Cannes is watching creative. The money is in media. And media is concentrating, not fragmenting.

Cannes used to be a cadre of French, Gauloises-smoking, rosé-drinking creatives, very parochial. It is still too Anglo-Saxon and too European, and it does not give APAC or LATAM enough prominence. About half of S4's 6,200 people are in Latin America, and the creative there, in Buenos Aires, is as good as anywhere, driven by the Boca and River rivalry.

Cannes has become tech-dominated. WPP comes off the beach and PMG, a data-driven media company, goes on it. Michael Kasman has made Cannes more like CES and CES more like Cannes, and over the next five years it will only get more concentrated.

The focus this year is the impact of AI on creative, and that is where it is going wrong, because it should be on media. We said AI wow one year, AI how the next, and we hope it will be AI now. That question mark is the problem.

Digital is 900 billion of a 1.2 trillion industry, and the shrinking 300 billion is driving deals like Omnicom buying IPG. The agencies say media is fragmenting. It is not, it is concentrating. Google is 300 billion, Meta 200 billion, Amazon heading to 100 billion, TikTok outside China 40, already more than half of all spend.

The hyperscalers will spend 5 trillion on AI infrastructure over five years, reinforcing positions that will not go away, and they are government-supported. Arthur Sadoun has done a brilliant job, but the profit is driven by media, and once media gets squeezed on take rates, as the Trade Desk episode showed, that is the real exposure.

There is real compression in visualisation and copywriting, faster and cheaper. If you are S4 or Monks with no traditional production, it is not a threat. If you are WPP with about 40 per cent of revenue from creative, you are disintermediating yourself.

Personalisation at scale raises the general level, but as James Quincy of Coca-Cola asked, if everybody is personalising, how do I make Coke different? So the human dimension matters more than ever, at least until machines become more powerful than people, which Eric Schmidt put at three years.

Claims that agencies are winning work on value-based, outcome-based pricing are BS. People are not winning on that basis yet. Marketing is keen, procurement is nervous and wants its safety blanket. Company profits are strong, yet the agencies have not prospered because the industry is flat and the 300 billion is shrinking. Unit and output pricing will come, but procurement resists change.

Companies do not move until they face an existential threat. Autos have it because of Chinese EVs, banks because of fintech, and packaged goods firms like General Mills and SC Johnson are squeezed on price and commodity costs. It was the existential threat that forced one company to transform.

After now, quantum is the next thing, the compute power to deal with health and climate. I have seen Google's labs in Santa Barbara, and these are early waves. If the industry thinks AI is disruptive, standby.

Every brand needs influencers, from the top 1 per cent to a Paris Hilton. But I saw a prospectus for a YouTube-heavy company with 30 million of revenue, 20 million of it payments to influencers, net revenue of 10, profit of 4. The money is flowing to the influencers, and the consumer is not a moron. You can only sell a bad product once.

Sponsorship and experience matter enormously. We are making the official FIFA World Cup film with Disney. With five years of football in America, soccer has replaced ice hockey among the top sports, and World Cup audiences rival the Super Bowl. FIFA is on top, F1 is doing very well, and the IOC has a chance to catch up.

It is really tough for CMOs. Very few become CEOs, it tends to be the ex-CFOs, and Google's Sean Downey says they are spending more time with CFOs. One new CEO asked his CMO why they were spending on upper funnel, the Olympics, F1, FIFA, when he has to report growth and margin every quarter. The problem for marketing is justifying its existence.

Agencies have to justify theirs too. We are validators, you cannot mark your own homework, which is why Publicis buys data and why owning verification is hard, as we found with Millward Brown and Kantar. Our venture fund backs Verido, which has helped move Australian outdoor share from 11 to 14 per cent. Eventually algorithmic analysis with quantum, blockchain and AI will make verification easier. For now it is extremely tough.