Cognizant, Infosys & TCS: The Outsourcing Trio by the Numbers

Three companies show up again and again at the top of any H-1B sponsor ranking: Infosys, Tata Consultancy Services (TCS), and Cognizant. They are IT-services and consulting firms that place large numbers of workers at client sites, and together they have filed hundreds of thousands of Labor Condition Applications (LCAs) over the years the Department of Labor has published.

This article compares the three directly — not against Big Tech, and not product companies against staffing firms, but the trio against each other. You'll learn how their filing volumes stack up, how their role mixes differ, and — importantly — how to read the numbers without drawing conclusions the data can't support.

First, what these numbers are (and aren't)

Every figure below comes from LCA disclosure data. An LCA is the wage-and-conditions filing an employer must certify with the Department of Labor before it can petition for an H-1B worker. That distinction matters more than any single statistic in this piece:

• An LCA is not a visa. A certified LCA is a prerequisite, not an approval. The actual H-1B petition is adjudicated separately by USCIS.
• A certification rate is not a USCIS approval rate. DOL certifies that a filing is complete and the wage meets the prevailing floor; it is not vetting whether a person ultimately gets or keeps a visa.
• "Withdrawn" is not "denied." A withdrawn or certified-withdrawn filing was pulled by the employer, not rejected by the government.
• The offered wage is not take-home pay. LCAs list the wage the employer attests to, before taxes, benefits, and the realities of bench time or project changes.

Keep those four guardrails in mind. They are why volume alone tells you how much a firm files — not how many people it actually brought in.

The volume gap: Infosys leads by a wide margin

Across the disclosure data, the three firms' cumulative certified-track filings look like this:

Infosys sits in a class of its own, with roughly 60% more filings than either of the other two. TCS and Cognizant, by contrast, are nearly tied — separated by fewer than 600 filings across the entire history of the dataset.

That near-tie is a useful reminder that these are three distinct businesses that happen to converge on a similar model: staff-augmentation and systems integration for large US clients, delivered by a globally mobile workforce. The H-1B program is a channel each of them leans on heavily, but the scale of the lean differs.

You can see how the broader sponsor leaderboard is shaped — and where these three sit relative to everyone else — in the live ranking below.

Role mix: what these firms actually file for

Volume is one dimension; the kind of work is another. Infosys's filings concentrate heavily in a handful of computing occupations. Its top roles by filing count are:

The dominance of "Computer Occupations, All Other" — a catch-all SOC code for computing jobs that don't fit a more specific title — is characteristic of the services model. Client engagements span configuration, integration, support, and analysis roles that resist tidy classification, so a large share lands in the residual category. That's a structural feature of how these firms describe their work, not a data error.

The concentration also explains why the trio's wage profiles skew toward the analyst and support tiers rather than the senior software-engineering bands you'd see at a product company. If you want to see how a single role's pay climbs across the DOL's four prevailing-wage levels, this ladder for Computer Systems Analysts shows the entry-to-senior spread.

Cognizant's certification numbers, read carefully

Cognizant offers a good case study in why raw percentages need context. In the disclosure data, Cognizant shows a certification rate of 53.6% with 0.00% denied.

At first glance those two figures seem to contradict each other — how can barely half of filings be "certified" if essentially none are "denied"? The answer is the certified-withdrawn effect. A large block of filings are certified by DOL and then withdrawn by the employer before use. Those withdrawals pull down the share counted as cleanly "certified" without adding to the "denied" bucket, because a withdrawal is an employer action, not a government rejection.

This is exactly why "certification rate" is a poor proxy for how selective or successful a sponsor is. A low headline rate driven by withdrawals says more about how a firm manages its filing pipeline — over-filing to preserve flexibility, then pulling back — than about visa outcomes. We walk through this mechanic in detail in Who Sponsors the Most H-1Bs — and Who Has the Cleanest Record, which is the right place to go before ranking sponsors on "clean" percentages.

How the trio differs from the stories you've already read

If you follow this site, you've seen two adjacent pieces. This one is deliberately different:

• Outsourcing Giants vs. Big Tech: The Two Faces of H-1B Hiring contrasts the services model with the product model — two categories of employer.
• Google vs. Infosys: Two H-1B Business Models puts one product company against one services firm.

Here, all three companies are on the same side of that divide. The interesting comparison isn't services-versus-product; it's the differences within the services tier — Infosys's outsized volume, the TCS/Cognizant near-tie, and how each firm's role concentration and filing-pipeline behavior show up in the data.

What a job-seeker should take from this

If you're weighing an offer from one of these firms, the filing data is a starting point, not a verdict. A few practical reads:

1. High volume means an established immigration pipeline, which can be reassuring for process familiarity — but it says nothing about your individual petition's odds, which USCIS decides.
2. A residual role code like "Computer Occupations, All Other" is common here; it doesn't signal a downgrade, but it does make it harder to benchmark your offer against a specific market wage. Compare against the nearest concrete SOC on the occupation pages.
3. Don't read a low certification percentage as rejection risk. As Cognizant's 0.00%-denied figure shows, withdrawals — not denials — drive those numbers.

You can pull any of these firms' full records directly: Infosys, Tata Consultancy Services, and Cognizant.

The bottom line

Infosys, TCS, and Cognizant built their US footprints on H-1B volume, but they are not interchangeable. Infosys files at a scale the other two don't approach; TCS and Cognizant are effectively neck-and-neck. Their work concentrates in computing occupations, much of it in a catch-all category. And their headline certification rates — Cognizant's especially — only make sense once you separate withdrawals from denials.

One last caution on any current-year figure: FY2026 is partial, reflecting only about two quarters of reported data, so year-to-date comparisons will understate full-year totals. Treat the cumulative history as the reliable picture, and the latest year as a work in progress.