Why AI May Cause a lot of Ad Tech Companies to Duck and Cover
It may be best to go through the great re-tooling in private
I spent much of last week in Sandpoint, Idaho at the 12th annual Kochava Summit, during which pushing CTV toward becoming more of a performance vehicle, and naturally - how are AI agents going to change everything - were hot topics.
Kochava’s roots are in mobile measurement and particularly attribution - but more recently the company has pushed into connected TV and other media. And now Kochava, which counts NBC Universal and Walmart, as clients is looking to marry measurement with media buying and planning via its AI platform Atlas Performance.
As I wrote about last week, that’s potentially a very different, far more integral role for measurement companies. We’ll see where things net out as AI further reshapes media buying, and really all of digital advertising.
Surely, there will be some winners and losers if this agentic revolution takes hold - which is something I asked Kochava CEO Charles Manning about during our live podcast interview at the summit.
He predicted lots of deals, consolidation, and maybe even something close to chaos in ad tech.
“There’s going to be a lot of collapsing,” he said.
“For every ad dollar, there’s historically been [an ad tech] tax. A lot of that middleware gets replaced agentically. You might spend more on a fleet of agents than before — but less than all the other middleware combined.”
By middleware - I think Charles means SSPs, DSPs, and all those ad tech middlemen. Such a collapse should theoretically be good for publishers and brands, but maybe not for the Lumascape.
“Publishers may begin building exclusive inventory access at ultra-premium rates once they understand their true value. And whereas programmatic was about auctions per impression, we may start to see auctions at the IO level — like programmatic guaranteed at scale.”
So what should ad tech companies do? Sell, or maybe even ride out the storm in the private markets.
“Being private over the next six to 16 months is going to be considered gold,” Manning said. That’s where ad tech companies may be able to regroup and retool. It’s not easy to ‘agentify’ all of your operations in the public eye while still trying to top last quarter’s revenue.
Otherwise, it may feel like a fire sale for some. “Tech is re-creatable really easily with AI. [So would-be acquirers] are going to be purchasing contracts [not companies or software].”
I told Charles that it’s hard for me not to be skeptical over AI and Agent Mania, given the recent hype around how we’d all be living and marketing in the metaverse while trading NFTs right about now.
But he was unfailing in his belief that digital advertising is set to be fully reinvented - rather quickly.
“If the last decade was about digitization of signal and media, and the next decade is about digitization of workflows,” he said. “That’s a really good representation. And it’s probably not a decade — it’s probably 16 months.”
Ok, but didn’t that already happen like 12 years ago, with the advent of programmatic and real-time bidding?
“Machine learning has already done a good job optimizing for reach and frequency. Increasingly, it’s about business outcomes — because CFOs want net impact,” Manning argued. “What’s new is automating the workflow steps — gathering spreadsheets, compiling data, preparing briefs. Instead of spending a week preparing a post-campaign brief 45 days later, that workflow can now happen daily.”
Compared to the programmatic era, this shift is happening, “tremendously faster,” he said. “Breathtakingly faster.”
“We’ve always been in the business of real time for signal. Now the processes will be close to real time as well.”
So what should a brand worry about? Charles listed two:
“First: their data is their proprietary differentiator. They need to use it for their benefit — not leak it into someone else’s workflows.”
“Second: the how is as important as the what. When everyone has access to AI, differentiation will come from how growth is executed.”
Other than that, we can all just chill and go about business as usual, right?
“Everything is happening super fast, and it is transformative to how people work,” said Manning. “When something happens quickly and impacts how people work, there’s going to be disruption.”
There is a lot more detail to unpack here, which is why I encourage you to check out the full episode here:
“Publishers may begin building exclusive inventory access at ultra-premium rates once they understand their true value. And whereas programmatic was about auctions per impression, we may start to see auctions at the IO level — like programmatic guaranteed at scale.”
Yes! It's called an insertion order, and it's been used by buyers and sellers for decades to convey dates, dollars, goals and constraints. PG is an IO routed through programmatic pipes, often unnecessarily. More than 1/2 of all CTV/OLV dollars transacted are still executed by IO, and it's been consistent for several years. Look for much greater automation of this without the programmatic fees. That's the baseline for agentic at scale.
The agentic collapse framing rings true, but the real shift is power, not efficiency. If agents compress the Luma layer, leverage moves to whoever controls the data and inventory access. Less ad tech tax disappearing, more a new toll booth forming.