Current State of Enterprise AI: Buy versus Build

#2 in a Series  

The MIT study The GenAI Divide: State of AI in Business 2025 starkly lays out that 95% of AI implementations fail, despite $30–40 billion in enterprise investment. As the report states:  “Just 5% of integrated AI pilots are extracting millions in value, while the vast majority remain stuck with no measurable P&L impact. This divide does not seem to be driven by model quality or regulation, but seems to be determined by approach.”

What Differentiates the 5%?

In our last blog post, we outlined six strategic decisions that drive real ROI in AI. One of the most important of those decisions is Buy versus Build, a choice that often determines whether a pilot ever makes it across the divide into measurable success.

In the words of the report,
“Implementation advantage: external partnerships see twice the success rate of internal builds.”

The Justification for Building

There are instances where there are obvious justifications for building a system internally. Regulated industries often face extraordinary hurdles for security and compliance. In other cases, companies may want to control their own intellectual property. And let’s face it, many technical teams relish the challenge of building a custom AI system or tailoring one from a generalized foundation.

Decision Framework

In buy-versus-build decision-making, the McKinsey framework is often referenced. This framework includes four key points:

• Strategic alignment – Is the capability core to the company’s business?
  Does it make sense to build an AI for legal work if a company’s core business is hospitality?
• Cost Analysis – Beyond development cost, what is the total cost of ownership (TCO) over the long term?
  Do you have AI experts who can maintain the systems you build?
• Time-to-Value– Do the benefits of customization outweigh the speed of an off-the-shelf solution?
• Resources – Does the company have the internal expertise and infrastructure to develop a solution, maintain and enhance the solutions, especially as it may distract from core business functions? And especially given the rapid pace of AI. Vendor partnerships bring external expertise while freeing internal teams to focus on revenue-generating work.

Why Building Often Fails

The MIT study shows a clear pattern: internal builds often struggle. They are slower, more expensive, and more likely to stall before delivering ROI. According to the report, external partnerships succeed at roughly twice the rate of in-house efforts. And employees are twice as likely to use external resources as internally built ones.

Common reasons include:

• Solutions that don’t fit smoothly into day-to-day workflows because builders don’t have the vertical expertise.
• Long development cycles that drain resources before value is realized.

One study participant summed up the frustration this way:  “It’s excellent for brainstorming and first drafts, but it doesn’t retain knowledge. For high-stakes work, I need a system that accumulates knowledge and improves over time.”

Why Buying (or Partnering) Works Better 

By contrast, vendor solutions are typically designed with:

• Prebuilt integrations into common workflows and systems.
• Adaptations and enhancements that come from product evolution by expert vendors.
• Business-focused metrics that tie performance to ROI.

This pragmatic approach explains why the 5% that succeed often buy or partner first. It gives them speed, focus, and measurable impact, while freeing internal teams to concentrate on core business priorities.

A Balanced View

This is not to say that building never makes sense. If compliance demands it, if IP ownership is strategic, or if the company has the resources and patience to invest, then building may be the right choice. But those cases are exceptions.

For most organizations, partnering with a solution provider whose entire business is understanding and developing the best-in-class AI solution that delivers greater value, more rapidly, and with far less risk.

Conclusion 

The lesson from the MIT study is clear: the companies that cross the divide into the successful 5% make smart choices. They buy or partner when speed and results matter most. They build only when strategic necessity demands it.

The takeaway: buy first, build when you must. That simple choice may determine whether you stay stuck in the 95% or join the 5% extracting millions in value.

Coming Up Next 

In the next blog in this series, we will look at another theme from the MIT study: why back-office functions are proving to be one of the strongest early sources of ROI for AI. We will explore how the 5% are using AI in finance, HR, and operations to quietly capture value while others are still chasing headlines.