Apple is reportedly considering a radical departure from its tradition of in-house development, exploring the use of AI models from rivals OpenAI and Anthropic to power the next generation of Siri. This potential pivot, reported by Bloomberg, suggests Apple may outsource the core intelligence of its voice assistant, a move that underscores the immense pressure the company faces in the AI race.
According to the report, Apple has asked both AI firms to develop and train versions of their large language models that can run on its cloud infrastructure for internal testing.
This news represents a potential admission that its own multi-year effort to modernize the assistant has fallen short. The project to graft generative AI capabilities onto Siri’s notoriously brittle and aging foundation was described by insiders as a “wreck,” with one employee noting that fixing a single bug would often cause three more to appear. This forced the company into a complete architectural overhaul.
A Foundational Reset and a History of Missteps
The decision to start over was seen as unavoidable by some industry observers. John Gruber of Daring Fireball argued that “rebuilding Siri on a new architecture was the ‘only way forward, even if it’s embarrassing in the short term.'” This public admission of failure culminated in Apple confirming that its most advanced Siri features—those enabling deep personal context and on-screen awareness—would be delayed until at least spring 2026, arriving on a new “V2” architecture.
The path to this point has been fraught with public missteps that have tarnished Siri’s reputation. These include embarrassing functional glitches, such as the assistant creating “phantom” restaurant reservations for users in late 2024. More seriously, Apple agreed to a $95 million legal settlement in early 2025 over allegations that Siri had made hidden recordings of user conversations without consent.
Rivals Race Ahead in a Crowded Field
While Apple has been mired in this foundational rebuild, its competitors have been aggressively pushing forward. Amazon’s generative AI-powered Alexa+ has already been rolled out to over one million early access users and is pioneering a new subscription model. Google has also been steadily integrating its conversational Gemini Live assistant across its ecosystem, from Android devices to Google Workspace accounts.
The competitive pressure is intensifying across the board. AI-native companies are entering the fray directly, with Anthropic launching a voice mode for its Claude AI and Perplexity securing pre-installation partnerships with smartphone makers. This creates a stark contrast in momentum and puts Apple in a difficult game of catch-up.
The Privacy Gambit and Geopolitical Headwinds
This intense pressure forces Apple to confront its core strategic dilemma. The company has built its brand on an unwavering commitment to user privacy, architecting its Apple Intelligence system around on-device processing and a unique “Private Cloud Compute” infrastructure. This system is designed to be ‘stateless,’ processing complex requests on secure Apple Silicon servers without storing user data.
However, this privacy-first approach can be a significant constraint in the world of large language models, which typically achieve greater power by training on vast, cloud-based datasets. By considering models from OpenAI and Anthropic, Apple may be signaling a pragmatic compromise. The potential move can be seen as a necessary compromise and an act of AI pragmatism, allowing Apple to leapfrog its internal development hurdles.
The challenge is not just technical but also geopolitical. In China, a market that represents nearly a fifth of its sales, Apple is required by local regulations to partner with a domestic firm for AI services. Its exploration of a deal with Alibaba has already drawn intense scrutiny from Washington, with U.S. officials expressing deep concern over national security.
Greg Allen of the Center for Strategic and International Studies put the government’s position bluntly, stating, “The United States is in an AI race with China, and we just don’t want American companies helping Chinese companies run faster.” Representative Raja Krishnamoorthi, a House Intelligence Committee member, called the potential deal “extremely disturbing.”
A Two-Pronged Strategy
Despite the high-profile delays with Siri’s most ambitious features, Apple’s broader AI strategy is not on hold. At its recent WWDC 2025 conference, the company unveiled a suite of more practical “Apple Intelligence” tools, such as Live Translation, which will ship this fall. In what may be its most significant long-term move, Apple is also empowering its massive developer community.
Software chief Craig Federighi announced that Apple is taking a “huge step” by giving developers access to its on-device foundation models. He explained, “We’re also taking the huge step of giving developers direct access to the on-device foundation model powering Apple Intelligence…”, allowing them to build powerful, private, and offline-capable intelligent apps. This strategy plays to Apple’s core strengths: its integrated ecosystem and a developer base eager to innovate on its platform.
Ultimately, Apple finds itself at a crossroads, navigating what technology analyst Jennifer Wong has termed a “a technology cold war”. The decision to potentially partner with its AI rivals for Siri’s brain is a clear admission of its internal struggles.
However, by simultaneously doubling down on its developer ecosystem and its unique privacy architecture, Apple is betting that a superior, trustworthy, and seamlessly integrated user experience will prove more valuable than winning the raw AI model race. The success of this multi-pronged strategy will determine if Apple can finally deliver the intelligent assistant it has long promised.
