On June 18, Accenture reported quarterly earnings, cut its full-year growth guidance to 3%-4% (down from 3%-5%), and watched new bookings fall 2% year over year to $19.32 billion. TD Cowen downgraded the stock to Hold, and the reasoning left no room for ambiguity: Accenture’s headcount-and-billable-hour model is being structurally cannibalized by rapid enterprise adoption of generative AI. The stock was still down 7.28% on June 22, four days after the report.

Two weeks later, on July 2, Microsoft announced Frontier Company: $2.5 billion and 6,000 people dedicated to implementing AI inside its largest customers. And there was Accenture, named a launch partner alongside EY, London Stock Exchange Group, Unilever, Land O’Lakes, and Novo Nordisk.

Read that in order. Accenture didn’t dodge the hit by being on the right side of the table. It took the hit first, in its own stock price, and two weeks later Microsoft invited it to the new table anyway. Sitting inside the labs’ AI ecosystem bought no immunity from the repricing. That’s the part of this week that actually matters, more than any single number: if your business still bills by the hour, it doesn’t matter who you partner with afterward. The repricing already happened.

Two new consultancies, $6.5B combined, weeks apart

Accenture isn’t the only proof point. May had already delivered the first one: OpenAI launched the OpenAI Deployment Company with $4 billion, led by TPG alongside Advent, Bain Capital, and Brookfield as founding co-lead partners, with OpenAI holding majority ownership and control. The same deal folded in the acquisition of Tomoro, an applied-AI consulting firm bringing roughly 150 deployment engineers and clients like Mattel, Red Bull, Tesco, and Virgin Atlantic. The deal is set to close in the second half of 2026.

Microsoft didn’t shy away from the comparison in its own announcement. Judson Althoff, CEO of Microsoft’s commercial business, put it directly: Frontier Company “goes beyond what has been labeled as Forward-Deployed Engineering, and will be the largest, most capable, outcome-driven engineering organization in the industry.”

Ben Appleton summed it up well this week: the consulting market now has four powerful new competitors: OpenAI, Anthropic, AWS, and Microsoft. Anthropic already runs its own Applied AI team, actively hiring Forward Deployed Engineers for enterprise rollouts. AWS committed $1 billion to its own Forward Deployed Engineering unit back in June, something I already covered when I wrote about Alex Karp’s attack on token pricing. None of the four are spending this quarter on better models. All four are spending on people who sit with the client and implement.

Why now? Ashwin Kadaru laid out the mechanism well: companies try AI in chat first, then build throwaway internal apps, and only once one of those tools becomes load-bearing do they ask someone to wire it into their real systems and keep it running in production. That’s exactly the job Frontier Company and the OpenAI Deployment Company are now selling directly, with no consultancy in between.

Being a partner didn’t save Accenture

Accenture’s official explanation for the guidance cut named three macro headwinds: the geopolitical impact of conflict in the Middle East, clients deferring large managed-service agreements while waiting on the Federal Reserve’s rate path, and persistent weakness in U.S. federal consulting sales. None of those three mention AI.

TD Cowen did, by name. And this isn’t Accenture’s first attempt to climb into the boat with the labs instead of competing against them: back in February, Fortune reported that OpenAI had already partnered with McKinsey, BCG, Accenture, and Capgemini to push its agent platform into large enterprises. That alliance didn’t prevent June’s hit either.

This isn’t only a big-four problem. James O’Dowd argued on LinkedIn this week that most time-and-materials professional services firms are about to see their margins compressed, because once any capable firm can produce multiples of the same output with AI, the market floods with capacity and price adjusts down, not gradually but all at once. Alice Lassman called it, in The Guardian, the slow death of the prestige career: the analyst-to-partner training pipeline that ran on billable junior hours for decades is losing the economic floor that funded it.

What AI compresses first isn’t the senior partner’s judgment call on the final recommendation. It’s the work that funded the whole pyramid underneath: market scans, first-draft analysis, comparing twenty vendors side by side, synthesizing a hundred interviews into a deck. That work bills by the hour today, and an agent can do a version of it in a fraction of the time. When that floor goes, the whole pyramid loses the base that paid for it, not just the bottom rung.

The pattern repeats across all three stories: it doesn’t matter whether you’re a $60 billion consultancy with a seat at Microsoft’s launch table or a boutique shop that never got invited to anything. If your invoice depends on hours and headcount, the market already decided that model is worth less. The question left for every company isn’t who to partner with. It’s how to stop billing by the hour before someone else forces the switch on you.

What we do at IQ Source

A mid-market company isn’t going to buy a seat in a $4 billion fund backed by TPG, and it doesn’t need Microsoft’s $2.5 billion war chest to justify its next AI project. What it needs is for one specific workflow to get implemented against a clear outcome, without its contract depending on which lab its vendor happened to sign with that week.

That’s why our Technology Partner model is scoped around the workflow, not headcount or hours. I made the underlying argument already when I wrote about why you are what you charge for, not what you install: charging for the outcome forces the change to actually happen, because your invoice depends on it, not on how many people got staffed onto the project. When we come in as Technology Partner for a software company subcontracting AI implementation, the scope gets defined by the workflow that has to change, not by a block of hours that bills whether real progress happened or not.

Microsoft and OpenAI are solving this for the largest accounts on the planet, with billions behind them. Most of the companies you’re going to compete against, or sell to, aren’t in that league. They’re deciding right now whether their next AI project gets paid by the hour or by the outcome, and this week the market already gave Accenture the answer, in the most expensive way possible.

Talk to us about a Technology Partner scope for your next AI project