I want to write something that I would not normally put on the public record, because I think the moment requires it. My job at Komodo X is to sit across from the people who run holding-company agencies, regional offices, in-house brand studios, and to figure out how we partner with them, compete with them, or quietly take their work depending on what each conversation calls for. I have done this for long enough now that I can hear the same anxiety repeated in seven different accents in any given month. So let me describe it plainly.
The holding-company model is cracking. The people running it know it is cracking. They are not yet sure what to say in public, because their share prices and their quarterly earnings calls require a more confident story than the one they tell each other in private rooms. So they hire AI consultants, they announce in-house AI studios, they restructure into one-team global models, and they hope the restructure outruns the structural collapse it is responding to.
I am writing this from the other side of that pitch wall. Here is what I am hearing.
What the holding companies actually believe in private
The first thing they tell me is that the AI productivity gains are real, and the gains are larger than they have publicly admitted. Internal data points to 30 to 50 percent of traditional production tasks coming out of the workflow inside 18 months. That figure is being publicly framed as opportunity. Privately it is being framed as headcount reduction, with the headcount question deferred until after Q4 numbers.
The second thing they tell me is that the new roles emerging, AI Workflow Producer, Prompt Engineer, AI Supervisor, are not staffed at the seniority the work requires. The people qualified to do these jobs at a senior level are not in the holding company. They are at AI-native studios, at platform companies, or freelance. Holding companies are hiring entry-level prompt operators and trying to manage them through a creative leadership that does not understand the work. The result is friction, and the friction is showing up in client delivery.
The third thing, and this is the one they say most quietly, is that the client does not need them in the loop the way the client used to.
A brand that previously needed a holding-company team of forty to deliver a multi-market campaign can now deliver it with an internal team of eight, an AI-native partner studio, and a freelance director. The holding-company economics depend on the larger team being necessary. The larger team is no longer necessary.
Why the in-house AI studios are not the answer they look like
Every major holding company has now announced an internal AI studio or capability. Publicis has CoreAI. Omnicom has its creative intelligence stack with Google. WPP has Open. The branding is excellent. The execution is uneven, and the reason the execution is uneven is structural, not tactical.
An in-house AI studio inside a holding company is fighting against the gravity of the parent organisation. It needs to bill at parent-organisation day rates, follow parent-organisation procurement, work through parent-organisation approval cycles, and hire under parent-organisation comp bands. Each of those constraints is incompatible with the actual operating physics of AI-native production. The fast iteration speed that AI cinema requires is in direct conflict with the slow approval rhythm that a holding company is built around. You cannot solve a velocity problem by adding a velocity team and routing them through a non-velocity organisation. The parent metabolism wins, every time.
"The holding companies are not failing because their leaders are not smart. They are failing because the operating model that made them dominant in 1995 is structurally incompatible with the operating model that wins in 2026."— Zulfiqar Ali
What the smartest brands are doing instead
I am watching a quiet but consistent pattern emerge in the GCC and South Asia. The smartest brand-side leaders are unbundling. They are keeping the holding company on retainer for one or two strategic services, typically media planning, sometimes brand strategy, occasionally crisis communications. For everything else, they are going direct to AI-native studios, freelance directors, and specialist partners.
The financial model that emerges looks nothing like the old AOR model. It is a hub-and-spoke model with the brand at the centre, retained partners on specific functions, and a flexible network of execution partners on everything else. We are increasingly the execution partner. Three years ago, we would have been a third subcontractor under an Omnicom or Publicis lead. Today we are signed directly to the brand, on terms the brand sets, with the holding company informed but not in the room.
If you are a brand-side reader of this piece, this transition is happening with or without your involvement. You can lead it, in which case you control the partner mix, the IP rights, and the workflow design. Or you can wait, in which case your competitors will reach the new equilibrium first, and you will be paying holding-company rates for what will become commodity work.
Where this lands in 24 months
I do not believe the holding companies disappear. They will retain media buying, scaled localisation, and crisis communications as defensible service lines. What they will lose is the assumption that they are the default partner for original creative work. That assumption has already broken in the rooms I sit in. It is just not yet visible in the public conversation, because public statements lag private behaviour by 12 to 18 months in this industry.
Twelve months from now, I expect to be writing this piece in past tense.
Today I am writing it as a field report, from the other side of the pitch wall, for any creative leader who would benefit from knowing what is actually being said in the rooms they are not in.
Zulfiqar Ali is Head of Creative Partnerships at Komodo X, where he leads the studio's relationships with brands, agencies, and platform partners across the GCC and South Asia.