Cognizant Splits 68,000 Staff Into Claude, OpenAI, and Gemini Units

Cognizant just did something no other IT services company has done at this scale: it carved its workforce into three distinct armies, each loyal to a different AI platform.

In the span of 22 days, the $5.7 billion IT services giant announced dedicated delivery units for Anthropic's Claude (25,000+ staff, November 11), OpenAI (25,000+, October 30), and Google's Gemini (18,000+, October 19), as ² first reported. Together, the three groups hold more than 68,000 professionals — nearly a fifth of Cognizant's 359,900 employees.

The groups are distinct and non‑overlapping. Each associate is assigned to only one AI unit at a time, though they can move between groups as skills and certifications change.

The Claude unit introduced something new: a managed services model that bundles dedicated staff, licensed Claude access, and Cognizant's proprietary AI Delivery Operating System into a single contract. The pitch is predictable AI spend — clients pay one price and Cognizant owns Claude in production, licenses and all.

This is the classic MSP playbook applied to a frontier model. As Channel Dive reports, Paul Smith, Anthropic's Chief Commercial Officer, framed the talent crunch directly: "The people with the skills to make it work inside a real business have become the hardest talent to find."

The initial focus areas are redesigning business domains around AI, modernizing underlying systems, operating agents securely at scale, and engineering AI‑native software.

CEO Ravi Kumar S told analysts on the October 29 earnings call that Cognizant now has "a lot of deals bundling Anthropic's Claude, OpenAI's Codex, and Cognizant's own inference units along with services." In high‑volume workflows, he said, clients "no longer want to buy software and services separately."

The bundling strategy helped lift Q3 revenue 4.8% in constant currency to $5.7 billion,.² But CFO Jatin Dalal cautioned that Q4 will see a slightly lower contribution from resales, and the full‑year growth outlook was trimmed to 4%-4.4%.

Cognizant is explicitly not betting on a single model provider. By creating three parallel units — each with its own leadership, training pipeline, and client engagement model — the company is positioning itself as the Switzerland of enterprise AI deployment. Whatever platform a client picks, Cognizant has an army ready.

Accenture made a similar move last December with a ~30,000‑person Claude training group. But Cognizant's three‑platform approach is more aggressive — it's essentially running a multi‑model experiment at the scale of a Fortune 500 company.

Omdia Chief Analyst Alastair Edwards told ² that he expects power to tilt toward partners within 12‑18 months as AI vendors lean on deployment skills they don't have in‑house. But Gartner warns buyers to watch for providers that rebrand traditional consulting as specialized AI deployment and charge a premium for it.

If you're a developer or engineer working in enterprise AI, Cognizant's restructure is a signal about where the jobs are going:

Platform specialization is becoming a career path. Cognizant's Frontier Certified Engineers are explicitly positioned as "model‑agnostic builders" — people who can work across Claude, Codex, and Gemini. The fastest‑growing job designation in tech right now is the Forward Deployed Engineer, with postings surging from under 200 in 2023 to over 7,300 in the first nine months of 2026, per ¹ citing Lightcast data.

The talent bottleneck is real and deepening. Only a couple thousand practicing FDEs exist, creating fierce competition. AWS funneled $1 billion into an FDE business unit. OpenAI acquired Northslope, an enterprise deployment company. Cognizant is betting that owning the deployment layer — not just consulting on it — is where the margin lives.