Why CIOs are consolidating their EA stack in 2026 (and what it means for you)

If you work in a large organization, chances are your CIO is currently staring at their stack of architecture and governance tools, wondering "wait, how many of these do we actually run?". Spoiler: a lot. Here is why tool sprawl has become the conversation of 2026, and why enterprise architecture is quietly moving back to the center of the board.

First, a number worth sitting with

In early 2026, Torii published its SaaS Benchmark Annual Report. One data point made the rounds on LinkedIn: the average large enterprise (4,000+ employees) now runs 2,191 applications. Yes, you read that right. Zylo reports a median of 305 apps across all organization sizes, for an average annual SaaS bill of 55.7 million dollars.

What is even more telling is that in most cases, nobody at IT knows the full list. Torii estimates that 61% of deployed applications fall under shadow IT, against only 15.5% that are formally sanctioned. Put plainly, when you ask a CIO "who owns this tool", there is a 6 in 10 chance they cannot answer.

And it is not just about the raw count. MuleSoft measured that the average enterprise runs 897 applications, with only 29% of them actually integrated. IT teams end up spending close to 40% of their time hand-wiring custom integrations between tools that were never designed to talk to each other.

All of this has a cost. Gartner estimates that organizations waste an average of 30% of their software budget on overlapping features and underused tools. On a 50 million euro IT budget, that is 15 million vanishing every year. Every single year.

OK, now that the scene is set, let us zoom in on what this looks like on the EA side specifically.

What the architecture stack of a large company really looks like today

Say you are an enterprise architect in a 5,000-person organization in 2026. Your environment probably looks something like this:

• An EA tool (LeanIX, Ardoq, HOPEX, Alfabet, Boldo, depending on the history of the house)
• A separate BPM tool to model processes
• A GRC tool to keep up with DORA, NIS2, and soon ISO 42001
• A data catalog that showed up during the GDPR wave
• An AI register required by the AI Act (applicable from August 2026)
• A process mining platform, often Celonis
• An IT cost management solution that used to be an Excel file three years ago
• A vendor management tool, often a repurposed HR module

Each brick landed for a good reason. Regulatory pressure, operational pain, a reorg, a strategic project. None of them are useless on their own. The problem shows up when you look at them together.

The same business objects end up modeled three times, in three repositories speaking three slightly different languages. Your business capability "customer onboarding" is defined in the EA tool, in the BPM tool, and in the GRC tool. With three glossaries that do not quite match. So when someone on the exec team asks a simple question like "how many critical applications support our onboarding process", it takes three weeks to answer because you have to reconcile three sources.

That is what tool sprawl actually looks like day to day. Not a "too many tools" problem. A "repositories that overlap without talking to each other" problem.

The costs nobody puts in the TCO

When you price out what this stack really costs, licenses are usually the first thing on the spreadsheet. Fair enough, it is the most visible line. But it is much less than half of the real bill. Here are the four places where money actually leaks out.

Duplicate contracts. Zylo measured that the average company pays for 26 applications in "multi-channel spend", meaning different teams bought the same tool multiple times without talking to each other. EA tools themselves range from 5,000 to 250,000 dollars a year according to Ardoq, so once you stack a premium EA, a BPM, a GRC, a data catalog, and a process mining platform for a 5,000-person company, you are easily sitting between 400,000 and 800,000 euros a year in licenses alone, implementation not included.

Custom integration work. The 40% of IT time spent integrating tools represents a serious amount of senior engineering salary going into plumbing that no framework can reuse. When someone leaves the team, the knowledge walks out with them. Shopify cites an independent consulting study pegging the TCO of fragmented stacks at 36% higher than unified platforms, and most of that gap sits in the integration layer.

Data duplication. If your business capability is modeled in three places with three different wordings, your reports to the executive committee will never be fully comparable. GBTEC calls the thing that is missing a "traceability mesh": a single map connecting processes, data, systems, and controls, letting you trace an incident back to its root cause in a few clicks. Without that map, every rationalization exercise starts from scratch. IDC estimates that 20 to 30% of applications in most companies are redundant, but you still need the ability to prove it.

The AI governance black hole. This is probably the line that hurts the most in 2026. The AI Act becomes fully applicable on August 2, 2026 for high-risk systems. Every AI system has to be classified, documented, traced, and tied to an owner. Gartner measured in 2024 that 47% of organizations without an AI governance framework saw their costs go up, and that 36% of their AI initiatives failed. OutSystems found at the start of 2026 that 97% of organizations are exploring agentic AI, but only 36% have centralized governance. The gap between those two numbers is pure risk.

Add those four buckets together and you land right back on Gartner's 30% waste estimate. There is nothing abstract about it.

What is happening on the vendor side (and why you should care)

EA vendors are not passive observers here. They saw this coming, and 2025 was the most consolidated year the sector has ever had.

