Many of the challenges organizations face with AI today will feel familiar to anyone who lived through the early years of cloud computing, writes Stephanie Sanchez, Chief Strategy Officer at Grass Valley. The difference is that AI is moving faster, making it more important than ever to establish the right operating models before complexity takes hold.
We’ve been here before.
Not with artificial intelligence, of course, but with another technology wave that arrived with similar momentum, similar excitement and many of the same promises. Two decades ago, cloud computing changed how organizations built, deployed and consumed technology. It opened the door to new levels of agility, lowered barriers to innovation and fundamentally altered enterprise IT.
It also exposed weaknesses in how many organizations governed technology, managed costs and made architectural decisions. Looking at the pace of AI adoption today, I am reminded of those early cloud years.
Across almost every industry, organizations are experimenting with new models, testing new tools and identifying opportunities to improve productivity, automate processes and accelerate decision-making. In many cases, they are seeing real value. The speed of innovation is remarkable.
The challenge is that operating models rarely evolve as quickly as technology. During the early stages of cloud adoption, most conversations centred on what the technology made possible. Far fewer focused on governance, financial accountability or long-term flexibility. Those discussions came later, often after organizations had accumulated complexity, unexpected costs or dependencies that were difficult to unwind.
Cloud ultimately delivered on its promise. But many organizations spent years addressing issues that could have been avoided had they established the right frameworks earlier.
AI gives us an opportunity to approach things differently. The question is not whether organizations should adopt AI; that decision has largely been made. The question is whether we can apply the lessons learned from previous transformation cycles before complexity becomes embedded.
Prioritize The Desired Outcome
When people reflect on the cloud era, there is sometimes a tendency to suggest organizations moved too quickly. I don’t think that was the problem.
Innovation requires experimentation. Organizations that waited for complete certainty often found themselves at a disadvantage. The companies that benefited most from cloud were typically willing to learn by doing. The real challenge was that governance, accountability and visibility often arrived after adoption had already scaled.
The same pattern is emerging with AI. Employees are discovering new ways to work. Development teams are incorporating AI into software delivery processes, and business units are exploring use cases that would have seemed unrealistic only a few years ago.
None of this is inherently problematic. What matters is ensuring that operating models evolve alongside the technology.
The organizations that create the greatest value from AI will not necessarily be those deploying the most tools. They will be the ones that establish clear links between AI adoption and measurable business outcomes.
Before selecting a model, implementing a platform or launching a pilot program, organizations should ask a simple question: what problem are we trying to solve?
The objective might be improving employee productivity, accelerating software development, enhancing customer experiences or increasing operational efficiency. Whatever the goal, the technology should remain connected to a defined outcome.
Otherwise, experimentation can quickly become ‘activity without impact’.
Choice Disappears Faster Than You Think
One of the most valuable lessons from cloud transformation was understanding how difficult it can be to regain flexibility once systems, data and processes become tightly coupled to a particular platform.
The AI landscape is evolving at an extraordinary pace. Models continue to improve, new providers enter the market regularly, and capabilities that seemed differentiated six months ago quickly become commonplace. A model that performs best for software development may not be the best choice for content creation, summarization or analytical tasks.
This does not mean every organization should adopt a multi-model strategy from day one. It does mean that organizations should think carefully before creating unnecessary dependencies while the market is still evolving.
Standardization delivers benefits. Simplicity has value. But flexibility matters too. The ability to evaluate alternatives, adapt to changing economics and select the right model for a particular workload may prove increasingly important over the next few years.
Many organizations learned this lesson during their cloud journey. The same principle applies to AI.
Create Financial Visibility Before Scale
If there is one lesson every technology leader remembers from cloud adoption, it is that visibility matters.
Many organizations only developed mature approaches to cloud cost management after consumption had already scaled significantly. By then, the challenge was often remediation rather than optimization. AI introduces a similar risk.
A handful of pilot projects can quickly become hundreds of users, thousands of daily prompts and growing consumption across multiple business functions. Without visibility, it becomes difficult to understand not only what is being spent, but what value is being created.
This is where AI FinOps becomes important. The conversation should not be limited to cost control but should focus on understanding the relationship between investment and outcomes.
Are employees completing tasks faster? Are development cycles becoming shorter? Are customer experiences improving? Are business processes becoming more efficient?
The objective is not to spend less. It is to create the transparency required to make informed decisions about where AI is creating value and where it is not. Innovation works best when it is supported by accountability.
Why AI Is Different
The comparison with cloud is useful, but it only goes so far. Cloud transformed where technology operated. AI is beginning to influence how work itself is performed. That distinction matters.
Cloud adoption was largely driven by technology teams. AI is being adopted by employees across almost every function within an organization. It influences decision-making, creativity, customer interactions, software development and knowledge work.
The leadership challenge is therefore broader. Beyond technical governance alone, success requires education, change management, clear policies and a shared understanding of where AI fits within the organization.
Most importantly, it requires organizations to recognise that AI is not a standalone technology initiative but a business capability.
At Grass Valley, we view AI through that lens. We don’t deploy AI for its own sake. The goal is to improve a customer, employee or business outcome in a measurable way. That requires experimentation and it also requires discipline.
We need the freedom to test new ideas, but we also need visibility into costs, confidence in our governance frameworks and flexibility to adapt as the technology continues to evolve.
Moving Forward With Purpose
No organization will get every AI decision right; the technology is moving too quickly for that. What matters is creating an operating model that allows organizations to learn, adapt and scale successfully over time.
The cloud era taught us that governance introduced too late becomes remediation. It taught us that visibility becomes more difficult once complexity takes hold. It taught us that preserving flexibility early is often easier than recovering it later.
Those lessons remain relevant. We have seen this type of transformation before. The difference is that AI is moving faster than cloud ever did.
As leaders, the challenge before us is to apply those lessons sooner, balancing the speed required to innovate with the discipline needed to scale.
