The promise AI has shown in pilot deployments has been impressive enough to grab the attention of engineers and executives alike. But there are also signs that this early success may be taking some executives’ eyes off the ball. As leaders demand measurable returns, the emphasis on optimized product design and longer-term innovation has fallen by a striking 23-point margin, putting the emphasis squarely on financial ROI. This marks a seismic shift in attitudes toward the technology’s promise. And the risk isn’t that ROI matters too much. It’s that a near-term payback mindset can quietly crowd out the transformational work required to sustain competitive advantage.

As customers and competitors make aggressive moves to implement AI, staying ahead means finding ways to embed AI into operations more comprehensively to reap its benefits at greater and greater scale. Given the cost of these types of investments, a near-term focus on financial returns could prove costly in the long run.

Capital investment and expectations are high

Many companies are willing to put substantial resources behind their AI strategies. But there’s a tension between the ambition driving this spending and the more incremental objectives toward which many companies have gravitated. Something fundamental needs to shift for companies to align that spending with larger-scale and longer-term ambitions.

At such a high level of spending, businesses will need a structural transformation to yield commensurate value. Embedding AI at scale will require both infrastructure investment and a redesign of the value chain. That’s one reason we are seeing the CFO function take on the new moniker of “Chief Future Officer” as CFOs make crucial decisions and investments to make this transformation possible.

The firms leading the way in this transformation aren’t just buying the latest AI models and looking for places to plug them in. Instead, they are making investments across their infrastructure to provide the secure data access and guardrails necessary to support embedded AI functions. Key areas of investment include cybersecurity, data privacy, process automation and optimization, and advanced data and analytics platforms. This is innovation, but not an overnight change. These ongoing cycles of infrastructure investment will pave the way for AI implementations that target processes throughout the organization, from the back office to industrial design to manufacturing and shipping.

Scaling AI must include people

Finding all the opportunities for AI transformation within the organization can’t be a top-down process. The people best equipped to identify areas for AI support and improved efficiency are the people currently leading those workflows. In many cases, we are seeing younger engineers who are more “digitally native” using agentic AI in ways that are teaching everyone new ways of working. Training, along with experimentation and sharing best practices, will be critical as organizations scale and evolve their AI strategies. Literacy alone isn’t enough. Companies must also systematize an approach to encourage employees to look for new use cases and separate essential human-led tasks from those ripe for AI automation, further supporting employee engagement and buy-in.

AI transformation is a marathon, not a sprint

The potential gains from AI are substantial, but they’re not going to be achievable if companies can’t deploy the technology at scale. Prioritizing short-term gains could create a long-term headache if it keeps a business from undertaking the infrastructure reset needed to take full advantage of AI across the enterprise. Companies that take a longer-term view and make the investments in resilient, measurable, at-scale infrastructure will set the stage for more and bigger opportunities over time.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.