In short. In 2026 brands will spend 1.3 trillion dollars on advertising, more than ever. But media buying is moving into platform dashboards, production into algorithms, brands are taking work in-house and clients refuse to pay for hours. Agencies earn exactly where the money is leaving. What remains is what cannot be automated: market understanding, data-driven decisions and knowledge that builds up from campaign to campaign.
In 2026 brands will spend more on advertising than ever before. Agencies will earn less from it.
This is not a paradox. The money is simply taking a different route.
There is more money in advertising
WPP Media estimates the global advertising market will reach 1.3 trillion dollars in 2026. That is 8.9% growth in a year, and advertising's highest share of global GDP since 1999.
But 57.6% of the market outside China is held by three companies: Alphabet, Meta and Amazon. And these companies now do what agencies used to be paid for.
According to The Wall Street Journal, Meta is building tools that will create and target ads almost without human involvement by the end of 2026. A brand provides a product image and a budget. Generative search is becoming an ad channel: 5.1 billion dollars in 2026 and more than 100 billion by 2030.
Media buying moves into self-serve platforms. Production moves into algorithms.
Brands are taking work in-house
In 2008, 42% of members of the US Association of National Advertisers (ANA) had an in-house agency. By 2023 it was 82%.
The shift is not slowing down. In ANA's 2026 study, experts were five times more likely to say brands are bringing more work in-house than ever, rather than scaling back. And it is no longer only production. Strategy is moving in-house too.
Even the giants are restructuring
At the end of 2025 Omnicom closed its 13 billion dollar acquisition of IPG and immediately cut more than 4,000 jobs. The legacy networks DDB, FCB and MullenLowe were folded into other structures.
The holding companies are open about the reason: AI is changing how advertising is produced, and platforms let brands produce it at scale without an agency.
Clients no longer want to pay for hours
In 2011 hourly fees were the dominant way brands paid agencies. Today, according to the World Federation of Advertisers (WFA), only 19% of global brands still use them as their main model.
63% expect more payment by results. 61% plan to change their commercial model with agencies because of AI. Only 20% have done it so far.
The gap between "want to" and "have done" is the window of opportunity.
What is leaving and what remains
Thales Teixeira, a former Harvard Business School professor, calls this decoupling. Customers break the value chain apart and give each link to whoever does it cheaper and faster.
Three links are leaving the agency. Media buying, where most agencies make their money. Production, which AI makes many times cheaper. And manual work: free strategies for pitches, reports, reconciliations.
What remains is what a platform dashboard cannot automate. Understanding of the market that goes beyond one client's data. The decision on where to put the next budget and why. Proof of results. And knowledge that builds up from campaign to campaign.
The problem is that most agencies earn their money exactly where it is leaving.
What happens if agencies do not change
Margins will keep shrinking. Media commission melts as buying moves to automated platforms.
The agency becomes a contractor. What stays is execution, where price wins, not expertise.
Knowledge walks out the door with people. Experience does not accumulate in a system, and every campaign starts from zero.
Why now
First. Knowledge only accumulates over time. A record of "what we predicted and what actually happened" in your own market cannot be bought. Whoever starts building it today will be the most accurate in two years.
Second. The rules are not set yet. Brands want a new way of working, but most do not have one. Whoever arrives with a ready model sets the standard.
Third. Our markets are the least visible to global models. Central Asia, the Caucasus, Eastern Europe. Local data is worth more here than anywhere, and almost no one collects it systematically.
What we are doing
For the third year now we have been building an AI Native agency.
The difference is simple. It is not an agency that added AI tools to familiar processes. It is an agency designed from scratch around those tools: how data flows through it, how decisions are made, how knowledge accumulates and what the client pays for.
Anyone can plug AI into the old model. Rebuilding the model around it is harder. That is exactly what we do.
Tools can be bought. Accumulated knowledge of your market cannot. It can only be built.
FAQ
Why are agencies earning less while the ad market grows?
Because the money flows through platforms. Media buying moves into the automated dashboards of Alphabet, Meta and Amazon, AI makes production cheaper, and brands take work in-house. These are exactly the links where most agencies make their money.
What is decoupling in advertising?
A term coined by Thales Teixeira: the customer breaks the chain of services apart and gives each link to whoever does it cheaper and faster. For agencies, that means losing media buying, production and manual work.
What is an AI Native agency?
An agency designed from scratch around AI tools: how data flows, how decisions are made, how knowledge accumulates and what the client pays for. Unlike an agency that simply added AI to its old processes.
Sources
WPP Media, This Year Next Year, June 2026. ANA, The Continued Rise of the In-House Agency, 2023. ANA, The Resilient Rise of the In-House Agency, 2026. Fast Company and Reuters on Omnicom job cuts after the IPG acquisition, December 2025. WFA and Agency Mania Solutions on agency remuneration, October 2026. The Wall Street Journal via Storyboard18 on Meta ad automation, 2025. T. Teixeira, Unlocking the Customer Value Chain, 2019.
Author: Sherali Djurabaev, founder of TAG.