In early July, Paul Meade left Apple to join OpenAI.

Having worked at Apple for 15 years, Meade rose from iPad project manager to head of Vision Pro hardware engineering, overseeing Apple's planned 2027 launch of screenless smart glasses. Over the past year, OpenAI has recruited more than 40 engineers and executives from Apple's hardware division, spanning roles in camera engineering, iPhone hardware, Mac hardware, chip development, industrial design, manufacturing, audio, smartwatches, and Vision Pro. Combined with earlier hires like Jony Ive, Tang Tan, and Evans Hankey, a significant portion of Apple's core hardware-defining team from the past decade has now shifted to OpenAI.

While this appears to be a talent war on the surface, it reflects a deeper industry shift: AI companies centered on model capabilities are systematically building out hardware expertise.

I. OpenAI's Hardware Strategy: Extending from "Cloud" to "Edge"

OpenAI's ChatGPT boasts over 500 million monthly active users, making it the world's most widely used AI application. However, a fundamental reality remains: users access ChatGPT through Apple's iPhone, Google's Android devices, or Microsoft's Windows systems.

This means that regardless of model strength, user experience is constrained by third-party platforms' hardware specifications, interaction logic, and ecosystem rules. Apple has integrated "Apple Intelligence" into iPhones, Google has deeply embedded Gemini into Android, and Microsoft has incorporated Copilot into Windows—platform owners are transforming AI into a "feature module" within their ecosystems.

For OpenAI, this poses a structural risk: if AI remains confined as an "app" on others' devices, its growth will be capped by app store distribution rules and platform revenue sharing. More immediately, platform owners could replace third-party services with in-house models—Apple with Apple Intelligence, Google with Gemini, Microsoft with Copilot.

OpenAI's public response: Build its own hardware.

In May 2025, OpenAI acquired Jony Ive's hardware startup io in a $6.5 billion all-stock deal, integrating 55 hardware engineers, software developers, and manufacturing experts. Combined with the 40+ Apple hires over the past year, OpenAI's hardware division now approaches 100 personnel, with expertise spanning industrial design, hardware engineering, chip development, manufacturing, audio, cameras, and spatial computing—covering the entire consumer electronics value chain.

In June 2025, OpenAI partnered with Broadcom to launch Jalapeño, its first custom AI inference chip for data centers. Developed in approximately nine months from announcement to tape-out, the move demonstrates OpenAI's rapid progress in establishing foundational silicon capabilities.

Models, chips, devices—within 18 months, OpenAI has assembled all critical components needed for a hardware company.

II. Apple's Talent Exodus: Leadership Transition and Strategic Uncertainty

Paul Meade's departure was triggered by clear internal changes.

In April 2025, Apple announced Tim Cook would step down as CEO in September, with hardware engineering SVP John Ternus succeeding him. Following Ternus's promotion, chip chief Johny Srouji took over the hardware engineering division, restructuring reporting lines: several hardware VPs who previously reported directly to Ternus now answer to newly appointed middle manager Tom Marieb.

According to Bloomberg, this adjustment reduced decision-making authority for some VPs, including Meade. OpenAI's opportunity was described as "catching someone who had already decided to leave" rather than mere high-salary poaching.

Apple has countered with retention efforts, offering iPhone product design teams up to $400,000 in restricted stock bonuses. However, the need for such incentives underscores that talent attrition has reached core team levels.

Deeper discussions focus on Apple's AI strategic pace. When Jony Ive left Apple in 2019, he cited founding independent design firm LoveFrom, but industry observers have long noted potential disagreements over Apple's AI strategy. Subsequently, Apple Intelligence lead Ke Yang departed for Meta in October 2025, followed by Human Interface Design VP Alan Dye joining Meta in December. Three VP-level departures in six months far exceed Apple's typical external VP attrition rate over the past decade.

These developments suggest Apple's appeal to top hardware talent is experiencing a temporary decline.

III. Path Dependency from Hardware Success

An overlooked angle in Apple's talent losses is that its hardware has been too successful, creating path dependency.

