Former Intel CIO Motti Finkelstein On How AI Is Changing The Build Or Buy Equation
As AI accelerates market change, CIOs must shift from one-time procurement choices to a continuous discipline of managing optionality. Which layer of your stack do you own to stay competitive?
"Everything built for competitive advantage eventually becomes a commodity. The question is whether a vendor or your company controls that timeline."
Recently, I’ve noticed that my conversations with fellow CIOs have drastically changed. The general consensus is that AI is moving too fast to work in the ways we used to. In the past, the executive suite operated within a manageable three-to-five-year roadmap. As a former Fortune 500 CIO, I helped architect these very structures, prioritizing predictable execution and stable enterprise capabilities. That era of predictability is now over.
Today, we’re operating in a state of what Axios CEO Jim VandeHei aptly calls post-processable velocity. AI-driven innovation cycles move faster than traditional organizations can process the change. The work of a CIO in this era has become a discipline of managing organizational optionality. Our role has shifted from technology provider to capability orchestrator.
Take the classic build vs. buy debate. This decision is no longer a one-time procurement task. We now need to view it as the continuous discipline of owning our enterprise’s future.
Generic vs specific problems
In the legacy IT model of the previous era, the Total Cost of Ownership (TCO) was everyone’s north star. We bought software because specialized vendors could amortize development costs across thousands of customers. That logic still holds for generic, already-solved problems like infrastructure, standard CRM workflows, and market data ingestion. Building those in-house is a waste of engineering talent you need somewhere else. But the TCO model breaks down when the market moves faster than either vendors or your own teams can adapt. Whether you buy or build, you'll end up paying more if you treat the solution as a one-time event rather than an ongoing commitment.
At Intel, our enterprise architecture team eliminated 321 in-house applications and 1,400 commercial products, avoiding $13 million in costs. Most of these were built years earlier, many of them long before I arrived, and the choices behind them were sound at the time. The trouble was that leadership had stopped revisiting these decisions as market conditions changed, turning them into liabilities.
In the same period, we moved in the opposite direction on employee IT support. AskIT had been a commercial product. We rebuilt it on our own AI platform. It now handles 26,000 conversations for 11,000 employees and has cut “How do I…” tickets by 30%. Two different moves during the same time period. That is why build vs. buy needs to be seen as a continuous discipline rather than a single decision.
However, when we move from generic to specific problems, especially those that involve your unique proprietary data and competitive advantage, the math changes. The modern CISO can’t wait around for a quarterly vendor patch to close an AI-driven vulnerability. At Capital One, that necessity for speed produced VulnHunter, a set of AI agents that hunt for exploits directly in source code. It is paired with what the team calls a falsification engine, a reasoning layer built to disprove the agents’ own findings, so only what survives that internal challenge reaches a developer for remediation.
Capital One built because no vendor could move at the speed their threat demanded. Then, rather than wait for the capability to be commoditized around them, they commoditized it themselves and kept the position that comes with authoring the standard.
The governance hub from Intel Forge to the modern stack
When we moved to a federated model at Intel, we had a major governance problem. Platform users went from 500 to 1,600 in a single month. Employees were not waiting for permission to use generative AI, and nothing on the market governed federated AI workflows at the speed we were growing. Our solution was building Intel Forge, the backend engine that ran and connected the models and data sources, and Intel GPT, the no-code layer on top. Security and privacy were part of the approval path from the start. With this, any business unit could build what it needed inside approved guardrails.
By 2025, the Intel Forge platform carried 30,000 monthly users and more than 7,000 employee-built assistants, with 274 AI use cases in production out of 436 tracked.
This choice was all about controlling what “safe to try” meant, and because the market was still deciding what the category was, we had to build it ourselves. Everything built for competitive advantage eventually becomes a commodity. The question is whether a vendor or your company controls that timeline. Today you would buy a solution like Intel Forge rather than build it. At the time, building was the right call, because the market had not caught up.
The line between build and buy is moving. Your job as a CIO is to buy the stuff that’s stable like your infrastructure and core platforms, and own the stuff that differentiates you. Everything else is just work that keeps you busy without making the enterprise better or smarter.
Constructing the decoupled, three-layer stack
The key to modern enterprise architecture is decoupling. You must ensure that buying software does not mean surrendering your strategic neutrality. If your business logic is locked inside a vendor's system, you’re beholden to them. You need to own the parts that matter.
Your stack has three layers. First is your core platform layer. This is where vendors have built real advantages. You're probably not upgrading SAP with AI. Instead, you build the connectors on top. The integration and orchestration layer is where you keep your flexibility. A lot of vendors want to lock you into a long term relationship with them. They want to own everything by building it all into their platform. This is limiting, and it compromises your optionality and strategic neutrality. You can’t afford to be locked in today with the speed of change. The third layer is your custom workflows, aka how your company actually operates. AI made this cheap enough to build so there is no need to wait for a vendor to build this for you, it changes too fast.
For each layer, ask yourself: Does this choice keep us agile and competitive? This is why build-vs.-buy decisions are now a continuous conversation, with decisions made often and layer by layer.
Making it continuous: who decides, and how often
At Intel, we ran a joint technology council: a standing monthly forum — not a quarterly build-vs-buy review — in which the question Would we make the same call today? was unavoidable because it was embedded in every assessment. In attendance were the CIO, the CISO, infrastructure and security leads, and the business relationship managers who knew what each unit was trying to build. We would rotate through one business area at a time rather than attempting to review everything at once.
Significantly, nothing on the agenda of our standing monthly forum was labeled “build vs. buy.” Proposals were assessed based on implementation cost, redundancy with what already existed, expected usage, and value. If a capability duplicated something already in the portfolio, the answer was usually to buy or to use what we had. If it was genuinely differentiating and nothing in the market fit, we built it.
The forcing function most organizations are missing is renewal. Every contract coming up for renewal or major upgrade is a free opportunity to ask whether you would make the same call today. Most of the time, the answer is yes, but when the answer is no, you have caught it before committing to another three years.
The executive audit: Three pressure-test questions
Before approving any major technology investment, answer me these questions three:
- Is this capability core to our competitive differentiation?
If yes, build and own the logic. Differentiation handed to a vendor is eventually sold to your competitors. - Is this still the best call today based on current market velocity?
The "half-life" of technology decisions has shrunk. Be prepared to pivot if the "buy" has become a commodity. - Does this purchase lock us into a vendor’s roadmap at the expense of our own intelligence layer?
Ensure you own the data and the decision rights.
The velocity of AI-driven innovation is a competitive weapon for those who avoid getting locked in and architect optionality instead.
Motti Finkelstein is the Founder, Chairman, and CEO of EMDO Inc, leading corporate-wide investment, advisory, incubation, and board engagement activities across global affiliates, focused on supporting next-generation market leaders through operational expertise, strategic guidance, and long-term value creation. He previously served as Corporate Vice President and CIO at Intel Corporation, where he led global IT through the company's IDM 2.0 transformation and sponsored Intel Forge, a centralized GenAI governance platform. Earlier in his career, Motti was Chief Technology Officer for Capital Markets at BMO, a Senior Advisor at McKinsey, and held senior technology roles at Citi. Motti is a 2026 CIO Hall of Fame inductee, a 2026 National ORBIE Super Global CIO finalist, a 2024 NYCIO ORBIE Super Global Award winner, and a Forbes Technology Council contributor.
If this caught your attention, that’s not accidental.
The best editorial systems don’t happen by accident. Outlever builds them.
