The Ghost in the Machine: Why a 1998 Language Still Runs the World

AUSTIN, TX – August 27, 2026 – In an industry perpetually chasing the next disruption, the most significant challenge facing global corporations isn't a futuristic AI or a quantum leap in computing. It's a ghost. A digital ghost, born in 1998, that haunts the operational core of thousands of businesses worldwide. This ghost is Visual Basic 6, a programming language officially unsupported for nearly two decades, yet it stubbornly refuses to die.

A startling new report, “The State of Legacy Modernization Demand 2026,” published today by GAPVelocity AI, pulls back the curtain on this hidden crisis. Analyzing over 1,100 inquiries from organizations in more than 120 countries seeking to modernize their aging software, the report delivers a jaw-dropping statistic: an overwhelming 67.6% of this demand is driven by a need to escape applications built with Visual Basic 6 (VB6). This single, obsolete technology appeared 12 times more often than any other, a stark reminder that in the world of enterprise technology, the past is never truly past.

The Zombie Language That Won't Die

How did a programming environment, whose last major release coincided with the launch of the iMac G3, become so deeply entrenched in the global economy? The answer lies in its original success. VB6 was a titan of the Rapid Application Development (RAD) era, enabling a generation of developers—and even business users with a technical inclination—to quickly build custom, forms-based applications to solve specific departmental problems. These were not grand, centrally-planned IT projects; they were tactical solutions built in finance, operations, and manufacturing to get a job done.

These applications worked. For decades, they have reliably processed invoices, managed inventory, and run critical operational reports. This reliability created a powerful inertia. As one IT leader noted, the prevailing attitude has long been, “If it’s not broken, why pour millions into fixing it?” Microsoft itself has played a role, maintaining a commitment to ensure VB6 runtime files continue to work on modern Windows operating systems, including Windows 11. This allowed the applications to function, even as the development environment itself withered on the vine.

The result is a massive, distributed portfolio of technical debt. Decades of crucial business logic, operational exceptions, and undocumented domain knowledge are now trapped inside these aging codebases. The people who wrote them are often long retired, leaving a dangerous knowledge gap.

Counting Code, Not Just Companies

The report from GAPVelocity AI, the AI modernization arm of Growth Acceleration Partners, provides a uniquely pragmatic lens by using first-party demand data rather than abstract surveys. This grounds the analysis in the reality of what companies are actively trying to fix. And that reality is more complex than a single headline figure.

While VB6 dominates the number of companies seeking help, the picture changes when you measure the size of the problem. The report analyzes a second dataset of 467 scoped projects, revealing that the median VB6 application was around 124,000 lines of code. In contrast, applications built with other legacy tools like PowerBuilder and Clarion were eight to ten times larger, with median codebases exceeding a million lines. When weighted by this code volume, VB6’s share of the problem drops from nearly 70% to a more modest 30%, while PowerBuilder and Clarion swell to represent a combined 32%.

“Counting organizations alone understates the largest enterprise systems, many of which were built outside IT,” said Darryl Worsham, General Manager of GAPVelocity AI, in the press release. His point is critical: the modernization challenge isn't uniform. It consists of a vast number of smaller, widespread VB6 applications and a significant number of monolithic enterprise systems built with other legacy tools. The report also highlights the global nature of this issue, with 58% of inquiries originating outside the United States, confirming that legacy software is a worldwide economic anchor.

The Mounting Costs of Inaction

For years, deferring modernization was a viable, if risky, financial strategy. That era is definitively over. The cost of inaction is no longer a future concern; it’s an active drain on resources and a growing source of catastrophic risk. Security is paramount. Unsupported software receives no security patches, leaving it wide open to known vulnerabilities that cybercriminals actively exploit. A single breach enabled by a flaw in a legacy COM component could cost a company millions in damages, fines, and reputational harm.

Beyond security, there is the strategic drag. Industry analysts estimate that 60-80% of IT budgets are now consumed by simply “keeping the lights on,” maintaining these fragile systems. This leaves scant resources for innovation, digital transformation, or adopting technologies like AI that competitors are using to gain an edge.

Then there is the talent crisis. The pool of developers with VB6 expertise is shrinking rapidly. Companies are finding it increasingly difficult and expensive to maintain these systems, creating a precarious operational risk. This technical stagnation prevents integration with modern cloud platforms, APIs, and mobile solutions, effectively walling off critical parts of the business from the modern digital ecosystem.

AI Modernization: Savior or Hype?

Into this landscape enters AI, promoted as the silver bullet for the modernization crisis. Platforms like GAPVelocity AI's VELO, which uses a system of coordinated AI “agents” built on Microsoft Foundry, promise to automate the painstaking process of code conversion. These agentic systems are designed to analyze a codebase, architect a modern structure, translate the logic, and verify the output, promising to reduce modernization effort by up to 40%.

However, as the report itself wisely cautions, AI modernization claims should be tested against the messy reality of these legacy stacks, not against simplified demo targets. The challenges are formidable. The proprietary user-interface frameworks and limited training data for these older languages challenge even the most advanced Large Language Models (LLMs). There is no standard mapping from a 1998 client-server architecture to a modern, cloud-native one.

AI tools can “hallucinate,” producing code that looks correct but contains subtle errors that can cause chaos in production. This introduces a “hallucination tax”—the cost of having human experts find and fix these AI-generated bugs. The consensus among pragmatic strategists is that a “human-in-the-loop” approach is essential. AI is a powerful accelerator, a tool that can handle the 80% of repetitive, predictable work, but human engineers are still required to handle the complex 20%, owning the architecture, verifying the business logic, and ensuring the final product is secure and maintainable.

The GAPVelocity AI report doesn't just highlight a problem; it provides a benchmark for the solution. It reveals the true nature of the enemy: a vast, fragmented, and deeply embedded layer of legacy technology. Confronting these digital ghosts is no longer optional. It is the defining strategic challenge for any organization that wishes to remain secure, agile, and competitive in the coming decade.