LUMA Partners’ Terence Kawaja on the new AI LUMAscape and why the tech will accelerate adland consolidation

For more than a decade, Terence Kawaja, founder and CEO of LUMA Partners, has been the ad industry’s unofficial cartographer, mapping its chaos through the LUMAscape.

Last month, he introduced the new AI LUMAscape, despite his promise to never release one, and LinkedIn was ecstatic. Within days, the announcement pulled in half a million likes, comments, reports and messages, with the industry harping over which companies made the cut and those that were sidelined.

“There’s not even that many people in and around the industry,” Kawaja told The Current.

The timing is ripe. AI is reshaping how marketing shows up. And in classic Kawaja fashion, the process of building the map wasn’t powered by technical posturing but by cold financial reality: If AI is truly driving a business, it will show up in the numbers.

In this candid conversation, Kawaja explains why the AI LUMAscape took four and a half grueling months to build, how AI is ushering in what he calls advertising’s “fifth wave” and why most logos on the map are destined to disappear. He also shares which categories are overhyped, which are undervalued and which AI-first companies are catching his attention.

What follows is Terence Kawaja unfiltered, just the way the industry likes him.

This conversation has been edited for clarity and length.

What is driving the timing around doing an AI LUMAscape at this time?

We actually said we would never do this as recently as a couple years ago, only because there are no LUMAscapes for ubiquitous, fundamental technology tools everybody uses. Does every company use electricity? Yes. Does that mean we should have an electricity LUMAscape? No! And I still stand by that. But it’s a differentiator in terms of who's doing what kind of AI tech and it gives the people what they want. And LUMAscapes have been around for 16 years, and so anything that revitalizes the map, we thought, might be a good idea.

It’s not an easy undertaking. How were you able to distinguish which companies made the cut? The competition must be fierce.

It took four and a half months. There’s a lot of marketing fluff. How do we parse which companies are real or not? I’m not a technologist, so I don’t think I’m qualified to test the tech. What I came to realize is, I don’t have to because it will manifest in the company’s financials. One of three characteristics will occur if they have true AI tech: No. 1: Faster growth. No. 2: Higher EBITDA margins. No. 3: Net revenue/FTE ratio.

When we first posted the AI LUMAscape, the one thing we knew was that it was incomplete, so we will be making additions and removals. I appreciate all the input from folks, but if your financials aren’t improving, then you’re like a tree falling in the forest. Nobody hears it. We’re going to take you off.

How did you determine which categories the companies should fall under?

Companies offer things that span between categories, and we get that. They only get one logo per page, so we’re going to place them in the most appropriate one. Unless your market cap starts with a T, then don’t talk to me about multiple logos.

The LUMAscape has been known to be a predictor of consolidation waves, do you agree?

If you look at all these logos, it’s clear we have a highly fragmented ecosystem. Most of those logos are irrelevant. There’s just no value add. You could get rid of 95% of the logos on the LUMAscape and nothing would change. Well, you’d have fewer players doing more volume at lower take rates with better quality, less resellers, less bullshit, less margin leakage.

I’m a big believer that we’re long overdue to have what I describe as rationalization and consolidation. Rationalization just means less logos, including failure. Not to sound mean again, but there hasn’t been enough failure in this industry. We still have way too many companies.

Do you think AI will help fuel this consolidation?

I believe AI will be a catalyst for bringing about this long-overdue consolidation or at least accelerating it. I think it’s going to be like a sword slicing through the LUMAscape because AI accelerates the move to outcomes.

How soon do you think we’ll start to see this consolidation?

Oh, it never happens as fast as you think, but it is going to happen faster than prior trends. I’m a realist, you know, it’s going be three years before it has a significant effect.

Are there any particular AI categories that you believe are undervalued or overhyped at this point?

There are some categories that I would say have proliferated so greatly that I’d be a little bit worried about how competitive they are. You could have 10 LUMAscapes just on content creation companies. If I were one of those companies, I’d worry about the sustainability of margins.

On the other hand, I like media optimization because that’s the lifeblood of a lot of the systems. Companies like Chalice, SWYM, InPowered AI and Zero to One, don’t have a lot of head count, nor a lot of costs. They have a kind of lightweight approach.

Are there any AI companies that have impressed you the most?

We’ve been pitched by everybody and their brother, trying to demonstrate their AI bona fides. I’ve seen a million media buying dashboards, I don’t need to see yet another one.

There’s some AI-first companies that really stand out. Of existing companies that have new blood, new growth, is this company Yieldmo. They have some sports content search tools that are so cool. In the category of AI-first, there’s one company called Flam. This is an Indian company that has completely revolutionized the ability to stream 3D imagery with no coding and brands are going cuckoo for Coco Puffs over this technology.

How do you see AI reshaping the power dynamics between advertisers, publishers and platforms?

For marketers, it’s just a huge win. There’s going be less waste and more amplification of great creative. On the other end of the spectrum, it’s a huge win for the consumer, because they’re getting super powerful technologies for free or close to free. Both the independents and Big Tech have thrived.

For the largest publishers, they’ll probably be OK because they’re large enough to negotiate an IP license. For the small publishers and news publishers, AI accelerates the challenges of their monetization models, and I don’t want to see that happen. We need to fix this to have a functioning democracy.

For agencies, this is either the best of times or the worst of times. It’s up to them. I said this to a group of agency execs a couple weeks ago, I said, “If you’re fearful, your clients are fearful. Help them! Be their Sherpa! Agencies have never had more agency. If you duck and hide under the covers, yeah, you’re going to fail, turns out.”

How do LLMs revolutionize advertising?

What AI does really, really well is it connects the bottom of the funnel to the top of the funnel. You go on ChatGPT, and you enter a 25-word prompt. You’re telling the LLM what you’re interested in. If you’re asking the LLM about the best jogging routes in Brooklyn. Guess what? They know you live in Brooklyn, and they know you’re a jogger, so it’s going to surface ads that are actually useful to you. It’s the golden elixir of intent data.

I believe this is going to usher in what I’m describing as the fifth wave in ad tech. We get the chance to do it all over again.