Can AI help solve the COBOL code crisis?

OpenAI said Monday that its latest AI models have made “substantial improvements” in migrating aging software written in older languages, including COBOL, to modern code. The company says this could help organizations maintain and modernize legacy systems used by institutions such as the IRS, banks and airlines; the article provides no quantified performance results or financial impact.

Analysis

The investable question is not whether AI can translate legacy code, but whether it can reduce the cost and risk of validating a production-ready migration. Hidden business rules, undocumented dependencies, regulatory controls, and testing remain bottlenecks; a working code conversion demo does not establish enterprise adoption or savings. OpenAI’s claim is not yet evidence of measurable customer economics.

If validation costs fall, banks, airlines, and public agencies could bring forward modernization budgets and reduce exposure to scarce legacy expertise. The second-order effect for services firms is mixed: Accenture, Capgemini, Cognizant, and DXC could win migration and integration work, while recurring maintenance revenue faces pressure if projects become shorter or require fewer labor hours. IBM is also exposed on both sides through its legacy-system ecosystem and modernization capabilities; treating it as a simple loser would be premature.

Days: the announcement may support a modest AI/software sentiment bid, but it does not justify repricing legacy-services earnings absent customer evidence. Over 1–3 months, watch for named deployments, independently verifiable migration-time reductions, defect rates, and contract economics. Over 6–18 months, sustained adoption could shift spending from maintenance toward testing, integration, cloud, and security—but regulated production systems are likely to migrate incrementally.

Contrarian view: the market may overvalue code generation and undervalue assurance, governance, and liability. The thesis strengthens only if customers demonstrate safe production deployment, not merely faster code conversion.

Key Decisions for Investors

• No immediate directional trade on the announcement alone. Treat the AI-modernization theme as a catalyst watch until customer deployments and realized project economics are disclosed.
• Monitor Accenture, Capgemini, Cognizant, DXC, and IBM for modernization bookings alongside legacy-maintenance revenue, delivery margins, and headcount intensity. Services firms could capture project demand while still facing pressure on labor-based recurring work.
• Potential relative-value setup: consider favoring firms with measurable testing, integration, and modernization exposure over labor-heavy maintenance exposure only after evidence of production adoption; do not infer that any named firm has already won work.
• Falsifiers: no credible regulated-sector deployments over the next 1–3 months; migration defect or remediation rates that erase labor savings; or services-company guidance showing no modernization demand and continued maintenance pricing strength.