There is a moment in every technology cycle when the CEO stops delegating and starts deciding. For cloud computing, that moment came around 2014. For digital transformation, around 2018. For AI, according to BCG's AI Radar 2026, that moment is now.
The headline finding: 72 percent of CEOs now identify themselves as the primary decision-maker on AI in their organisation — double the share of a year ago. AI is no longer an IT initiative that the CTO presents to the board once a quarter. It is a strategic agenda item the CEO owns personally, alongside revenue targets, capital allocation, and market positioning.
This is not a survey about sentiment. The AI Radar 2026, published in January, draws on 2,360 executives across 16 markets and nine industries — including 640 CEOs. The pattern it reveals is structural: AI has crossed the threshold from a technology project to a leadership discipline. And the implications for DACH Geschäftsführer and Vorstände are specific, actionable, and uncomfortable.
The leadership model has shifted
Twelve months ago the pattern in most DACH mid-market companies was predictable: the CTO or CDO identified use cases, IT evaluated vendors and ran pilots, results reached the Geschäftsführung in a quarterly update, and the CEO approved budgets and asked about ROI timelines. That made sense when AI was a technology question — which platform, which vendor, which pilot. But AI has become an operating-model question. It touches pricing, workforce planning, customer experience, supply-chain architecture, and competitive positioning at once — decisions no CTO can make alone, however technically able, because they require the strategic authority and cross-functional visibility only the CEO holds.
BCG's data confirms this shift quantitatively. Four in five CEOs are now more optimistic about AI's ROI potential than they were a year ago. This is not the cautious optimism of a technology evaluation phase. It is the operational confidence of leaders who have seen enough deployments — and enough agentic capability mature — to believe AI creates measurable business value, and who now want direct control over where and how that value is captured. Notably, 94 percent say they will keep investing at current or higher levels even if the returns are slow to arrive. This is no longer treated as a bet that has to pay off this quarter.
The urgency is equally personal: half of CEOs believe their own job depends on getting AI right. That level of stake changes behaviour. It changes meeting agendas, budget authority, hiring priorities, and the questions asked in board sessions. When a chief executive's tenure is tied to AI outcomes, AI ceases to be a project and becomes a programme. The risk in DACH is not that Geschäftsführer fail to feel this pressure — it is that they respond to it by committing capital before they have built the judgment to direct it.
What CEOs are actually deciding
The shift from delegation to decision raises questions most CEOs have never had to answer — not technical questions about model selection or deployment, but strategic ones about how AI reshapes the business, demanding a decision framework that does not yet exist in most DACH organisations.
Investment magnitude and allocation. BCG finds that companies plan to double their AI spending in 2026, bringing it to roughly 1.7 percent of revenues — more than twice the increase seen in 2025. For a €200m-revenue Mittelständler, that is on the order of €3-4m a year: no longer a discretionary innovation budget tucked inside IT, but a line item that competes with capital expenditure, R&D, and workforce development for board attention. The CEO must decide not just how much to spend, but where the spend creates the highest operating leverage — which workflows generate the most value when redesigned with AI. That is a question most CFOs are only beginning to formalise, and it is not one a vendor can answer for you.
The agentic AI allocation question. The most consequential decision in BCG's data is not total spend but composition. The report sorts CEOs into Trailblazers (roughly 15 percent), Pragmatists (about 70 percent), and Followers (roughly 15 percent). Trailblazers direct around 60 percent of their AI budget to agentic AI — systems that take multi-step actions autonomously rather than merely respond to prompts — while Pragmatists and Followers sit closer to 25 percent. That gap is not a preference; it is a different theory of what AI is for. Trailblazers build organisations where AI acts as an operator, Followers where it acts as an assistant. This maps directly onto the difference between augmentation and delegation in our Three Levels framework: a quarter of budget on agentic AI is a bet on helping people do their jobs better; 60 percent is a bet on AI performing work teams used to do, with human oversight as a governance rather than a production function. Roughly 90 percent of executives expect agents to deliver measurable returns in 2026 — but expectation and the operating model required to capture it are not the same thing, and the CEO who makes this allocation is choosing the organisation's operating model for the next three to five years.
