Back to BlogTechnology
Gary Brach
2/25/2026
8 min read

Anthropic Says Claude Code Can Read COBOL. Here's Why Reading Isn't Understanding.

Anthropic's February 2026 announcement sent IBM's stock sliding and the enterprise world buzzing. Claude Code can 'read and understand' COBOL, they said. We've spent years building a system that actually does. Let us explain the difference.

On February 23, 2026, Anthropic published a sweeping claim: Claude Code is "exceptionally proficient at deciphering and modernizing legacy languages like COBOL." IBM's stock took a hit. The enterprise media cycle went into orbit.

So let's talk about what Anthropic actually announced — what it means, what it doesn't mean, and why the distinction matters enormously for enterprises with billions of dollars of business logic locked inside their mainframes.

What Anthropic Is Claiming

Claude Code, Anthropic's agentic coding assistant, has been positioned as a tool that can automate the "exploration and analysis phases of COBOL modernization." Specifically, Anthropic and its partners (including Deloitte in a November 2025 webinar) have claimed the tool can:

  • Map dependencies across thousands of lines of code
  • Document "long-forgotten workflows"
  • Translate COBOL logic to modern languages
  • Reduce modernization timelines from years to quarters
  • These are not small claims. And to be clear: we respect Anthropic enormously. Claude is a remarkable language model. But there is a profound difference between a language model that has read COBOL and a purpose-built system that truly understands it. That difference is not academic — it is the difference between a project that succeeds and one that fails catastrophically in production.

    The Probabilistic Problem

    Every large language model — Claude included — is fundamentally a probabilistic system. It produces outputs that are likely to be correct based on patterns learned from training data. This works remarkably well for general programming tasks, documentation generation, and even surface-level code reading.

    Modernizing a COBOL transaction processing system that clears billions of dollars annually is not a general programming task.

    A typical Fortune 500 mainframe application contains:

  • Millions of lines of COBOL spanning 40+ years of ad-hoc modification with complex memory implications
  • Hundreds of copybooks with implicit data contracts
  • GOTO-heavy control flow that creates "spaghetti" logic, requiring careful formal analysis to trace
  • Business rules embedded in compute logic, often with no corresponding documentation
  • JCL job streams, DB2 binds, VSAM file structures, and CICS transactions form a dependency web that no single developer holds in their head
  • When an LLM "reads" this code, it pattern-matches against its training distribution. It will produce output that looks correct. It may be correct 90% of the time. But for mission-critical banking, insurance, and government systems that process trillions of dollars annually, a 10% error rate in understanding is not a rounding error. It is a catastrophe waiting to happen.

    IBM itself noted it plainly in response to the Anthropic announcement: "merely translating COBOL code does not fully address the complex architectural and integration challenges of mission-critical mainframe workloads."

    What Understanding Actually Requires

    COBOL Colleague was built on a fundamentally different premise: that true understanding of legacy code requires deterministic analysis, not probabilistic inference.

    Our approach constructs a mathematically-verified knowledge graph of your application. Every relationship in that graph is derived from formal analysis of your actual source code — not from a model's training data, not from statistical likelihood, but from formal analysis and graph traversal over your specific system.

    This means:

  • Every data dependency is traced, not inferred. We know exactly which fields flow into which computations, across every copybook boundary and FILE-CONTROL entry.
  • Every causal path is verified, not guessed. Impact analysis tells you the exact blast radius of a change — not a probable one.
  • Every business rule is grounded in actual source code and linked bi-directionally, so a business analyst's question gets an answer traceable back to a specific line of COBOL.
  • Confidence is 100%, or it is nothing. We do not produce answers that are "likely correct." We produce answers that are provably correct, or we tell you we don't know.
  • The Real Risk of "Good Enough"

    Perhaps the most dangerous outcome of the Anthropic announcement is not the technology itself — it's the expectation it may set.

    Enterprise buyers are already under enormous pressure to modernize. Boards and CFOs are excited by the prospect of AI shortcutting a process that has historically been expensive, slow, and risky. An AI that claims to "understand COBOL in quarters, not years" is exactly what they want to hear.

    But when that AI produces a translation that is subtly wrong — when the edge case in the premium calculation logic is misread, when the copybook inheritance chain is incompletely traced, when COBOL's difficult handling of memory is exhaustively understood, when the GOTO that looked like dead code turns out to control a critical batch job — the consequences are not a bad code review. They are regulatory violations, financial system failures, and data loss events that take years to remediate.

    We have seen this pattern before. In the 2010s, a generation of COBOL "automated translation" tools promised the same thing with rule-based compilers. The projects they launched still litter the enterprise landscape as cautionary tales.

    Where Claude Code Fits (Honestly)

    We don't believe the right answer is to dismiss Claude Code entirely. Agentic coding tools are genuinely useful for specific phases of the modernization lifecycle:

  • Performing "Code Archaeology": Analyzing program flows and mapping data dependencies to understand legacy subroutines.
  • Extracting Business Logic: Reading through legacy programs to document hidden rules and verified facts that have been lost to staff turnover.
  • Accelerating Test Creation: Analyzing existing code to understand expected behaviors and edge cases, ensuring functional equivalence after migration.
  • Generating Documentation: Translating complex, undocumented architectural patterns into fluent prose for non-technical stakeholders.
  • COBOL Colleague uses LLM capabilities — specifically for natural language generation — to surface the knowledge our graph has already deterministically established. The LLM translates our verified facts into fluent prose. While Claude Code struggles to present the facts, it excels at narrating them.

    That is the appropriate role for a probabilistic model in a zero-hallucination system: acting as the fluent voice of a deterministic mind.

    What This Moment Means

    Anthropic's announcement — and IBM's stock reaction — confirm what we have believed for years: the market now recognizes that AI-driven COBOL modernization is not only possible but also urgent. That is genuinely good news for the industry and for the enterprises that have been waiting for a credible path forward.

    Our job, as the company that has actually built the right solution for this problem, is to make sure buyers understand what they're evaluating. The question is not "Can AI read COBOL?" Claude Code has demonstrated it can. The question is: "Can you stake your core banking system on a probabilistic reading?"

    Phase Change is dedicated to providing the essential logical verification layer that anchors the high-velocity, probabilistic outputs of Generative AI agents in deterministic fact.

    COBOL Colleague is available for an enterprise demonstration. If your organization is evaluating COBOL modernization approaches in light of recent announcements, we welcome the conversation. Contact us to schedule a technical briefing.

    Share

    Related Articles