Summary: On April 22, 2026, OpenAI announced a partnership with Infosys to embed Codex in the Topaz AI platform, beginning with software engineering, legacy modernization, and DevOps. No financial terms were disclosed, and the deal extends OpenAI’s Codex Labs channel strategy through systems integrators across more than 60 countries. Adoption will still depend on enterprise code governance, security review, and existing toolchains rather than distribution alone.
OpenAI broke the news on April 22, 2026: it is teaming up with Infosys, the Indian IT services heavyweight, and plugging its tools into Infosys’s Topaz AI platform. Software engineering, legacy modernization, and DevOps come first, and the pitch to enterprise clients is helping them move “from experimentation to large-scale deployment.” No financial terms were disclosed.
Treat this as more than another licensing announcement. Enterprise software has an old playbook for accounts the in-house sales force cannot crack: hand distribution to systems integrators and their delivery networks, and that is exactly the script OpenAI is following. Seen from Infosys’s side of the table, the deal is a bid to prove it remains indispensable at a moment when generative AI has the entire Indian IT sector anxious — anxious that the technology these firms are hired to deploy will also automate away the outsourcing work that pays their bills.
What the Deal Actually Includes: Codex Embedded in Topaz
Mechanically, Topaz will carry OpenAI’s tools — the coding assistant Codex is the one product reports name outright — with Infosys’s delivery capabilities across more than 60 countries attached, so clients can overhaul how software gets built, automate their workflows, and stand up AI systems at scale. The opening areas — software engineering and legacy modernization — are a tell: this is the volume work Indian IT majors already perform in bulk for Fortune 500 clients, which lets Codex slot into engagements that already exist rather than a new budget line a customer has to be talked into opening.
One scale figure deserves a bookmark: on X, Sam Altman noted that Codex has cleared 4 million weekly active users.
Codex Labs: The Larger Play Announced That Same Week
The Infosys deal is one piece of Codex Labs, the program OpenAI unveiled that same week. Its design sends OpenAI engineers straight into client engagements to help get the tools deployed; the first partner roster reads like a who’s-who of systems integration — Accenture, Capgemini, CGI, Cognizant, Infosys, PwC, and Tata Consultancy Services.
Read it as channel design, plain and simple: instead of calling on enterprise customers door by door, OpenAI converts seven major SIs into a combined resale and implementation network. This follows an earlier, similar enterprise agreement OpenAI signed with HCLTech.
Where Infosys Stands: AI Revenue Climbing, Shares Sinking
Infosys has every reason to lean in hard. The upside: AI-related services revenue reached 25 billion rupees (about $267 million) in the December quarter, some 5.5% of total revenue — proof that AI work is a genuine revenue line already, not slideware. The downside: the stock has lost more than 22% this year, and sector coverage shows HCLTech, Infosys, and TCS all selling off sharply in tandem, squeezed by clients reining in spending, macroeconomic turmoil, and the dread that generative AI will simply eat traditional outsourcing demand.
The backdrop counts as well: as recently as February 2026, Infosys signed with Anthropic to build enterprise-grade AI agents. Spreading bets across multiple model camps at once is the standard posture for IT services firms in this cycle — none of them wants its fate chained to a single model vendor, and none of the model vendors wants to be shut out of the SI channel either.
The Channel Play: Why Model Makers Ride on Integrators
What OpenAI is buying here is reach, not technology. When enterprises stall on adopting AI, the API is never the bottleneck; integration, data governance, and process rework are — the daily bread of the SI business. Codex Labs stations OpenAI engineers on the front line, so vendor presence lowers adoption friction while OpenAI tooling gets wired into the delivery pipelines of hundreds of large enterprises as the default.
Takeaways for Enterprise Buyers
Three concrete consequences. One, picking models is turning into a decision made at the SI level: the camps your outsourcing partners have backed will go a long way toward deciding which tools land inside your organization. Two, the shift from AI experimentation to large-scale deployment is the defining enterprise theme of 2026 — our opening-of-year outlook drew exactly that line — and SI networks are now speeding up its arrival. Three, developers should not read Codex’s arrival in the enterprise as a foregone conclusion: 4 million weekly active users is a strong number, but code governance, security review, and the toolchains already installed inside each company are where these deployments will actually be won or lost.