AI build vs. buy decisions are often driven by the wrong factors

Build-versus-buy decisions in AI have a profound influence on the long-term prospects of your business. In new research from Modus Create, 550 product and technology leaders share how these choices are playing out as AI moves deeper into real-world use. The findings reveal that many leaders are letting a short-term factor—the availability of internal skills—drive what should be a long-term strategic decision. Read the full report on AI in product development -->

One of the most important AI decisions that many organizations are making right now is also the one they often misjudge.

35% of enterprises say they are building proprietary AI solutions, 29% are buying off-the-shelf tools, and 32% are taking a blended approach.

​​On the surface, this looks like a thoughtful distribution based on different requirements. But when you look into how these decisions are actually made, the choices are often less intentional than they appear.

In our new research, we found that nearly half of executives make their build vs. buy decisions based on the availability of internal skills.

That means a decision that should be anchored in where the business needs to create advantage is often based on what the team is capable of today. Once the rationale shifts from strategy to convenience, AI adoption starts to move away from business goals.

What build, buy, and blend actually mean

Before we explore the reasons behind the decision, let’s define what build, buy, and blend AI actually mean. This is important because many organizations define these terms too vaguely.

1. Build: You own the intelligence

Building means the logic that underpins AI is defined by your models, training approach, and system design. In other words, if your implementation disappeared, the capability would not exist elsewhere in the same form.

Example: Training or heavily fine-tuning an open model on your internal data to power a fraud detection system or recommendation engine that behaves uniquely to your business.

2. Buy: You rely on external intelligence

Buying means you use an off-the-shelf AI solution that controls the model, its behavior, and its improvements. You might provide inputs, but have no role in shaping the quality of outputs beyond your own prompts.

Example: Sending customer queries to ChatGPT or Gemini APIs to help craft personalized support responses.

3. Blend: You direct the intelligence

In the case of blending AI, the model itself is not yours, but how it is used and its output are controlled by you.

Example: Using ChatGPT or Gemini as the language engine, but controlling retrieval (your internal documents), validation (business rules), and workflows (approvals, compliance checks) before showing any response to a user.

Before you decide how AI should be adopted, you need to be far more explicit about exactly what you should own, what you can rent, and where a hybrid model is justified by the problem itself rather than by internal ambiguity.

Internal skills should not guide your decision

Nearly half of executives cite the availability of internal skills as the primary driver of their build-versus-buy decisions. On one level, that is unsurprising. AI is evolving quickly, leadership wants results, and teams are under pressure to move from AI experimentation to execution. In that environment, it is natural to start with what the organization can realistically do now.

However, internal skills are a poor proxy for what the business should own. Skills evolve quickly. They can be hired for, developed, supplemented through partners, or deprioritized as the business changes.

On the other hand, strategic advantage is much harder to rework. Once you decide to own a capability that does not deserve ownership, it inherits cost, complexity, and maintenance that can persist for years. Similarly, once you outsource a capability that should have remained closer to the product’s core, you give up leverage that may be difficult to recover later.

Therefore, a build-versus-buy decision should begin with the business model and the degree of control your company needs. Internal skills should shape how you execute your choice, not the choice itself.

The illusion of “building”

When we saw the numbers in our research (35% build, 29% buy, and 32% blend), one thing immediately stood out. It’s highly unlikely that more organizations are building their own AI solutions than they are buying or even blending.

So what explains this? Not a flaw in the data, but how organizations interpret what it means to build.

In practice, what many organizations describe as “building AI” is not really building in the strategic sense, but blending/assembling. Teams are integrating models, tuning prompts, orchestrating workflows, and layering interfaces on top of third-party systems. That work can be useful and, in some cases, commercially important. But it is not the same as creating a differentiated AI capability that the business owns.

This distinction is easy to miss because the work still feels substantial. There is complex engineering involved. There may also be custom workflows, governance logic, evaluation layers, or specialized interfaces. But the core intelligence, including the model and infrastructure, often remains external.

This is why so many build-versus-buy conversations feel strangely inflated. Organizations often believe they are building a moat when they may only be improving access to a common utility. That gap between perception and reality can lead to overconfidence about control, differentiation, and defensibility.

The right lens to make your decision

If the availability of skills isn’t the right lens, how should you decide whether to build, buy, or blend AI solutions? While this is a complex question, a simple framework can help you make an informed choice.

• Build when the AI solution is closely tied to a competitive differentiation or operates in a high-sensitivity environment.
• Buy when time-to-market matters more than exclusivity, and the market can provide a reliable solution.
• Blend when the problem requires both scale and specificity. This is often the most sensible approach in enterprise settings, especially in regulated industries, because it allows companies to rely on mature external platforms while retaining control over the outputs.

AI will keep evolving, and today’s capabilities will quickly become table stakes tomorrow. What will endure is how deliberately you chose to position your business in that landscape. Get that wrong, and AI becomes another layer of complexity that appears to be progress but never quite delivers on it. But get that right, and AI becomes a meaningful driver of business performance.

This blog features findings from our latest report, AI in product development: A reality check, a comprehensive study of how 550 product and technology leaders are actually deploying AI in their organizations. Access the full report on AI in product development →

More insights from our research

This article is part of a series of findings from AI in product development: A reality check, Modus Create's study of 550 product and technology leaders. Explore the other articles in the series:

• Rushing AI, missing ROI: What's slowing product innovation
• AI maturity mirage: What leaders believe vs. what teams see
• AI is widening the product strategy/delivery divide
• AI has made modernization non-negotiable
• Shift governance left or slow AI down
• Pressure to prove ROI is rising
• AI is reshaping product teams