Apple and Google’s New Venture: A Familiar Narrative
Apple Inc.’s latest collaboration with Google LLC to utilize Gemini as the cornerstone for Apple Intelligence, encompassing Siri, may evoke a sense of déjà vu. This scenario is reminiscent of similar agreements we have witnessed in the past, with predictable outcomes.
In 2016, Google entered into a comparable arrangement with Apple known as the Google–Apple Internet Services Agreement, or ISA, which established Google as the default search engine on Apple devices.
Although users ostensibly had the option to explore alternative search engines or download different browsers, the reality proved contrary.
A federal judge determined in 2024 that such distribution agreements were anticompetitive, as the default settings engendered foreclosure effects within the market.
The Gemini initiative appears to replicate this model. It places Gemini at key access points within Apple’s ecosystem, notably Siri.
Initial reports indicate that the agreement may not be exclusive, suggesting that Apple could engage with other foundational models and that consumers might still opt to download various chatbots or browsers integrating different foundation models.
However, this was also a feature of earlier search agreements. The essential takeaway from the Google search case is straightforward: in the realm of digital markets, defaults wield more influence than nominal exclusivity.
Once established, such defaults become remarkably entrenched. Evidence from the Google case illustrated that even when Microsoft Corp. proposed to share 100% of Bing’s search revenue, Apple ultimately concluded that no financial incentive sufficed to warrant a switch from Google.
Judge Amit Mehta, in his ruling, dismissed the notion that a contract must entirely preclude alternatives to be deemed exclusive in effect.
Drawing on the Microsoft case, he articulated that merely closing off a substantial portion of distribution opportunities suffices. The Apple–Gemini deal risks reproducing this dynamic concerning AI distribution.
Being the foundational model that powers Apple’s ecosystem could potentially yield even greater value than being the default search engine on Apple devices. Transitioning away from a default search engine or downloading a different browser remains within the realm of possibility.
Conversely, it remains uncertain whether users will be able to alter the underlying foundation model that empowers their iPhone’s Intelligence features. If changing this model is possible, such a process may involve more friction than switching search engines.
Some distinctions do exist between the ISA and the Apple–Gemini partnership. While Safari served as a critical gateway to the internet, Siri and other Apple Intelligence features are not yet the predominant medium for AI-assisted inquiries, as a majority still utilize standalone applications or browsers to access AI tools.
Nevertheless, as AI chatbots and agents become increasingly sophisticated and ubiquitous, the significance of voice assistants and operating system-level AI features may grow.
As long as users remain tied to iPhones and MacBooks as their primary devices, whichever foundational model underpins Apple’s ecosystem will secure a considerable distribution advantage.
Even amidst shifts in consumer habits, which may see chatbots surpass search engines as the primary internet gateway, the prevailing hardware ecosystem could still play a pivotal role in determining the victor among chatbots or AI assistants.
Proponents of the deal might contend that this constitutes mere product design—similar to Apple’s choice of its camera lens supplier. Users do not select those components; why should the same not apply to AI models?
The response hinges on our understanding of what foundational models represent. If viewed solely as components, Apple should indeed have broad latitude.
However, if they are perceived as general-purpose infrastructure akin to search engines, operating systems, or app stores, then default control assumes competitive ramifications.
Considering the considerable influence these models exert on information access, creativity, and economic prospects, the latter viewpoint is difficult to dismiss.
The arrangement also underscores a broader dilemma regarding the efficacy of antitrust measures in rapidly evolving technological marketplaces. Section 2 of the Sherman Act necessitates a demonstration of anticompetitive impacts.
At present, the foundation models market appears robust and competitive (with players like ChatGPT, Gemini, and Claude), meaning clear anticompetitive effects associated with the Apple–Gemini collaboration remain elusive.
However, once a single entity gains control over a crucial distribution channel, competition may dissipate rapidly. Yet such effects often emerge only after defaults have solidified their positions.
We find ourselves ensnared in an antitrust catch-22: Misconduct is challenging to contest before markets reach a tipping point and arduous to rectify afterward. By the time monopolistic behavior becomes evident and quantifiable, reversing such patterns is often implausible.
The Google search case aptly illustrated this asymmetry. Following years of litigation and a liability determination, remedies proved modest and anticlimactic, leaving Google’s search dominance largely unscathed.
Simultaneously, Google’s inherent advantages are progressively translating into the realm of AI. The emergence of entities like OpenAI and Anthropic PBC in recent years was sometimes heralded as evidence of a thriving competitive landscape, suggesting that established tech leaders could indeed face challenges.
Yet, such narratives underestimated the intrinsic advantages held by established players. Google brings to Gemini a wealth of publisher data aggregated from its search index, seamless integration into search results via AI overviews, Android as a launchpad, and now, Apple’s distribution channels.
Moreover, the Gemini arrangement alters Apple’s strategic incentives. If Apple has access to a foundational model via Google, will it be motivated to invest in developing or collaborating with a rival model?
This dynamic is not novel. In the search case, the court found that the distribution agreement with Google diminished Apple’s incentives to cultivate its own search engine. It effectively functioned as a non-aggression pact between titans, shielding each from competitive pressures posed by the other.
Having missed the initial wave of AI advancements, Apple may now perceive it as more efficient to integrate Google’s models rather than strive to create its own.
Although this may represent a logical business decision, it fundamentally reinforces a troubling trajectory wherein dominant firms fortify their respective moats and share monopoly profits instead of competing to dismantle them.
Source link: News.bloomberglaw.com.
