CEO of Ascendion, Karthik Krishnamurthy has led significant business operations & transformation in software engineering for over 20 years.
getty
We are currently living through what I call the "Great AI Disconnect." Since 2022, enterprise tech spending has surged by 8% annually, yet labor productivity has flatlined at less than 2%. To the board, it looks like a bubble; to IT, it looks like a mess.
But the truth is more nuanced: Value is being created, but it’s being "absorbed" into legacy contracts, outdated rate cards and entrenched workflows. When work moves to software—AI agents and platforms—the economics of technology services shift: higher productivity, lower costs, faster time to value, lower risk and freed-up capital to drive growth.
While most leaders are hoping for a productivity miracle, a small group of pioneering firms are quietly finding real value.
This is not an incremental 5% productivity bump. We are starting to see a reshaping of our traditional ROI curve as a new economic model—what I call AI arbitrage—comes into play. Here are some new examples of what I've been able to accomplish with clients when leaders are focused on outcomes:
● An online bank needed to modernize a 40-year-old wealth management platform with no documentation and no available experts. Agentic AI reverse-engineered 700,000 lines of legacy code and delivered the full modernization in 18 months for $9 million.
● An integrated healthcare company had spent $10 million on a cloud data migration with nothing to show for it. AI-orchestrated workflows were able to complete the job in nine months for $9 million, with 85% automation and a 60% reduction in total cost of ownership.
● A global technology firm, which stalled mid-transformation, used a six-month AI pilot to reset delivery benchmarks, building the case for $500 million in savings over five years, 40% to 50% faster delivery and 30% to 35% better defect detection.
These results sound exceptional, and not long ago, they would have been impossible. But they're not impossible. I've seen results like this firsthand. While frontier firms capitalize on this AI arbitrage, the broader market seems stalled: MIT reports a 95% pilot failure rate, and PwC finds most CEOs have yet to see any financial return.
If AI is generating this much value, why isn't it showing up in your profit-and-loss statement? I'd argue it's because the economic model hasn’t caught up to the capability. That's the problem AI arbitrage solves by changing how value is priced, delivered and captured.
Based on what I've learned in daily client work over the past three years, leaders I've spoken to are at an inflection point. In order to succeed in this space, they'll need to embrace three critical tactics.
Thousands of global capability centers, GCCs, are operating today. These centers work as subsidiaries that perform business functions for parent companies, including IT, engineering and more. In the next 24 months, nearly a quarter of enterprises featured in a recent ISG report said they're planning to establish a GCC, and 40% are expanding their GCC initiatives.
Most leaders think of GCCs as cost-optimization engines. They are, but they're also an underutilized asset in the AI value equation.
GCCs concentrate on exactly the kind of work where AI delivers the most immediate return: process-oriented, documented, data-intensive, routinized business workflows. IT operations, financial processes, HR workflows and customer service all run through these centers at scale. The legal, procurement and risk frameworks are already in place. The talent and workflows are already there, ready to be disrupted.
This is why AI arbitrage is already working inside global capability centers. Companies capturing real value from AI today aren't all starting from scratch. They're embedding AI agents and agentic workflows into GCC operations where the processes, governance and commercial structures already exist. The result is measurable productivity gains, cost reduction and innovation flowing through established financial models, not pilot programs.
There's an old saying I know all too well: "If you want to hang a picture, you end up cleaning the garage." We've all been there. A simple task turns into three days of fixing, taping and oiling 30 other things before the picture actually makes it onto the wall. Those distractions probably needed doing, but they didn't need to happen first.
We're seeing the same pattern with AI. Leaders get stalled by prerequisite work, redoing contracts, rearchitecting tech stacks, modernizing decades-old data and solving other problems before capturing any near-term value. These are essential investments over time. But they are not preconditions for getting started.
You don't need picture-perfect data, processes or infrastructure to begin generating AI value. You need a high-impact starting point and a willingness to run AI alongside existing systems, rather than after a wholesale overhaul. The companies pulling ahead right now aren't the ones with the cleanest architecture. They're the ones that hung the picture first.
Find a partner, internally or externally, willing to commit to the impact AI delivers. Your initial target should be business outcomes in half the time at half the cost of solutions without AI. That's not always achievable, but it should be the standard you work toward. Pay for impact, not effort.
On the cost side, your AI arbitrage investment should include LLM infrastructure, trained engineers, data access and the “last-mile” platforms that align AI agents to your existing technologies and workflows. On the benefit side, demand capital unlock, velocity (a platform released 12 months ahead of plan has real economic value) and revenue growth through better software, de-risked targets and improved consumer experiences.
If you're getting heart surgery, you're not paying the surgeon for the hours they spend in the operating room. You're paying for decades of training, mentorship and practice that improve your personal outcome. The same logic applies to AI-powered technical operators. Don't box yourself out of upside value with rate cards designed for a pre-AI world.
Skeptics are right about this: the old way of buying tech is dead. If you apply 2026's AI to a 1990s commercial framework, you’ll keep getting a 2% return on a trillion-dollar investment. But by leveraging GCCs as high-velocity on-ramps and demanding a 50-50 value proposition from partners, you can stop "cleaning the garage" and start hitting the P&L.
The technology isn't going back in the box—it’s time we changed the box to fit the technology.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。