The Consulting Shakeout: Why 8,586 New Firms Cannot All Survive the AI Restructuring of Corporate Advice

The Market That Is Growing and Dying at Once

Here is the counterintuitive fact your chairman will not enjoy at the next strategy day: the consulting industry is experiencing record client demand at exactly the same moment that the business model underpinning most consulting firms is being systematically dismantled. These two things are not in contradiction. They are the same event, viewed from different floors of the building.

AIBD analysis of Companies House data shows 8,586 new SIC 70.22 management consultancy entities registered in Q2 2026 alone. A number that sounds like vitality. It is not. That figure represents a 33.2% collapse versus the prior period, the steepest single-quarter drop in new advisory firm formations we have recorded. The army of independents and micro-boutiques that flooded the market in 2024 and 2025, riding the AI advisory goldrush, is no longer forming at anything like that pace. The retreat has begun.

Drucker observed that the purpose of a business is to create a customer. What he did not add, because it was once obvious, is that the customer must actually need what you are selling.

What the M&A Data Actually Says

PwC's Global M&A mid-year outlook, published this week, contains a figure that should be stapled to every boutique strategy firm's partnership retreat agenda. In the first half of 2026, just 17% of the 100 largest corporate M&A transactions cited AI as part of the stated strategic rationale, down from approximately one-third of deals in 2025. The slide from a third to a sixth in twelve months is not a sign that AI's importance is fading. It is a sign that buyers have stopped paying a premium for the word itself. They are becoming, in PwC's measured phrasing, "more disciplined about where AI creates durable demand."

Translation: eighteen months ago, a CFO could walk into a board meeting, announce an AI transformation programme, and watch the share price nudge upward and the consulting invoices get signed without excessive scrutiny. That era is over. The same PwC data shows global deal value on track to hit $4 trillion in 2026, up roughly 13% year-on-year, but with a savage catch: strip out megadeals above $5 billion, and deal value is actually down 4%. The market is concentrating at the top with extraordinary speed. Megadeal values are on track to rise 40% year-on-year if the current pace holds.

This is a K-shaped M&A market, and it has an exact consulting industry analogue.

The Qualcomm Signal

Yesterday, Qualcomm announced a $4 billion all-stock acquisition of AI software startup Modular, a vendor-neutral software layer designed to run AI models across diverse hardware architectures without code rewrites. The stated strategic target: Nvidia's dominant CUDA platform. Read past the chip industry framing, and what you are actually observing is a large corporation spending four billion dollars to acquire genuine software infrastructure, not a consulting engagement, not a transformation roadmap, not a pilot programme. Actual defensible technical architecture.

The same day, SpaceX signed a $6.3 billion computing infrastructure deal with AI startup Reflection AI, with payments of $150 million per month beginning July 1. Reflection AI joins Google and Anthropic as tenants of SpaceX's Colossus 2 data centre in Memphis. SpaceX, a company that builds rockets, is now a hyperscaler competitor.

Porter's Five Forces has a category for this: substitute products. The substitute for a strategy consulting engagement on AI architecture is increasingly a capital transaction. You do not hire a firm to tell you about your compute strategy. You acquire it.

The Bifurcation Is Structural, Not Cyclical

The Management Consulted industry report, also published this week, puts the dynamic cleanly: "scaled ecosystem integrators and highly specialised execution firms continue gaining share, while mid-market generalists without clear differentiation face increasing pressure." The consulting industry's transition toward what the report calls "platformised AI execution" is accelerating. Firms are building governed workflows and repeatable digital assets rather than selling standalone advisory hours.

The 33.2% drop in new SIC 70.22 registrations shows that the supply-side excitement following every AI product launch of 2024 has met the demand-side reality of 2026. Clients, as Grant Thornton's Q2 CFO survey found this month, are moving away from spreading AI thin and want proven use cases with defined ROI. A CFO who has internalised that discipline does not retain a generalist advisory firm to run twenty AI pilots. She retains nobody, or she retains a highly specialised executor with a measurable track record.

Writer's 2026 enterprise AI adoption survey, conducted with Workplace Intelligence across 2,400 respondents, found that 97% of executives report their company deployed AI agents in the past year. Yet only 29% see significant organisational ROI. The gap between deployment and value is not a technology problem. It is a strategy and governance problem. And crucially: it is a problem that most of the 8,586 new advisory entities registered this quarter are not equipped to solve.

The Translation Problem at the Top

McKinsey's Global Tech Agenda 2026 found that nearly two-thirds of top-performing companies have their technology leaders "very involved" in crafting enterprise strategy, versus 52% at other organisations. The elevation is real. But what it means for the consulting market is less flattering than the headline suggests: if the CIO is now a co-architect of enterprise strategy, the need for an external firm to provide that strategic architecture diminishes. The client is building the capability in-house.

At precisely the moment the externally-commissioned strategy document becomes less valuable, the cost of producing it with AI tools collapses to nearly zero. That is a margin catastrophe for the generalist middle. The price the client will pay and the cost basis justifying that price are moving in the same direction simultaneously.

Sun Tzu, on the rare occasion he earns a citation, noted that invincibility lies in defence, the possibility of victory in the attack. The mid-market generalist consulting firm is doing neither. It is holding a defensive position that no longer exists, against a competitive threat it has framed as an opportunity.

The Firms That Will Survive

Big Four firms moving into agentic AI delivery, KPMG's global alliance with Anthropic being a recent and clean example, are repositioning as execution infrastructure rather than advice vending machines. Accenture's $4.175 billion acquisition of Dragos, RunZero, and NetRise in the OT cybersecurity space signals the same logic: acquire defensible domain expertise and delivery capability, not market adjacency.

The boutiques that survive the next eighteen months will be those whose value cannot be replicated by a well-prompted language model plus a junior associate. Forensic expertise. Sector-specific regulatory navigation. Implementation accountability with reputational skin in the game. The rest face an uncomfortable Christensen moment: they are not being disrupted from below by a cheaper inferior product. They are being disrupted from above by clients who have decided they no longer need the product at all.

Prediction

By Q4 2026, new SIC 70.22 formations will have contracted a further 15 to 20% versus Q2's already-reduced base. The firms that survive will be unrecognisable as consulting firms by 2028: part software company, part staffed delivery operation, part specialist data asset. The ones that respond by rebranding their PowerPoint practices as AI transformation studios will not survive the rebranding.