The 2026 results season turned a two-year argument into a set of hard numbers. Here is what the annual reports and the share prices actually say, and what they mean if you buy consulting, work in it, or run a firm.
On 18 June 2026, Accenture fell almost eighteen per cent in a single day, and at one point in that session it was down nearly twenty. It was the worst day in the firm's history as a public company. The trigger was not a scandal or a lawsuit, it was a results statement: revenue came in slightly light, new bookings had fallen thirteen per cent from the previous quarter's record, and management quietly trimmed the full-year growth outlook to three or four per cent. The market did the rest.
For the last couple of years, whether AI would reshape consulting was an ongoing conversation, and I heard people on panels and in LinkedIn comments arguing it either would or it would not. I think we can now see that the argument is over, not because either side won it, but because these results, and plenty of others alongside them, have turned it into something we can finally quantify and try to understand.
If Accenture were on its own, you could write it off as a wobble. It is not on its own. On the same day Accenture cut its guidance, Capgemini fell nearly nine per cent. Globant, one of the market's more recent favourites, lost almost seventy per cent of its value across 2025. Endava has cut its full-year guidance twice in six months and almost halved its earnings guidance in the process, with earnings per share down about eighty-five per cent year on year, and its shares are now down more than ninety per cent since the start of 2024. Thoughtworks, a business that came to market in 2021 worth more than six billion dollars, was taken private at four dollars and forty cents a share, roughly eighty-seven per cent below its peak. And it is not confined to the listed players either, because Deloitte UK posted its first revenue decline in about fifteen years, with its Technology and Transformation business down ten per cent.
Put Accenture and Cognizant together and the two of them have shed more than a hundred billion dollars of combined market value in around two years. That is not a rounding error in a rough quarter, it is the market rewriting what it thinks these businesses are worth.
Here is where most takes go wrong, and where the numbers get genuinely surprising. If AI were simply killing consulting, the money would be draining out of the sector. Consultants love to solve problems, it is in our DNA, and throughout my career we have always found a way to turn a challenge like this into growth, so I see no particular reason why this time should be different.
The spending data backs that up, and rather more strongly than I expected. Gartner now expects global IT spending to grow more than fourteen per cent in 2026, to about six point four trillion dollars. The number worth noticing there is not the forecast itself but the direction of travel, because last October Gartner was saying under ten per cent and it has revised that figure up three times since. Whatever AI is doing to consulting, it is not shrinking the amount of money organisations are spending on technology.
Nor is every firm shrinking, despite the wild headlines we keep seeing. Cognizant is still growing and still hiring, on record bookings. TCS added more than nine thousand people in a single quarter even as it grew its AI book to two point six billion dollars annualised. Capgemini, which was going backwards this time last year, is now growing at eleven per cent. The line that AI destroys consulting headcount simply is not true everywhere, it is far more nuanced than that, and any honest look at the results has to sit with it.
So the story is not collapse at all, it is rotation. The money is not leaving consulting, it is moving around inside it, and the market is repricing which kind of consulting has a future.
Once you sort the winners from the losers, the pattern is pretty clear. The firms and business lines being punished have one thing in common: they sell people by the hour, or they sell advice without owning the delivery. The pure build shops, the legacy IT outsourcers, the standalone technology-consulting arms. The ones staying strong and growing are the ones selling outcomes, or embedding AI directly into the work rather than writing a report about it.
Publicis is a good illustration, because both halves of it sit inside a single company. The group grew almost five per cent organically in the first half of this year, while Publicis Sapient, its technology-consulting arm and about thirteen per cent of net revenue, went backwards. The same building, two directions, and the market can read the difference.
The clearest version of the same signal is DXC, where the order book itself is telling you what is coming. Its book-to-bill ratio has fallen below one, which means the backlog is shrinking faster than the business can refill it, and the company is guiding to another year of decline.
The most revealing story of the whole season belongs to Accenture, and it is not the share price. Most of the coverage settled on one number: headcount fell by about twenty-two thousand between February and August 2025, from roughly eight hundred and one thousand people down to seven hundred and seventy-nine thousand. Accenture never actually published that figure, as it happens, it is simply what you get by subtracting one quarterly disclosure from another, but the fall itself is real enough.
The trouble is that it is only a decline if you stop looking in August. Accenture had climbed to that February peak from seven hundred and seventy-four thousand the previous summer, and by the end of May this year it was back up to seven hundred and ninety-nine thousand, within a couple of thousand of the peak it had supposedly fallen from. What got reported as a collapse in headcount was, on a longer view, a dip.
What happened in between is the part worth paying attention to. The firm ran a six-month optimisation programme that finished last November and cost nine hundred and twenty-three million dollars, the bulk of it severance. Julie Sweet was strikingly direct on the earnings call about who was going, describing it as exiting people, on a compressed timeline, where reskilling was not a viable path for the skills the firm needs. And in that same financial year Accenture booked five point nine billion dollars of new generative-AI work, close to double the year before.
