Gartner Warns: AI for Mainframe Migrations is a Bubble

The enthusiasm for applying artificial intelligence to complex mainframe migration processes may be destined to deflate. According to analyst firm Gartner, the use of AI to facilitate the transition of legacy code from these platforms to alternative systems is a genuine bubble, set to burst with significant consequences for businesses and vendors alike.

Gartner's forecast is stark: an impressive 70% of projects relying on AI for mainframe migrations are destined to fail. Moreover, 75% of vendors operating in this specific market segment could cease operations. This scenario paints a picture of profound disappointment for most mainframe users who are placing their hopes in AI as a quick solution to modernize their infrastructures.

The Complexities of Legacy Migration and AI's Role

Mainframe migrations have always represented one of the most arduous challenges in the enterprise IT landscape. These are robust systems, often developed with proprietary languages and monolithic architectures, managing critical workloads and sensitive data. The legacy code, accumulated over decades, is imbued with complex business logic and interdependencies that are difficult to untangle.

The appeal of AI, particularly Large Language Models (LLMs), lies in the promise of automating the analysis, translation, or rewriting of this code. However, an LLM's ability to fully comprehend the semantic context and functional implications of a mainframe application, often lacking updated documentation, is still limited. Simple syntactic translation does not guarantee functional fidelity or the desired performance on new platforms, whether on-premise or in the cloud. Managing data sovereignty and compliance, crucial for companies operating with mainframes, adds another layer of complexity that AI alone cannot magically solve.

Implications for TCO and Deployment Strategies

The failures predicted by Gartner will have a direct and significant impact on the Total Cost of Ownership (TCO) for businesses. Migration projects that fail to meet objectives result in additional costs for correction, redesign, or, in the worst case, a return to the original solution. This negates initial investments in AI tools and consulting services, increasing exposure to financial risk.

For organizations evaluating on-premise or hybrid deployment strategies, integrating AI solutions for mainframe migration presents further challenges. Often, AI tools require specific hardware infrastructures, such as high-performance GPUs, which may not be readily available or easily integrated into existing mainframe environments. This necessitates significant investments in new infrastructure, impacting CapEx and OpEx, and the need to manage new development and testing pipelines. The choice between a self-hosted deployment to maintain data control and reliance on cloud services for computing power becomes a critical trade-off, especially when the promise of AI does not materialize.

Future Outlook and Necessary Caution

Gartner's predictions underscore the need for an extremely cautious approach to adopting AI for mainframe migration. Companies should conduct thorough due diligence on vendors, evaluating not only the technical capabilities of their AI solutions but also their proven experience in complex migration projects. It is crucial to understand the current limitations of the technology and not fall into the trap of unrealistic expectations.

Instead of viewing AI as a magic wand, organizations should see it as a supporting tool, part of a broader, well-planned modernization strategy. For those evaluating on-premise deployment for AI/LLM workloads, AI-RADAR offers analytical frameworks on /llm-onpremise to assess the trade-offs between control, security, and costs. Caution, strategic planning, and a clear understanding of AI's capabilities and limitations will be crucial to avoid ending up among the 70% of projects destined for failure.