Why Barclays sees more downside risk for digital IT services By Investing.com

Barclays downgraded EPAM Systems and Grid Dynamics to Hold and said the IT services demand backdrop has more downside than upside as AI reshapes the sector. The bank noted Q1 results were in line, but full-year guidance remains back-half weighted, suggesting clients are delaying near-term spending. Barclays also flagged EPAM's difficult path back to sustainable double-digit organic growth and said Grid Dynamics' retail exposure adds risk to growth acceleration.

Analysis

The key message is not that these IT services names are cheap; it is that the market is repricing the durability of their revenue model. When clients delay discretionary work to wait for AI-driven deflation, the first-order hit is backlog conversion, but the second-order hit is even more important: utilization falls, pricing power weakens, and every incremental revenue miss becomes a margin miss. That combination tends to compress terminal multiples faster than headline growth decelerates, which is why “value” screens can stay broken longer than expected.

EPAM looks like the cleanest expression of that risk because its recovery thesis depends on re-accelerating organic growth back to a double-digit regime. In a market where buyers are becoming more selective on transformation spend, that hurdle is high: even a modest slowdown can force another leg lower in estimate revisions. GDYN is more exposed to cyclical budget scrutiny because retail clients tend to pause non-essential modernization first, making revenue stabilization more fragile than the current multiple implies.

GLOB is the better relative short only if investors start distinguishing between quality and beta within the group. Its discount to the broader peer set may already reflect some caution, so the downside from here is more about estimate cuts than outright multiple compression. The contrarian risk is that the market may be underestimating how quickly AI implementation can re-open spend in specific verticals; if clients move from “wait-and-see” to pilot-to-production within one or two quarters, these names can re-rate sharply because positioning is already light.

Near term, this is a catalyst-light tape until guidance proves otherwise, which argues for expressing the view with options rather than outright equity shorts. The best setup is a time-based trade: the next 1-2 earnings cycles should reveal whether back-half weighted guidance is real deferral or just pipeline timing. If commentary improves on conversion and bookings, the shorts can unwind quickly because the stocks have already de-rated substantially year-to-date.

Key Decisions for Investors

• Initiate a bearish pair: short EPAM / long XLK for the next 1-2 earnings cycles. Thesis: EPAM’s need for re-acceleration makes it more vulnerable to estimate cuts than the broader software complex; downside is another 10-15% if bookings remain sluggish, while XLK provides macro/AI beta insulation.
• Use GDYN put spreads rather than outright short stock into the next print. Prefer 3-6 month maturity; the risk/reward is favorable if retail spending stays soft, but defined risk is important because the name can squeeze on any AI-related pipeline commentary.
• Hold GLOB as a relative long versus EPAM only if you want exposure to the sector with lower idiosyncratic downside. This is a tactical pair, not a structural buy; if there is no evidence of organic growth inflection in one quarter, exit and rotate to cash.
• For higher-conviction bearish exposure, sell a basket short of EPAM/GDYN and hedge with a long position in AI infrastructure beneficiaries. The idea is to own the spenders, not the intermediaries, over the next 6-12 months if enterprise AI adoption continues to pressure services demand.
• Set a trigger to cover shorts if management commentary shifts from ‘timing uncertainty’ to ‘budget reallocation toward AI initiatives’ with measurable bookings conversion. That would signal the market has moved from deferral to redeployment, which can reverse the thesis within one reporting cycle.