When I work with executives, leadership teams, and AI tasks forces on their AI journey and stratgey, sooner or later a key question comes up: should they build AI solutions in-house, purchase them from vendors, or adopt a hybrid approach?A recent panel discussion featuring leaders from Dell, A.Team, and Salesforce offers valuable lessons for navigating this decision - and the answer, increasingly, is "yes to both."
When asked directly where he stands on build versus buy, Chris Bennett, Dell's CTO of AI and Data Solutions, responded simply: "I am on the yes." This isn't fence-sitting - it's recognition that the question itself has evolved.
The traditional framing assumes a binary choice. But AI capabilities exist across multiple layers, each requiring different strategic approaches:
Infrastructure Layer: Unless you're Nvidia or a hyperscaler, you're buying here. As Raphael Ouzan from A.Team noted, "You're not assembling your computer to run your website." Semiconductors, cloud compute, GPU clusters - these are commoditized building blocks.
Platform Layer: Enterprise AI platforms, foundational models, and governance frameworks. Phil Mui from Salesforce made a compelling point: building a true "system-two reasoner" - the deliberative, accurate reasoning that enterprise applications require - is extraordinarily difficult. When accuracy, security, and compliance matter, proven platforms often deliver what custom builds struggle to achieve.
Application Layer: This is where differentiation lives. Domain-specific workflows, proprietary data integration, and the "last mile" of customer value. Ouzan calls this "the plethora of building blocks that are being created literally every minute" that enable you to focus your build efforts on what actually brings value to your stakeholders.
Here's the framework shift that caught my attention. Ouzan articulated what I believe will define AI budget allocation in the coming years:
The Old Model (Past Decade):
80% of budget → Buying software licenses
20% of budget → Building customizations on top
The New Model (AI Era):
20% of budget → Foundational AI building blocks (getting cheaper daily)
80% of budget → Building the features that differentiate your company
If you're still allocating budget like it's 2015, you're probably over-investing in commoditized capabilities and under-investing in differentiation.
There is a pattern I've seen repeatedly with clients:
Company buys AI tools (copilots, assistants, enterprise licenses)
Productivity improves incrementally (faster emails, better summaries)
Anxiety sets in - competitors are doing the exact same thing
Realization: commoditized AI doesn't create differentiation
Ouzan described this journey precisely: "There's a little bit of anxiety mixed with excitement, then the 'let's buy a bunch of tools because we got to get on the AI train.' You roll out those tools and realize the improvement is only incremental - and everyone else has them too."
You need to recognize that "buy for basics, build for differentiation" requires actually building - which means allocating real resources to the capabilities that create your moat.
Use these questions to categorize each AI initiative:
Does this capability provide competitive advantage? If everyone will have it within 12 months, buy the best available and move on. If it encodes your unique value proposition, build.
Does your proprietary data create value here? "Bring AI to your data, don't bring data to AI." If your data is your moat, you need to build capabilities that leverage it - not export it to generic platforms.
What's your time-to-value requirement? When speed matters and proven solutions exist, buying enables rapid experimentation. When you need deep customization, allocate time to build.
Do you have the talent to execute? The "last mile" talent are the product managers, designers, data scientists who can translate domain expertise into AI-powered products. Assess whether your organization possesses this capability, and whether investing in that talent is feasible given your timeline.
Is accuracy and compliance critical? In healthcare, finance, and other regulated industries, enterprise platforms with built-in governance may outperform custom builds. Building that yourself is harder than most assume.
The most practical approach combines three modes:
Buy: Infrastructure and foundational capabilities where you have no differentiation. Platforms that handle "undifferentiated heavy lifting."
Assemble: Combining building blocks - open APIs, partner integrations, modular services - to create workflows specific to your business. This is where the "Lego" metaphor applies.
Build: Custom applications that encode your unique domain expertise and data. This is your investment in differentiation.
Map your current and planned AI initiatives into these three categories. Then examine your budget allocation: if you're spending 80% on buying and 20% on building, you may be investing in parity rather than advantage.
The journey to leveraging AI for business value is nuanced. A thoughtful consideration of when to build, when to buy, and how to blend both strategies is essential for fostering innovation and maintaining competitiveness. Navigate this territory with a holistic vision - ensuring AI not only enhances operational efficiencies but also drives transformative growth.
For a deeper dive into these strategic considerations, watch the Insight Partners panel "Build vs Buy: Considerations for a Strategic Approach to Innovating with AI" featuring executives from Dell, Salesforce, and A.Team discussing how enterprises navigate these decisions in practice.: