Fix legacy operating models before they block your AI ambitions

Application modernization isn’t new; what’s new is that the future of the business now depends on it. The HFS Horizons: Legacy Application Modernization Services, 2025 research revealed that enterprises still burdened by mountains of technical debt will essentially be locked out of the next wave of AI-native value creation. Enterprise tech leaders must rethink their IT operating models to stay in the race for AI-driven market dominance.

Legacy is incompatible with the new way of the AI-driven world

Most enterprise tech leaders already know that we’ve passed the point where legacy is just a few clunky systems and processes or a hindrance to transformation. They are the silent corroders of AI’s return on investment. Every AI pilot that fails to scale can be traced back to brittle architectures, fragmented data, and project-centric delivery models that simply don’t align with the demands of real-time, AI-enabled decision making and orchestration. The resulting technical, process, data, and culture debt slows time-to-market and exacerbates compliance risks from outdated controls. Business cases don’t materialize, and AI is unable to scale beyond proofs of concept (POCs) because the tech architectures can’t support it (see Exhibit 1).

Exhibit 1: Legacy debt will continue to drain enterprise AI ambitions unless it is repaid

Source: HFS Research, 2025

Worse still, the subject matter experts who built the legacy systems are retiring, and the next generation is reluctant to work on them. Organizations are losing the knowledge they need to maintain and transform these systems. The explosive growth of AI in the enterprise makes it clear that the future of the business is AI, and good AI demands connected, governed, real-time systems with API-first architectures.

AI is raising expectations for speed, quality, explainability, and agility that legacy environments can’t meet. The message for enterprise tech leaders is clear: If your tech stack isn’t AI-ready, your transformation agenda is already behind.

Fix the operating model, and tech initiatives will fix themselves

Most enterprises are stuck not because they don’t know what to modernize, but because they’re modernizing the wrong way (see Exhibit 2).

They treat modernization as

• A cost center, rather than an investment in future capability.
• A one-time program, rather than a continuous life cycle.
• A tooling decision, not an operating model shift.

Exhibit 2: Unless you rethink how you buy and use tech, your AI ambitions will remain in POC purgatory

Sample: N=608 Global 2000 enterprises, 2025
Source: HFS Research, 2025

Future-forward enterprises are doing the opposite. They’re converging build and run, deploying digital guardrails, measuring outcomes against key performance indicators, and moving toward AI-infused product-aligned teams that ship faster and operate with confidence.

For instance, ING moved to cross-functional “squads and tribes” that own products end-to-end, resulting in faster time-to-market, higher engagement, and productivity. Goldman Sachs adopted a platform and site reliability engineering model with scalable guardrails and centralized observability, allowing product teams to own reliability while operating safely. In both these examples, we see a pivot away from treating modernization as a remediation exercise and elevating it to the center of AI roadmaps.

Without centralized modernization, your AI strategy is just an expensive science project

By centralizing modernization, enterprises can unlock value at scale from their AI deployments. HFS sees repeatable, enterprise-scale results when organizations adopt these foundational shifts:

• Embrace outcome-first modernization roadmaps anchored to business KPIs that treat modernization as a change in architecture and introduce a product-thinking approach that is foundational for AI readiness. Treat architecture as the North Star, with an emphasis on API-first, event-driven, composable platforms.
• Adopt AI-augmented software development life cycles and operations that use intelligent platforms for discovery, development, and test generation, AIOps, and defect prediction, improving coverage, consistency, and confidence of teams.
• Build platform guardrails and shared enablement to standardize build and run, thereby reducing variance, enforcing policy, and enabling developer self-service.
• Build partnered modernization factories that combine hyperscaler funding, curated refactoring toolchains, and ecosystem expertise to industrialize delivery, de-risk transformation, and reduce services drag.
• Build a continuous modernization pipeline, not “big bang” initiatives, to deliver continuous business value during the journey. The best enterprises treat modernization as a continuous product evolution, not an episodic event.
• Measure modernization success in business terms, such as
  • Velocity and quality, including lead time, change failure rate, escaped defects, and AI-assist penetration.
  • Business impact, including time-to-feature, revenue funnel improvements, and cost-to-serve.
  • Debt and risk retirement, including percentage of monolith decomposition, brittle integration removal, and AI control coverage.

The Bottom Line: The future belongs to enterprises that fix the foundation and redesign for continuous change. Anything else is just technical debt deferred.

Enterprises that break out of application modernization inertia do so by treating it not as migration, but as the first step in becoming AI-native.

• They use AI to accelerate modernization.
• They modernize to enable scalable, responsible AI.
• They shift from project-based code cleanup to product-based platform reinvention.

Transformation leaders who change how the enterprise builds and runs technology, rather than just swapping tools, separate themselves from the rest. Reframe application modernization and its role in your AI strategy to avoid getting stuck in the past.