In January 2025, Bizzdesign closed its acquisition of MEGA International, followed by Alfabet, picked up from Software AG. The combined group, now trading under the Bizzdesign brand, represents 110 million euros in revenue, 2,000 customers, and 600 employees. The combined offering spans EAM, Strategic Portfolio Management, BPM, and GRC under one roof. Forrester, in its January 2026 report Enterprise Architecture Tools Become A $1 Billion Strategic Battleground, openly calls out a market shifting from fragmented to concentrated.

In parallel, SAP has folded LeanIX into its ecosystem, ServiceNow is pushing its Application Portfolio Management on top of the CMDB, and the independent pure-players that remain (Ardoq, Boldo, and a few others) are investing heavily to broaden their footprint into adjacent disciplines.

The thesis behind all of these moves is the same: buyers no longer want seven separate tools that communicate poorly. They want a core platform where EA, processes, risks, and governance all live in the same graph.

GBTEC puts it well in its 2026 report: EA is becoming a living map and rules engine, tightly paired with BPM as the operational execution layer. The clean separation between those disciplines, which has structured the market for 15 years, makes less and less sense in a world where AI agents need a unified repository to act correctly.

Consolidating, yes, but not "one tool for everything"

Heads up: consolidating does not mean centralizing everything into a single tool. No vendor perfectly covers every discipline, and the vendor lock-in you would create by forcing it is its own problem. That is not the idea.

Consolidating means picking a core platform that holds the reference layer of your transformation, and rationalizing whatever plugs into it around that. That core platform is enterprise architecture. For one simple reason: it is the only place where business objects (capabilities, processes, applications, data, infrastructure, risks, AI systems) all live in a unified semantic repository.

When your process mining tool needs to know which process to analyze, it queries the EA. When your AI register needs to classify a new model under Annex III, it inherits the business context from the EA. When your GRC tool assesses the impact of a new regulation, it leans on the application map held in the EA.

Concretely, a consolidated stack in 2026 looks like this:

• At the center, an EA platform holding the unified repository
• Around it, specialized tools consuming that repository via API
• No duplicate repositories, one single source of truth on business objects

Xenoss, who works on a lot of data stack consolidation projects, measures that this model cuts between 40 and 60% of integration time. It removes the modeling duplication. And importantly, it makes the AI Act inventory actually tractable, which is not a minor topic with August 2026 around the corner.

Where to start if this sounds familiar

If your company is in this situation (and statistically, it probably is), here is a small framework you can apply before your next contract renewal. It is not magic, just structured common sense.

Run the full inventory. List every architecture and governance tool, its fully loaded annual cost (licenses, implementation, integration, training, internal support), the number of actually active users rather than provisioned seats, and the business objects it models. You will likely be surprised.

Find the overlaps. For every capability, process, application, and AI system, ask yourself how many tools record it. As soon as the answer is "more than one", you have a duplication to arbitrate.

Stress-test your candidate core platform. A real core platform needs to be queryable via API by all the satellite tools. If your current EA tool cannot do that, it is not yet a core platform, it is a modeling tool. That is fine, it can become one, but you need to make it an explicit selection criterion.

Phase the transition. No big bang. Pick two or three tools to sunset in the next 12 months, aligning the decommission with a natural contract renewal. It is a lot easier to defend at the executive committee that way.

Quantify the saving over 3 years. Licenses saved, custom integration hours avoided, architect time reclaimed. On a mature large-enterprise stack, you are rarely below one million euros over 3 years. That is the number that gets the project signed, not the word "governance" in bold on a slide.

What it changes for the EA team

The biggest benefit of consolidation, honestly, is not financial. It is political. An EA team that holds the core repository of the company no longer needs to justify its existence at every budget review. It becomes the context provider for every other discipline: IT, risk, data, AI, transformation.

The Zachman Institute, in its January 2026 report, frames this shift as the move from "maps to decision services". EA teams that survive 2026 no longer ship diagrams. They ship quantified answers to questions like: where is our dependency risk concentrated, which capability gaps should we close first, which applications can we decommission without breaking the critical chain, what is the architectural impact of this new regulation.

That kind of value only holds up when the repository is unified. It falls apart in a fragmented stack.

To sum up

Tool sprawl is not a death sentence, but it is worth taking seriously. The 2026 numbers are clear: 30% of the software budget wasted on average, 61% of the stack sitting in shadow IT, and mounting regulatory pressure with the AI Act. The CIOs who are pulling this off are the ones who stop adding a tool every time a new constraint lands and start asking whether their EA platform can already carry that discipline.

The reflex of buying a dedicated tool as soon as a new topic shows up (AI Act, DORA, NIS2, post-M&A reorg) is quietly becoming the most expensive pattern on the market. Every brick added without a prior mapping widens the gap between what you pay for and what you actually control.

At Boldo, we think of enterprise architecture not as yet another tool to add to your stack. We think of it as the one that makes every other tool negotiable.

Map before you buy.