AI at work: why only 1 in 10 companies scale it in 2026

Two reports published in 2026 — one from Deloitte, one from EY — arrive at the same diagnosis from different angles: almost no company still asks whether it should use artificial intelligence, but very few have moved it past the pilot stage. At Evicron, an AI and custom software studio based in Barcelona with over 200 projects delivered across 12 industries since 2019, it’s the conversation that comes up most in first meetings with clients: “we already use ChatGPT, but we don’t know what to do with it beyond that.” Here’s the full picture, with the numbers, of where the market really stands and what separates the companies that move forward from the ones that stall.

The report that splits companies into three even thirds

Deloitte’s 2026 State of AI in the Enterprise report, based on a survey of 3,235 business and IT leaders across 24 countries, sorts organizations into three near-equal groups. 37% use AI at a surface level, with little or no change to existing processes. 30% are already redesigning key processes around AI. And 34% are using it for deep transformation — building new products and services or reinventing entire processes and business models (full Deloitte Global report).

Workforce access to AI tools has also grown fast: from 40% to 60% in a single year, per the same survey. But the uncomfortable number sits right behind it: of those who already have access, fewer than 60% use it in their daily work. Handing out licenses is not the same as changing how work actually gets done — and that’s where the three thirds part ways.

Only 1 in 10 reaches an advanced stage

EY’s Challenges and Trends for Companies in Mexico and Latin America 2026 study adds the other half of the same picture: AI already dominates the boardroom agenda, but only 1 in 10 companies has pushed its adoption to an advanced stage (summary via El Cronista). According to Jaume Sués, EY’s emerging-technologies partner for financial services in Latin America, companies are shifting from an early phase focused on learning and experimenting with AI to one where they demand business cases, scalability, and measurable results before investing further.

That’s exactly where many small and mid-sized companies get stuck: someone on the team tries a tool, it works well in an internal demo, and it stops there — because nobody turns that pilot into a process with an owner, a budget, and a return metric.

The problem isn’t purely technical: there’s a governance gap

Deloitte’s second 2026 report, Global Human Capital Trends, points to a concrete reason the jump is so hard: 60% of executives already use AI to support their decisions, but only 5% say they manage it well (Deloitte Insights report). Deloitte calls this “culture debt”: the cost an organization accumulates when it scales AI use without also maintaining the accountability structures, norms, and trust needed around it. 34% of companies admit their own internal culture is holding back their AI transformation goals.

This ties directly into what we already track at Evicron on the regulatory side: Spain’s AI law working its way through parliament and AESIA’s compliance guides aren’t just a legal box to tick before an inspection. They are, in practice, the governance structure that most companies — the 95% Deloitte identifies — still lack.

What separates the companies that actually scale

Cross-referencing all three reports, the pattern that separates the 34% transforming processes from the 37% stuck at the surface isn’t the AI model they use — most use the same ones — but three management decisions:

• A concrete process with an owner, not a generic tool handed to the whole team with no more guidance than “use it when you can.”
• A return metric defined before starting, not a nice demo nobody measures again six months later.
• Someone accountable for governance: what data the system uses, who reviews its outputs, and what happens when it gets something wrong.

None of the three requires a huge budget. What it requires is treating AI as a project with a defined scope — like any other piece of software development — rather than a subscription you switch on and forget.

How we work on this at Evicron

In our AI consultancy for companies we always start with the same thing that separates the 34% that move forward from the rest: identifying one concrete process — not “adding AI to the company” in the abstract — and defining upfront who owns it and how the result is measured. If your company is stuck at “we tried it, but we don’t know how to scale it,” our guide to AI consultancy walks through when outside help is worth it and what to ask before signing; and if budget is the sticking point, we cover real price ranges in our guide to the cost of implementing AI.

In summary

Two international reports from 2026, one from Deloitte and one from EY, confirm the same gap from different angles: almost every company has already tried AI, but only around one in ten has turned it into a real process with measurable return. What separates that group from the rest isn’t the technology — it’s three management decisions: a process with an owner, a return metric, and governance, all of which any company can put in place without multiplying its budget.

Has your company been “trying out” AI for months with nothing to show for it? Get in touch: the first discovery session is free and we reply within 24 hours.