Grid Dynamics: AI Revenue Crosses 30% as Physical AI Moves from Pilot to Business

Q2 2026 beat guidance, the AI mix shift became measurable, and Ekumen + Doosan give the robotics bet real moat — at ~1.2x revenue

GDYN · Earnings Call · 2026-07-30

AI revenue crosses a threshold — and the mix shift becomes measurable

Grid Dynamics' Q2 2026 was the quarter where the AI narrative hardened into an accounting fact. Consolidated revenue of $108.2 million landed above the high end of guidance, with AI revenue reaching 30.7% of the total — up 54.6% year-over-year and the first time it crossed the 30% threshold. The CEO, Leonard Livschitz, was explicit that this is structural, not cyclical.

Two consecutive quarters of the year-over-year growth over 50% tells us something important. This is not a spike. It's a sustained shift.
Leonard Livschitz, Chief Executive Officer · 2026-07-30

The composition underneath matters more than the headline. Top accounts in technology and financial services — precisely the sectors with the deepest AI adoption — are now embedding the GAIN platform as core infrastructure, and vendor consolidation has made GDYN a preferred supplier rather than one of many. That is the enterprise scale transition leadership keeps referencing: clients moving from isolated pilots to platform-grade deployments. The application modernization engine (Horizon 1) is being re-accelerated by agents — at one Fortune 30 home-improvement retailer, seven COBOL services moved to a modern stack with roughly 90% of code agent-generated, all seven in production this quarter. The profitability math is also changing. Revenue grew 7% while total headcount dropped from 4,964 to 4,838, and the CFO was explicit about the driver: “We continue to rationalize our overall headcount as well as align our skill sets and geographic mix” — Anil Doradla, Chief Financial Officer · 2026-07-30 — overwhelmingly non-engineering, non-billable roles. GAAP gross margin stepped up to 36.6% from 34.1% a year ago, and non-GAAP EBITDA margin to 13.6%. For a company that has been range-bound in the mid-30s on gross margin for two years, this is the first visible step toward the 300-basis-point margin-expansion commitment made last year. GAAP gross margin of 36.6% in Q2 2026 vs 34.1% a year ago — the clearest evidence yet that AI-native delivery and cost rationalization are translating to the P&L.

Physical AI: the genuinely new strategic vector

If the AI-revenue story is about acceleration, physical AI is about optionality. The Ekumen robotics team joined in May, and this is not a typical tuck-in: Ekumen's engineers are contributors to and maintainers of the Robot Operating System, the open-source foundation used by most of the world's industrial robots. A strategic partnership with Doosan Robotics — which deploys cobots across 45 countries — adds a distribution channel for GDYN's GAIN-for-Physical-AI platform. The company's own keyword trajectory shows how new this is: "Robot," simulation, and "Consultancy" all jumped to the top gainers in the current quarter, alongside brand-new terms like Ekumen, Rosetta, and Allium. Yury Gryzlov, head of Physical AI and Robotics, framed the moat in terms of accumulated engineering experience rather than capital:

We don't believe another services company matches it today. Closing that gap isn't a matter of hiring a team. It's years of hard-won experience.
Yury Gryzlov, Head of Physical AI and Robotics · 2026-07-30

Early engagements are real: a multi-year autonomous-driving and excavation program at a large construction/mining equipment maker, humanoid robots piloted for pharmaceutical intralogistics, and Doosan embedding GDYN's platform in its own offering. This is nascent revenue, but it positions GDYN as one of the few services firms whose robotics bet is on software and simulation-to-hardware validation rather than body-shop labor.

Allium, governance, and the banks — where the margin thesis compounds

The second new theme is spec-driven agentic engineering aimed at regulated financial institutions. Allium, part of the GAIN SDLC suite, is being piloted across five major banks and moving into first commercial engagements; at one Tier 1 bank it is supporting a security modernization program spanning more than 100,000 systems. The pitch is governance — giving regulators evidence from AI-generated code — and the CTO, Eugene Steinberg, distilled the differentiation cleanly: “An agent can write a code. A person still makes a call and stands behind it” — Eugene Steinberg, Chief Technology Officer · 2026-07-30 — the accountability layer is the actual product, turning powerful agents into dependable enterprise systems. This was a live theme on the prior call, when leadership told analysts that fixed-price and outcome-based AI work was already priced at a premium — “non-T&M in general being higher... very positive outliers” — Anil Doradla, CFO · 2026-05-01 — and the current quarter is evidence that thesis is compounding. The same book of business that, in management's words on an earlier call, held “the highest billable headcount in the history of the company” — Leonard Livschitz, CEO · 2025-10-30 is now being run with fewer, more productive people. The CEO's own framing of the model change is that “clients who came to us for platform deployments now ask us to stay” — Leonard Livschitz, Chief Executive Officer · 2026-07-30 — an advisory and ongoing-operations vector that didn't exist two years ago. The market has already begun to price this in: GDYN is up roughly 51% over the past 90 days, though the full trend still shows an 81% drawdown from the 2021 peak to recover. On roughly 1.2x trailing revenue, the market is pricing GDYN as a services company — asymmetry if the physical-AI channel pays off. Management kept full-year revenue guidance at $435–465 million, which implies a meaningful H2 ramp into Q4 — a pattern analysts pressed on in prior calls. The bullishness rests on vendor consolidation, deeper AI wallet share, and an Agentic AI pipeline that, per the COO, has "strengthened over the last few years" as customers cut loose the long tail of vendors. In a quarter when much of the market's attention was on tariff refunds and proof of concept economics, GDYN's story is the rarer kind: a services company whose differentiation is software it actually controls.