Upskilling as a CEO discipline, not an HR programme. The defining behaviour of BCG's Trailblazer CEO is personal: more than eight hours per week spent on AI upskilling. That is not a lunch-and-learn. It is a full working day, every week, given over to understanding what AI can do, where it fails, how it reshapes workflows, and what organisational design it demands. These CEOs are not learning to code; they are building the judgment to make investment, risk, and operating-model decisions without depending entirely on advisors — internal or external — whose incentives may not align with the business. That personal discipline scales: Trailblazer CEOs direct around 60 percent of their AI budget to upskilling and retraining the existing workforce, against 27 percent for Pragmatists and 24 percent for Followers, and it shows in the result — roughly 70 percent of their people have been upskilled or reskilled for AI, versus about 41 percent at Pragmatists and 35 percent at Followers. The compounding cost of inaction is not a technology gap; it is a capability gap that widens every quarter a competitor is upskilling and you are not.
This lands hard in the DACH Mittelstand. In owner-managed and family-led businesses the Geschäftsführer often holds a breadth of authority exceeding that of a listed-company CEO, making the final call on capital allocation, key hires, and strategic partnerships. That makes them the natural AI decision-maker — but the Geschäftsführer who has not invested in personal AI literacy will either delegate decisions they should own or make them without the context they require. Both are expensive.
The IT-business relationship changes
When the CEO owns AI decisions, the relationship between IT and the rest of the business restructures itself. This is not an abstract organisational design point. It has immediate consequences for reporting lines, budget authority, and the role of the CTO.
AI becomes a board-level agenda item. In the delegated model, AI reached the board only with a budget request or a deployment milestone. In the CEO-led model it is a standing item, reviewed alongside revenue, market strategy, and talent — and understood as an operational performance metric, not a technology update. Which workflows are in production? What is the measured impact on cost per unit, cycle time, or error rate? When the board asks, the CEO needs answers, and the organisation builds the reporting to generate them: metrics once tracked informally in IT — model accuracy, adoption rates, integration uptime — become part of the management reporting package, and their quality determines the quality of the CEO's decisions.
The CTO role evolves from decision-maker to decision-enabler. Where the CTO once decided which platforms to adopt and which vendors to engage, in the CEO-led model the CTO supplies the technical context that lets the CEO decide — translating technical possibility into business options framed by cost, risk, timeline, and strategic fit rather than architecture. This is not a demotion but a role clarification, and for many DACH organisations an uncomfortable one. Repositioning a CTO's authority within a CEO-led structure — not reducing it — requires explicit conversation, adjusted expectations, and often a new definition of how the CTO's success is measured.
Cross-functional AI governance becomes mandatory. When AI was an IT project, governance meant model-risk management and data-privacy compliance. When AI is a CEO-level programme, governance must span the organisation and answer concrete questions: who approves a workflow before deployment? What is the escalation path when an agent produces an unexpected result? How are regulatory, reputational, and operational risks surfaced to the CEO fast enough to intervene? For DACH companies this is not abstract — the EU AI Act's obligations for high-risk systems (including AI used in hiring, creditworthiness, and access to essential services) apply from 2 August 2026, with general-purpose-model obligations already in force since August 2025. The governance frameworks most mid-market companies have built are regulatory in focus: risk classification, impact assessments, model documentation. What they miss is operational governance — the structures to run AI workflows reliably, catch failures early, and make continuous-improvement decisions. That is the gap CEO ownership exposes, and the more an organisation leans toward agentic delegation, the wider the agentic governance gap opens.
Differentiating from the capability question
BCG's earlier research — "The Widening AI Value Gap: Build for the Future 2025," from September 2025, which we analysed in The 5% Blueprint — answered a different question: what capabilities separate AI leaders from laggards. That study assessed maturity across 41 foundational capabilities and found that the 5 percent of "future-built" companies achieve 1.7 times the revenue growth and 3.6 times the three-year total shareholder return of laggards.
The AI Radar 2026 answers a different one: who decides, and what do they need to decide well? Capability and leadership are separate variables, and both are necessary while neither is sufficient. An organisation with strong data infrastructure and a skilled technical team still underperforms if the CEO delegates AI decisions to a CTO optimising for technical elegance over business impact. Equally, a CEO who owns AI decisions without grasping what capabilities the organisation needs will make confident decisions it cannot execute. The AI Radar tells us CEOs are now claiming the authority. The open question is whether their organisations are building the capabilities — data pipelines, governance models, talent development — to make that authority productive.