For 30 years, Apple's core strength has been packaging complex technologies into elegant consumer electronics. The iPhone, Mac, AirPods, and Apple Watch each exemplify a closed-loop ecosystem of "hardware + software + services." But this closed loop (closed loop) presupposes Apple controls everything.

In the AI era, this presumption imposes constraints. With billions of iPhone users, mature interaction paradigms, and strict privacy policies, any radical AI transformations must occur within existing frameworks, limiting disruptive experimentation. Apple Intelligence's "conservative" reputation among some users may stem not from technical limitations but from Apple's inability to risk user experience on its flagship cash cow.

Vision Pro was expected to pioneer AI-era devices but fell short of sales expectations. Apple now bets on screenless smart glasses, with Meade leading the project—his departure during its critical phase substantially impacts Apple's hardware roadmap.

Conversely, OpenAI carries no legacy hardware baggage. Ive and Sam Altman's first AI device—a $200–300 camera-equipped smart speaker with no screen, focusing on ambient interaction—would face questions like "how to integrate iOS ecosystems" or "maintain brand premium" if developed internally at Apple. For OpenAI, the sole question is: Can this device make large models' capabilities tangible to users?

Freed from legacy hardware obligations, OpenAI gains freedom to define new form factors—an opportunity window and structural challenge for Apple.

IV. An Industry Signal Spreading

OpenAI is not alone in bolstering hardware capabilities.

Anthropic has publicly recruited hardware engineers, while Meta has made AI glasses a core strategy, iterating its Ray-Ban collaboration. Chinese large model firms are also accelerating "edge deployment" and AI hardware exploration. The closed-loop capability of "models + hardware" is emerging as a new competitive dimension in AI.

This trend carries direct implications:

1. Pure model companies face thinning competitive barriers. Gaps between leading models like GPT-5, Claude, and DeepSeek are narrowing, with models commoditizing. Future differentiation may lie not in parameter scale but in "how models reach users, collect data, and form loops through hardware."

2. Talent with AI and hardware expertise is entering a premium window. OpenAI's compensation packages for Apple hardware engineers reportedly exceed previous levels, combined with equity and the career appeal of "defining next-gen devices from scratch," creating a siphon effect on traditional hardware giants.

3. Consumer electronics' defining logic may shift. For 30 years, hardware firms shaped device forms while software adapted. Over the next decade, model companies may instead define "what hardware forms AI needs," with hardware design optimizing around model capabilities rather than models serving as hardware features.

V. 2027: The Pivotal Year for Form Factor Competition

Ive and Altman's first AI device, originally slated for late 2026, has delayed to early 2027. Analyst Ming-Chi Kuo's supply chain reports note OpenAI is concurrently developing an "AI agent phone," projected to reach 30 million units shipped by 2027–2028.

2027 will also see Apple's screenless smart glasses and potentially Meta's third-gen AI glasses launch. The battle over next-gen computing device forms will intensify that year.

But beneath the form factor competition lies a power struggle: Who gets to define "what AI devices should look like"?

Apple argues devices must prioritize user experience, craftsmanship, and ecosystems. OpenAI contends devices should maximize model capability expression. Neither logic is inherently superior, but history shows the party controlling core production resources typically sets definitions.

In the PC era, the core resource was operating systems. In mobile, it was app ecosystems. In AI, it's models—but if models always run on third-party devices, model firms remain ecosystem participants, not rule-makers.

OpenAI's hardware strategy essentially seeks to transition from "participant" to "rule-maker."

Epilogue

Apple remains the world's most valuable company. Its iPhone, Apple Watch, and AirPods lines remain robust, with supply chain and brand moats intact for the near term.

However, as its top hardware talent continues flowing to OpenAI, a critical question arises: Can Apple's 30-year hardware expertise remain a competitive moat if the logic defining next-gen devices shifts?

The answer won't emerge from CEO transition speeches.

It will appear in 2027, through user choices after companies' AI devices collectively launch.

Disclaimer: This article represents financial commentary from Zhihu and does not constitute investment advice. All corporate data and regulatory events cited are from public information and for reference only; official announcements shall prevail. Image sources are networked; please contact for copyright removal.

AI Financial Insights Daily