Meta chose licensing over acquisition with Midjourney, tapping the profitable AI lab's aesthetic technology while preserving its independence. The deal signals a new model for successful AI startups to monetize expertise without surrendering control.
Apple approaches Google to power Siri's AI brain, marking a potential shift from internal control to external partnerships. The talks reveal constraints facing even tech's most valuable company when core technologies evolve faster than resources can follow.
OpenAI quietly scrapes Google Search results to power ChatGPT while positioning itself as Google's rival. The practice reveals messy dependencies behind AI's search ambitions as user behavior shifts toward longer, conversational queries.
Apple approaches Google to power Siri's AI brain, marking a potential shift from internal control to external partnerships. The talks reveal constraints facing even tech's most valuable company when core technologies evolve faster than resources can follow.
💡 TL;DR - The 30 Seconds Version
👉 Apple approached Google to develop custom AI model for redesigned Siri launching in 2026, with Google already training models for Apple's servers.
📊 Internal "bake-off" tests Apple's "Linwood" models against external "Glenwood" approach as Siri upgrade faces year-long delay from spring 2025.
🏭 Engineering setbacks forced leadership reshuffling, moving AI chief John Giannandrea away from Siri while Federighi and Rockwell take control.
💰 Anthropic was preferred partner but pricing disputes led Apple to expand search, while key AI architect left for Meta with $200 million package.
⚡ Apple's trillion-parameter model testing still trails OpenAI's several-trillion-parameter systems, highlighting competitive gap in AI capabilities.
🚀 Platform companies increasingly face build-versus-buy decisions where internal development timelines conflict with competitive market pressures.
Apple recently approached Google to develop a custom AI model for powering a redesigned Siri next year, with Google already training a version that could run on Apple's servers, according to Bloomberg's Mark Gurman. The discussions mark Apple's broadest search yet for external AI partnerships as engineering delays push the promised Siri overhaul from spring 2025 to sometime in 2026.
The talks represent a strategic inflection for a company that has traditionally maintained tight control over core AI capabilities. Apple's exploration of Google Gemini follows earlier discussions with Anthropic and OpenAI about using Claude or ChatGPT to power Siri's next iteration. The company now faces a fundamental choice between preserving its integrated approach or accepting external dependency to accelerate market competitiveness.
Apple is conducting an internal evaluation between two development paths: "Linwood," powered by Apple's own models, and "Glenwood," running on external technology. The company remains several weeks from making a final decision, with no formal commercial negotiations currently underway with any potential partner.
The original Siri upgrade was architected around Apple's Foundation Models team technology—the same group that created on-device language models powering Apple Intelligence features. Technical setbacks forced a complete restart and leadership restructuring that moved AI chief John Giannandrea away from Siri development. Software chief Craig Federighi and Vision Pro creator Mike Rockwell now oversee the project.
Anthropic emerged as the initial preferred partner, but financial terms demanded by the AI company led Apple to expand its search. The pricing dispute illustrates the negotiating leverage that leading AI companies now hold over hardware manufacturers seeking to integrate advanced capabilities.
Apple's AI team continues experiencing significant departures. Foundation Models chief architect Ruoming Pang left for Meta in July with a $200 million compensation package, taking several colleagues with him. Many remaining team members are interviewing elsewhere, driven by either concerns about third-party dependency or the AI talent market's unprecedented compensation levels.
From Apple's perspective, external partnerships provide a path to competitive parity while internal teams continue developing proprietary alternatives. The company has rarely led new technology categories initially but has historically delivered superior integrated experiences. External AI models would operate on Apple's Private Cloud Compute infrastructure using Mac chips, preserving architectural control while accessing cutting-edge capabilities.
From Google's perspective, powering Siri would expand AI partnerships beyond current Android and Samsung relationships. Google already pays billions annually to remain Apple's default search provider—a partnership under antitrust scrutiny that could face dissolution. An AI modeling partnership could establish alternative revenue streams and deeper ecosystem integration.
