfront costs for making generative AI, it is clear that the marginal cost of writing incremental software codes will fall significantly, approaching zero in the not-too-distant future. While analysts can suck their thumbs, this will neither be lost to the clients nor to their vendors. If the output of the AI models is largely similar (in terms of coding efficiency), software pricing will be commoditized, and industry revenues will stagnate or even deflate. If our hypothesis is correct, the IT Services industry is at near-peak employment or close to it. Dipesh, our software czar, argues that deflating prices will open up more use cases (read: more volume) and is thus not a material threat. I cannot disregard this entirely, but find it difficult to believe that the cause and impact will sync in so well as to keep the industry growth intact.

Secondly, it is almost a given that Generative AI models will become more powerful and more disruptive with time. It is endemic to agentic AI that the improvement beyond a threshold will be non-linear and in ways that are not completely comprehensible now. The odds of that eventuality, though, are high. I think it is critical to understand what billions of dollars in datacentre investments and a few trillion dollars in AI market-cap mean. What does a 30% increase within one day, in the market cap of a USD700bn Salesforce mean? Most forecasts for AI’s impact on GDP growth in developed markets such as the US or Europe estimate productivity gains of 1-3% per year, with cumulative GDP lifts of 10-15% over a decade. If ball-mark projections are correct, the mega-tech trillion-dollar market-caps are essentially building-in value migration away from other sectors of the economy. For every legal AI model, scores of legal firms will be impacted. For every Animated AI model, the business model of production studios will be impacted, and so on. The main point I want to make is that over the next few quarters, we are likely to witness a sudden 20% drop in prices of sectors or companies that are being disrupted. If AI is real, this is certain to happen; the only question is—when? If I see a 10,000-tonne truck coming at me from afar but am unsure about its cruise speed, I will prefer to stand away rather than attempt to guess the rate of acceleration.

I invite you to read an interesting book on technology investing, titled ‘Engines that move the market: Technology Investing from Railroads to the Internet and beyond’, which is a deep research on all major technology innovations in the past century, specifically from an investment perspective. Among the 11 commandments mentioned in it, I am highlighting two that I found most relevant: i) All new technologies veer from capital starvation to capital surplus and back again and ii) spotting the losers is easier than spotting the winners.

For More  Emkay Global Financial Services Ltd Disclaimer http://www.emkayglobal.com/Uploads/disclaimer.pdf & SEBI Registration number is INH000000354