Google masters two major technological levers: the most used search engine in the world and one of the most advanced artificial intelligence systems. However, the company is in a strategic impasse. By fully integrating the generative AI to its engine, it threatens the main source of its turnover: advertising.
In an exchange with Peter Diamandis, Kai-Fu Lee, former president of Google China, sums up the problem: “Google will not manage to make the tilting, as any innovator confronted with the dilemma of the innovator. »»
The impossibility of a turn to the single response
The founding principle of Google, formulated by Larry Page, aimed to offer a single, correct and immediate answer to any question. What major language models now make it possible to do on a large scale. But such an interface eliminates the space reserved for sponsored links, the base of the current economic model.
The cost of inference of a model like GPT-4 for a request is estimated at 10 cents.
The average income generated by Google research is 1.6 cent.
The disproportion is obvious. An AI engine would cause immediate financial loss, without a proven alternative model.
A structural lock
Google’s position on the financial markets prevents any sudden rupture. A switch to an IA-FIRST search engine would cause a fall in turnover, stock market devaluation and a shareholder distrust. And as Kai-Fu Lee points out, “it is not a transition that a listed company can afford”. The management is therefore trapped between The need to innovate and the impossibility of compromising short -term results.
An opening for new entrants
Actors like Perplexity or Bego adopt a reverse strategy: from the outset a search engine based on the generative AI, without dependence on advertising. Their absence of economic legacy allows them to move forward quickly. They bet on a refined, fast experience, based on a single response, possibly enriched with factual sources.
A question of governance more than technology
Google has technical, human and financial resources to remain a leader. But without reinvention of the economic model, the company may find itself exceeded not by the quality of the AI, but by the rigidity of its decision -making structure.
“Having the best technology is not enough if your internal incentives are discontinued,” recalls Kai-Fu Lee.
Google’s case illustrates the limit of a model maturity. The company faces a strategic alternative that other actors have been faced with Kodak and Coners: sacrificing part of its income to invest in a future aligned with emerging uses, or protect its model at the risk of being overcome by more agile actors.
Answers to come.
