Marco Santos is Global CEO of GFT. I write about topics at the intersection of AI, financial services, manufacturing and leadership.
This is a scene that I’ve watched play out in enterprise boardrooms across four continents. A CIO delivers a compelling modernization proposal: Exit the mainframe, gain speed and scalability, and reduce infrastructure spend by 30% or more. The numbers are real, the case is sound, and the board often still says no because they don’t understand the risk of not modernizing.
For many enterprises, touching mission-critical systems that process countless daily transactions is a serious undertaking. So when it’s not clear that modernization is no longer optional—but instead, key to a company’s AI readiness and ability to innovate at the speed their competition already is—most boards will choose the certainty of what they have over unclear potential every time.
Most enterprises have been stuck in this very scenario for decades, but AI has now magnified the consequences of staying there. It has also collapsed the cost and timeline of doing so.
Based on what I’m seeing across global engagements, enterprises have 12 to 18 months to modernize before the market surpasses them. This window accounts for the mounting pressure on CEOs to deliver AI solutions, emerging regulations, the accelerating retirement of legacy tech talent and the compounding advantage that early movers are gaining right now. After these dominoes fall, organizations that haven’t moved won’t just be behind; they’ll be structurally unable to compete.
The Risk Has Changed
Risk used to be assessed by disproportionately weighing the cost and hassle of modernizing. Now, the biggest cost is opportunity and the risk that results from companies being locked out of what comes next (and what’s already here).
Multi-agent AI systems, real-time data pipelines and agentic automation aren’t future capabilities. They’re being deployed right now by organizations with modern, composable architectures. And none of them work on top of monolithic legacy systems, even those that have been partially modified or equipped with patchwork work-arounds. They require an entirely different infrastructure foundation.
Every quarter that a competitor spends building a modern infrastructure results in new AI capabilities that a legacy-bound organization cannot replicate by spending more money. The only way to close the gap, rather than compounding it, is to modernize first.
The 18-Month Window
While it might seem that every financial institution must have already deployed AI modernization projects to address these issues, there are still many that haven’t. That’s due in part to the fact that they haven’t caught up with the reality that AI has made it more affordable and faster to modernize than ever.
In my experience, programs that used to take five years and nine-figure budgets can now be delivered in a fraction of that time and cost. AI-powered discovery can map an entire legacy estate in weeks, while multi-agent AI systems can reverse-engineer business logic and accelerate productivity efforts by, in some documented cases, 40%.
There are four other immediate reasons why the window for catching up won’t stay open indefinitely.
1. CEOs know that their success depends on AI. According to a recent survey of 900 CEOs across eight countries, 77% say failure in AI can cost them their job, while 62% are under board pressure to deliver results. AI is no longer a cost-saving initiative.
2. The engineers who understand COBOL, VB6 and mainframe architectures are retiring. I believe many COBOL experts will retire in coming years. In fact, as far as I’ve seen, universities largely stopped teaching the language decades ago. With the pipeline of replacements essentially nonexistent, there’s even more pressure to modernize. Some argue that mainframes and banks are evolving together, but the reality is that major global banks are shifting from mainframe to AI-driven cloud infrastructures.
3. Modernization is becoming the prerequisite for operating legally in regulated markets. DORA, NIS2 and emerging AI governance require capabilities like real-time resilience reporting, digital auditability and AI risk controls that legacy systems simply were not designed to support.
4. The organizations that modernize first will gain a market advantage. First movers are already attracting the best AI talent, the strongest cloud partnerships and the most favorable market positions, creating a flywheel that late movers won’t be able to catch.
Modernization Is Foundational To Every AI Initiative
The organizations I’ve seen successfully fund and execute modernization are framing the case for doing so across multiple dimensions. Cost reduction is one of them, but now it’s paired with time savings, faster speed to market, risk reduction, and the productivity and revenue gains that follow. Together, these factors can make the case that modernization isn’t an infrastructure expense; it’s the precondition for every AI initiative on the company’s road map.
CIOs need to walk into the boardroom and say, “Modernization is what’s enabling our competitors to deploy new AI capabilities that we can’t, and every quarter we wait makes the gap wider.”
Rather than attempting a high-risk overhaul of core systems on day one, they can prove the model and build confidence by beginning with peripheral applications. This is different from bolting on work-arounds that don’t solve the underlying architectural limitation. It instead builds momentum toward the core transformation that ultimately unlocks AI readiness.
It’s also the approach behind almost every successful modernization program I’ve been involved with.
The Best Time To Begin Modernizing Was Five Years Ago: The Second-Best Time Is Right Now
Though generative AI only recently emerged for businesses, enterprises that modernized their legacy infrastructure to cloud infrastructure years ago are now far better positioned to introduce AI use cases than those stuck on dated systems. For those companies, the cost of inaction will show up when a competitor building on modern infrastructure launches AI-driven products at three times their speed, long before it shows up in their quarterly review.
Modernization is the foundation for how companies will innovate, compete and scale in the AI era. The CEO’s job is to laser-focus on this strategic reality. In 18 months, it will be too late for many organizations to catch up.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?