So the firm spent the better part of a billion dollars taking people out, hired steadily back in, and finished roughly where it started on headcount. That is not a business shrinking. It is a business swapping one kind of person for another, in public, and paying handsomely for the privilege. If you want the clearest single illustration of what AI is actually doing to consulting, it is not a firm getting smaller, it is a firm changing who it employs while the total barely moves. The market's problem is that it cannot yet decide whether to applaud the reinvention or worry about what it implies, so for now it is doing both at once.
Analysts have started calling it AI deflation, and having spent my career inside the model it describes, the logic is uncomfortably familiar. Consulting was built as a pyramid, with a few senior partners at the top carrying the relationships and the judgement, and a wide base of junior people underneath doing the analysis, the research and the slides. The work that filled the bottom of that pyramid is exactly the work a good model now does in minutes, so the price of it falls, and with it the revenue that base used to generate. An HSBC sector note this summer warned that Indian IT services could see anaemic growth for six to eight quarters on this alone, and Bain has put a number on the stakes, estimating that carrying on as usual could erase thirty per cent of revenue and up to half the enterprise value of technology-services firms over five years.
The pricing model built on that pyramid is going the same way. The billable hour only ever made sense when the hours were the cost. Around three-quarters of BCG's largest AI cases now carry variable fees tied to results, though the firm is honest that outcome-based pricing is still under a third of its work overall, and McKinsey's global managing partner has said that about a third of the firm's revenue now comes from underwriting outcomes. When the biggest names in strategy are dismantling their own hourly rate, the direction of travel is not really in doubt.
This is not just in consulting, it's working its way through professional services more broadly, one of the most ironic was an auditor asking their auditor for a discount because of AI! Earlier this year the Financial Times reported that KPMG International had spent the previous year pressing Grant Thornton UK, its own auditor, to hand back the savings its AI rollout was producing or it would go elsewhere. I checked and sure enough, the accounts filed at Companies House show the fee for the 2025 audit came in at $357k vs. $416k the year before, a cut of about 14%. It's just an example (and not a particularly complex audit as it is the KPMG network umbrella org vs. the whole business) but it shows it's happening. Almost everything else in this piece is the price of advice inferred from revenue, utilisation or a day rate. This one is a negotiated invoice, in public, with AI named as the reason for the discount. When a Big Four firm makes that argument to the people who audit it, it becomes very hard to claim the same argument will not be made to it.
A credible version of this argument has to separate the AI story from everything else moving these numbers, because not every fall is about algorithms. Accenture's first big drop, back in early 2025, was a US federal spending shock rather than a model. Booz Allen's troubles are a government-contracting story from start to finish. Companies still make bad decisions, run projects that go wrong, lose clients and get caught by a change in public spending, and if you file every red number under AI disruption you will simply be wrong. The AI effect is real, but it is sitting inside a lot of ordinary noise that has always been there, so if you are the one leading the business, the question worth sitting with is how you tell the two apart.
If you buy consulting, the results are telling you what to stop paying for. Information you could now get from a model, and a team of juniors assembling a deck, are being repriced towards zero in front of you. What still commands a premium is a firm that will own an outcome and stay accountable for it long after the presentation is over.
If you work inside one of these firms, do not wait to be told where you fit. The squeeze is landing on the middle, on the layer that translated information into recommendations, and the people who come out of this well will be the ones who move first rather than waiting for a reorganisation to decide for them. Work out where your actual strengths sit, whether that is judgement and client trust at one end or hands-on delivery with AI embedded in the work at the other, then go and shape a role around them. The firms are rebuilding themselves in public right now, which is precisely the moment when a role can be reshaped by the person doing it rather than by a slide in someone else's operating model review.
And if you run a consulting firm, the results have handed you both a warning and an instruction. The part of your business that was information, leverage and a deck is being marked down in public, and no amount of AI branding on the proposal will change that. The part that is owned outcomes is where the value went, which means the uncomfortable work is not choosing an AI strategy but redesigning your own pyramid, your pricing and your delivery model around results you are willing to be accountable for. That is the whole thesis of escaping the deck, and for once you do not have to take my word for it. You can read it in the results.
The market has not decided consulting is over. It has decided that advice without delivery is over, and it is putting a price on the difference.
Alex Collins
Co-founder & COO at RAI Digital & Infinite PL · Ex-EY Consulting Partner · Writing on agentic AI, venture building, logistics platforms and transformation leadership.
One honest read every fortnight on what agentic AI is doing to consulting, and what to do about it. For people moving from advice to outcomes, rebuilding a firm, or done buying decks. No hype, no fluff.
My old university asked me back to lecture on the future of consulting. The honest version is that the "scissors of death" are closing on advice-only work, faster now that AI has grabbed one of the blades, and what replaces it is a better job than the one it ends.
AI has not killed consulting. It has killed the part of it that was only ever a nicely formatted opinion. The premium has moved from the cleverest deck to the owned outcome.
What I learned moving from advisory work into operating and venture building: what transfers, what does not, and what catches most people off guard.