The operating partner model
The convergence of CEO authority and organisational capability gaps creates a specific structural need: a strategic counterpart who operates at the intersection of business strategy and technology execution. This is not the CTO, whose remit is technical architecture and platform management. It is not a management consultant, who delivers a strategy deck and leaves. And it is not a body shop supplying developers without strategic context.
What the CEO needs is an operating partner — a counterpart who understands the business objectives, translates them into AI programme and change-management design, ensures the organisation builds the capabilities to execute, and supplies the technical judgment the CEO has not yet built. The operating partner speaks both the CEO's language — revenue, margin, competitive positioning, risk — and the technology team's: architecture, data pipelines, model selection, deployment. When 72 percent of CEOs own AI decisions but few have invested eight hours a week building the judgment to make them, that gap between authority and competence is where execution risk lives, and it is what the operating partner closes — not by deciding for the CEO, but by structuring the decisions so the CEO can make them with confidence.
For the DACH Mittelstand this addresses a real constraint. Most mid-market Geschäftsführer have neither the budget nor the organisational complexity to justify a Chief AI Officer, and do not need another C-level hire. They need a fractional AI leadership function — close enough to understand the business deeply, expert enough to make sound technical calls, and disciplined enough to build capability rather than create dependency.
The decision framework DACH CEOs need
The practical takeaway from BCG's AI Radar 2026 is not that CEOs should make all AI decisions. It is that CEOs need a structured framework for the decisions that only they can make — and the discipline to build the judgment required to make them well.
Decision one: investment magnitude and horizon. At roughly 1.7 percent of revenues, AI investment is a material commitment. The CEO must decide not just the total amount but the time horizon. AI capability is a multi-year build, not a quarterly experiment — and BCG's finding that 94 percent of executives will keep investing even without near-term returns only makes sense if the decision is framed as a trajectory rather than a one-off allocation.
Decision two: augmentation versus delegation. The agentic allocation split is a proxy for a deeper strategic choice. Will AI help your people do their current jobs better, or change what your people do? The answer shapes every downstream decision — workflow redesign, role definitions, performance metrics, and the governance you need to run it safely.
Decision three: build versus partner. Few DACH mid-market companies have the internal capability to design and run an AI programme at the level BCG's Trailblazers operate. The CEO must decide whether to build that capability in-house — hiring, training, organisational redesign — or to bring in an operating partner who supplies strategic and technical capability while the organisation builds its own.
Decision four: governance scope. Governance in most organisations stops at regulatory compliance. The CEO who owns AI outcomes needs operational governance too — workflow approval, performance measurement, exception handling, continuous improvement — and that cannot be designed by the compliance team alone. It requires cross-functional input and CEO-level authority to implement, and it grows more urgent as the EU AI Act's high-risk obligations land in August 2026.
These four decisions interlock: magnitude determines what is possible, the augmentation-versus-delegation choice determines what is built, build-versus-partner determines how fast capability is constructed, and governance scope determines whether the programme runs reliably once live. None of them can be made for the CEO by a CTO, a CDO, or an external consultant — they require the strategic authority and personal accountability that 72 percent of CEOs are now claiming. The question is not whether to claim that authority, but whether to exercise it with the judgment and support structure it requires.
A Fit Call maps the four decisions above against where your organisation actually stands — so you direct AI investment with the judgment a CEO-owned programme requires, before the spend is committed and the operating model is locked in.
References: BCG, "AI Radar 2026: As AI Investments Surge, CEOs Take the Lead," January 2026 — https://www.bcg.com/publications/2026/as-ai-investments-surge-ceos-take-the-lead (survey of 2,360 executives across 16 markets, incl. 640 CEOs; 72% CEO AI ownership, four-in-five more optimistic on ROI, 94% sustaining investment, ~50% job-stability link, doubling of spend to ~1.7% of revenues, Trailblazer/Pragmatist/Follower segmentation, 8+ hours/week CEO upskilling, ~60% vs ~25% agentic AI allocation, 60/27/24% upskilling-budget split); BCG, "The Widening AI Value Gap: Build for the Future 2025," September 2025 — https://www.bcg.com/publications/2025/are-you-generating-value-from-ai-the-widening-gap (5% "future-built", 41 capabilities, 1.7x revenue growth, 3.6x TSR); European Commission, "AI Act — Regulatory framework," high-risk obligations applicable 2 August 2026 — https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai.