While OpenAI and Anthropic engage in direct model competition, Apple has opted to leverage its ecosystem to capture value across the board. This suggests a new form of "Apple tax" is on the horizon.
According to a report by Bloomberg's Mark Gurman citing internal sources, a major shift coming with iOS 27 at the WWDC on June 8, 2026, will be the transformation of Siri into an open platform through the introduction of a key feature called "Siri Extensions."
In the future, users will be able to send query requests directly from the Siri voice assistant to any competing AI chatbot installed via the App Store. Services like Google Gemini and Anthropic's Claude will gain access equal to Apple's native services.
This indicates Apple intends to replicate the successful logic of the App Store—by establishing a de facto "AI App Store" and levying a commission of up to 30% on third-party AI subscription services transacted through its platform.
Historically, Apple's software strategy has always been subordinate to hardware sales, a fact reflected most clearly in its financial reports. In the first quarter of fiscal year 2026, iPhone revenue reached $8.527 billion, a 23% increase year-over-year, accounting for approximately 59.3% of total revenue. Combined with Mac, iPad, and wearables, Apple's total hardware revenue reached $11.37 billion, constituting about 79% of total revenue.
From an external perspective, the primary purpose of native applications like iMessage, Maps, and Photos is to enhance device stickiness and prevent user migration to competing ecosystems, even though these services also generate tens of billions in revenue annually.
The competitive logic in the AI field is fundamentally different. Major model providers like ChatGPT and Google Gemini are transitioning towards platform and tool-oriented models through subscriptions.
Unable to offer a cutting-edge model itself, Apple is compelled to participate in the AI era through a different approach: the "Siri Extensions" service, which allows users to freely switch between AI models.
Offering a sufficiently rich array of third-party AI services to prevent user churn is crucial for protecting the foundation of hardware sales—arguably more important than proving "Siri is smarter than Gemini." Furthermore, generating revenue through service subscription commissions opens up a highly promising incremental income stream.
Apple is already sharing revenue from ChatGPT Plus subscriptions by providing payment and subscription systems for OpenAI. In the Chinese market, companies including Alibaba and Baidu have been linked with potential partnerships with Apple.
It is worth noting that localized cooperation between Alibaba, Baidu, and Apple's Siri has been slow to materialize, potentially due to coordination with the unified rollout plan for the "Siri Extensions" service.
To some extent, Apple is competing on a technical level in the model race, but with Siri's integration of third-party AI services, its strategy shifts to an ecosystem battle, built upon a vast ecosystem of billions of activated devices.
During a previous earnings call, Apple disclosed a key figure: the global installed base of active devices, including iPhones, exceeds 2.5 billion.
Therefore, if one must distinguish between developing a proprietary model and integrating third-party services, it can be seen as a choice between two different financial and valuation models. Ultimately, however, investors want a model that generates profit.
Apple's fundamental overhaul of Siri also adopts an open approach by embracing Google—accessing the full Gemini model in its own data centers and using it to distill smaller models capable of running on-device.
Gurman has also confirmed that Apple is utilizing Gemini's technology and engineering teams to rebuild its own foundational model.
However, it must be emphasized that "openness" has never been a defining characteristic of the Apple ecosystem.
While this move is understandable in the context of the Cook era, under Steve Jobs' leadership within the closed iOS and macOS ecosystems, one could have found countless reasons to reject such openness: it would compromise our user experience, it poses risks to user privacy and security, and so forth.
Apple is quietly stepping out from behind the walls of its closed ecosystem not because experience and privacy are now guaranteed, but because its own models are not yet competitive.
Senior analyst Ming-Chi Kuo pointed out in an interview in November 2025 that Apple's on-device models would struggle to catch up with cloud-based large models in handling complex logic in the short term. To ensure the user experience of Apple Intelligence in 2026 does not fall too far behind competitors, Apple is introducing Google Gemini technology to rebuild Siri's underlying infrastructure.
Gurman compared this process to "improving Windows PCs with Apple chip technology"—leveraging mature external technology to build a deeply integrated intelligent system that is truly Apple's own.
As mentioned earlier, Apple's choice to integrate third-party services in the AI era, even incorporating Gemini at Siri's core, is essentially driven by the "ecosystem fortress" built by billions of activated devices.
In the view of seasoned observers like Gurman, AI's role within the Apple ecosystem is more akin to a high-performance "lubricant"—finding new, indispensable reasons for consumers to purchase expensive hardware.
Driven by the AI wave, Apple's hardware lineup is undergoing a profound upgrade. The biggest suspense this year surrounds the long-rumored foldable iPhone.
Kuo believes the significance of a foldable iPhone extends far beyond form factor innovation; it lies in providing the "large canvas" urgently needed in the AI era. For professional users requiring AI-assisted writing, deep image editing, or video generation on mobile devices, the screen size of traditional phones has become an efficiency bottleneck.
However, before foldable screens see mass adoption, Apple first needs to bolster the AI computing foundation within its existing product lines. In other words, the foldable form factor is an upgrade in shape, while computational power is the prerequisite; both must advance simultaneously.
To this end, Apple is proactively strengthening the hardware capabilities of its traditional slate iPhones. Kuo revealed that production estimates for the iPhone 17 series are nearing 96 million units, approximately 10% higher than the 16 series. The Pro Max models are expected to come standard with 12GB RAM across the board to support more complex multimodal AI tasks. This generation is essentially laying the groundwork for subsequent, more radical form factors, including foldables.
Upon this foundation, the foldable iPhone would truly have the conditions for volume sales. Priced potentially over $2,000, this flagship model is not just a form factor innovation; it is also expected to boost the overall Average Selling Price (ASP)凭借其 enhanced AI capabilities and broader use cases, becoming a significant profit driver for Apple in the next phase.
Beyond phones, Apple is deploying AI across a wider physical world. Gurman revealed that a lightweight smart glasses product without AR functionality is expected to be released in the first half of 2026, aiming to capture the visual interaction interface.
Simultaneously, a secretive Apple robotics team is researching various robotic products, with the first possibly being a desktop smart display equipped with a robotic arm, usable for scenarios like video calls and home security.
Different types of hardware collectively form an "ecosystem fortress." The services users inside this fortress can experience depend on which external services Apple chooses to open a gate for, allowing them in, and then charging a toll for each service admitted.
Observe this, and it becomes clear: this is the new "Apple tax" of the AI era.
At the 2024 Apple WWDC, besides the debut of Apple Intelligence, a more memorable impression was the personal attendance of OpenAI CEO Sam Altman.
At that time, Altman was the undisputed rising star of the industry—the launch of GPT-4 in late November the previous year had directly propelled the industry into the ChatBox era.
Altman's presence foreshadowed the subsequent cooperation between OpenAI and Apple on Siri, and even more so, the upcoming Siri Extensions: shifting from OpenAI having exclusive access to traffic, to everyone having the opportunity to convert traffic from the Apple ecosystem, with Apple holding the "priority of choice."
But this is only Apple's first step, and the most easily understandable one, given its lucrative experience with search engine defaults on iOS. More critical is the cooperation between Siri and Gemini. This venture extends beyond conventional gateway and traffic business. Regardless of the freedom Apple grants in server setup or the security safeguards and model distillation applied to Gemini, the underlying technology retains its Google essence—the "external monk" mentioned earlier.
From providing access points to controlling the underlying infrastructure, the strategy offers users more choices while creating cash flow for Apple. While the common adage is to "trade time for space," regarding Siri's "connection" with AI, the summary here is: trading space for time—buying enough time to catch up with frontier models.
Standing in 2024, Altman was the clear winner. But standing in 2026, OpenAI and Altman might feel not victory, but a sense of betrayal. There can be only one true winner: the creator of the ecosystem and the maker of the rules—Apple.
Regarding Apple's series of phased strategies, Ming-Chi Kuo believes Apple's "remarkable performance" will begin from 2027. The substantial construction and deployment of Apple's self-built AI servers are also expected to make significant progress after 2027. For 2026, aside from the foldable phone, there might not be many groundbreaking updates. However, Apple cannot afford to underperform for two consecutive years. The revamp of Siri at WWDC will be a critical step in demonstrating its resolve.
Comments