73% of enterprises have already launched AI, including generative AI, somewhere in their marketing organization. That is the headline figure from the Infosys CMO Radar survey, and it confirms what most operators already see in budget lines: AI in marketing has moved from pilot talk to deployed software.

The second number matters more. Only 52% of those deployments are achieving business value, according to the report. In other words, nearly half of the AI projects that enterprises have actually shipped inside marketing are not yet paying their way.

What do the numbers actually say?

The Infosys CMO Radar covers enterprise marketing organizations and their use of artificial intelligence, with generative AI explicitly included in scope. Two figures anchor the findings:

• 73% of enterprises have launched AI, including generative AI, across marketing functions.
• 52% of those deployments are achieving business value as measured by the study.

Read together, the numbers sketch a familiar adoption curve. Deployment is broad; returns are not. An enterprise can run dozens of AI-assisted workflows — copy generation, campaign segmentation, creative variants, audience scoring — and still struggle to tie them to revenue, cost reduction, or cycle-time gains that a CFO would sign off on.

Why does the 52% figure matter for operators?

For marketing and technology leaders, the gap between 73% and 52% is the working agenda. If roughly three-quarters of enterprises have AI live in marketing, then AI adoption itself is no longer a differentiator. The differentiator is execution quality: which deployments clear the business-value bar, and why the rest fall short.

The Infosys CMO Radar does not break out the failure reasons in its headline figures, but the framing is instructive. The study measures deployments against actual business value, not against experimentation or usage. That is a stricter test than most vendor case studies apply, and it is the test that determines whether budgets survive the next planning cycle.

Enterprises that have not yet deployed AI in marketing are now a minority — roughly one in four, if the 73% figure holds across the surveyed base. Late adopters face a different problem than early ones: the tooling is proven, the vendor market is crowded, and the internal expectations are already set by competitors who shipped first.

What should readers take from the report?

The Infosys CMO Radar is a status check, and the status is mixed. Adoption is effectively mainstream — 73% is no longer an emerging-technology number. Value capture is not. Just over half of deployed AI in marketing clears the business-value threshold.

For CMOs and CIOs planning 2025 budgets, the practical reading is straightforward:

• Count your own AI deployments in marketing, including generative AI use cases.
• Apply a business-value test to each one, the same way the study does.
• Expect roughly half to fail that test, based on the enterprise-wide pattern.
• Reallocate spend from the failing half toward the use cases that demonstrably perform.

That is less dramatic than the usual AI narrative, and more useful. The report's value is not in declaring AI a success or a bust in marketing. It is in putting hard percentages on both sides of the ledger: 73% deployed, 52% delivering. Enterprises that close that gap fastest will set the benchmark everyone else gets measured against.

Infosys published the CMO Radar findings as a survey-based report on enterprise marketing AI adoption. The figures cited here — 73% deployment and 52% business-value achievement — come directly from the study's headline results.