On 30 April 2026, Deloitte released its Global Technology Leadership Study 2026, drawn from more than 660 senior technology executives surveyed between December 2025 and February 2026. The headline finding does not flatter the C-suite. Eighty-one percent of tech leaders say they are confident they can scale AI. Forty-two percent, in the same survey, report low or no ROI on the AI investments they have already made. Confidence is high. The number underneath it is not.
The wider picture sharpens the contrast. Seventy-nine percent of tech leaders cite driving business outcomes as their top priority. Seventy-five percent say the operating model must fundamentally change. Forty-one percent say the business sees them as unable to keep up with demand. Deloitte’s framing for the shift is the title of its press release: from operators to orchestrators. The tech leader of 2026 is being asked to step out of system management and into enterprise value creation, cross-functional coordination, and the redesign of how the business actually runs. Anjali Shaikh, the Managing Director who led the study, puts it directly: "Today’s CIO isn’t just leading technology; they are being asked to redesign the very fabric of how the business runs."
Why It Is A Behaviour Problem, Not A Stack Problem
Read alongside the longer track record of large transformations, the 42% lands harder. McKinsey’s transformation research has shown for more than a decade that roughly 70% of large change programmes miss the goals they were set. The technology is rarely the proximate cause. The friction is in the conversations that orchestration requires: with sceptical business unit heads, with finance teams who want a five-quarter payback, with engineers who are being asked to share roadmaps with marketing for the first time.
DDI’s Global Leadership Forecast has been making the same point in adjacent territory. Only 11% of organisations say they have a strong leadership bench, and half of leaders report feeling unprepared to lead. Drop a CIO into the orchestrator role, with the AI ROI clock already ticking, and the gap between the confidence number and the outcome number is exactly what shows up.
Edmondson’s 1999 paper in Administrative Science Quarterly sits underneath this. Psychological safety, defined as the belief that speaking up about a worry will not be punished, is what determines whether a sceptical finance director actually tells the CIO that the pilot has not landed, or whether the bad news arrives via a board paper three quarters too late. In tech leadership rooms where the orchestrator behaviours have not been rehearsed, the bad news arrives late.
What Most Organisations Try
The default response to a 42% gap is more strategy. Another framework, another operating-model diagram, another investment in tooling. None of that is wrong. None of it changes what a tech leader does in the next executive-committee meeting when a peer challenges the AI business case in front of the CEO.
This is a known behavioural pattern, not a strategy failure. Roediger and Karpicke (2006), in Psychological Science, showed that being tested on material lifts long-term retention by roughly 50% compared with re-reading. A two-day senior-leader off-site of slides and panel discussions is the re-reading kind. A CIO rehearsing the actual sentences they will use when the CFO says "show me the return", against feedback, is the testing kind. One of them holds when the boardroom asks. The other does not.
Anders Ericsson’s Peak (2016) extends the point. The behaviours that make orchestration work, naming a disagreement in plain language, asking a peer what would change their mind before answering, sitting with a difficult silence rather than filling it, change only through deliberate practice. They do not change with insight.
What Actually Works
In Sidestream’s own academic behaviour-change work, building on research from UCL, Cambridge and Bocconi, immersive role-play was roughly 20% more effective than passive modalities at building communication skill, and self-rated confidence did not predict observed performance. Senior leaders, in particular, scored their own skill well above what observers later measured. The design solves that by replacing self-reports with behavioural measurement.
Applied to the orchestrator mandate, the format is a small executive group, one or two professional actors playing a sceptical CFO, a hostile business unit MD, a risk-averse general counsel. Three or four full conversations, replayed with feedback against named behavioural anchors. By the third rehearsal, the CIO’s first move is to ask the question, name the disagreement, and stay in the silence. That is the behaviour that turns a 42% ROI signal into a defensible plan rather than a defensive deck.
What We Do About It
Our immersive simulations and leadership workshops are designed for exactly the rooms where the orchestrator conversation has been postponed in favour of another strategy paper. Read also our piece on BCG’s false-consensus 70% for the wider transformation angle, or our Microsoft 26% piece for the manager-modelling layer underneath AI adoption.
81% are confident they can scale AI. 42% cannot yet show the return. The gap is not the technology. It is the orchestrator behaviour the role now demands, and it has not been rehearsed.
The CIOs who close the 42% gap over the next eighteen months will not be the ones with the cleverest reference architecture. They will be the ones who have rehearsed, out loud, the executive-committee conversations the orchestrator role actually consists of. Book a call to look at what that rehearsal would look like in your leadership team.
Book a free 30-minute diagnostic call → or read about our research-backed approach.