From the broader market's perspective, the discussions signal Apple's acknowledgment that its AI capabilities lag meaningfully behind competitors. Both Apple and Google shares rose on news of the potential partnership, with investors viewing external collaboration as preferable to continued delays in AI feature deployment.
The structural challenge runs deeper than engineering timelines. Apple faces talent retention pressures that even $200 million packages can't fully address, while external AI providers gain leverage from their technological leads.
Apple recently began testing its first trillion-parameter model, representing a substantial leap from the 150 billion parameter systems currently running in its data centers. The company remains well behind AI leaders—OpenAI employs several-trillion-parameter approaches. Parameters measure AI models' complexity and learning capacity, making this gap a concrete indicator of Apple's competitive position.
The company has no immediate plans to deploy the more powerful trillion-parameter system to customers, restricting it to research applications. This conservative approach reflects Apple's traditional emphasis on shipping refined products rather than experimental capabilities.
Some Apple software leaders have discussed replacing models used for non-Siri Apple Intelligence features with external alternatives. Such a shift would break from Apple's preferred approach of maintaining ownership over consumer-facing AI capabilities. The company isn't actively pursuing this broader outsourcing strategy.
The Siri partnership discussions occur alongside separate negotiations to integrate additional chatbots into Apple Intelligence. Apple already incorporated ChatGPT as a fallback for general knowledge queries—an area where Siri has historically underperformed. Planned Gemini integration suggests Apple is constructing a multi-provider AI ecosystem rather than committing to a single external partner.
This approach mirrors Apple's broader platform strategy of maintaining multiple supplier relationships to avoid vendor dependence. In AI, however, the technical complexity of model integration and the specialized nature of leading capabilities limit Apple's traditional negotiating leverage.
The potential Google partnership also carries geopolitical dimensions. As AI capabilities increasingly influence national competitiveness, Apple's reliance on external providers—whether American companies like OpenAI or international ones—could complicate its global market access and regulatory relationships.
In recent company meetings, CEO Tim Cook emphasized that Apple must succeed in AI and is accelerating investments. He noted that Apple rarely leads new markets initially but eventually provides superior products. The external partnership exploration tests this philosophy against market realities that may not accommodate Apple's traditional development timeline.
Apple's shift from internal development to competitor partnerships signals a fundamental recalibration. When a company that has built its competitive advantage on vertical integration starts shopping for external AI models, the market has moved faster than even deep resources can follow.
Why this matters:
• Platform companies increasingly face build-versus-buy decisions in AI where internal development timelines conflict with competitive pressures, potentially reshaping how integrated technology experiences are created
• Market concentration in AI capabilities forces traditional technology leaders to reconsider vertical integration strategies that previously defined their competitive advantages
Q: What exactly is Apple's "bake-off" approach between Linwood and Glenwood?
A: Apple is simultaneously developing two versions of next-gen Siri. "Linwood" runs on Apple's own AI models, while "Glenwood" uses external technology from partners like Google, Anthropic, or OpenAI. The company will test both approaches before choosing which one ships to customers.
Q: How much does Google currently pay Apple for their search partnership?
A: Google pays Apple billions of dollars annually to remain the default search engine on iPhones and other Apple devices. The exact amount isn't disclosed, but estimates range from $15-20 billion per year, making it one of Apple's largest service revenue sources.
Q: What is Apple's Private Cloud Compute and why would external AI models run there?
A: Private Cloud Compute is Apple's server infrastructure that uses Mac chips for remote AI processing. External AI models would run on these Apple-controlled servers rather than directly on devices, allowing Apple to maintain some security and privacy control while accessing advanced AI capabilities.
Q: How far behind is Apple in AI model size compared to competitors?
A: Apple's current production models use 150 billion parameters, with new trillion-parameter models in testing. OpenAI uses several-trillion-parameter systems for ChatGPT. Parameters measure model complexity and learning capacity, so Apple trails by roughly 3-5x in computational scale.
Q: When will the new AI-powered Siri actually launch to consumers?
A: Apple is targeting spring 2026 for the revamped Siri launch, representing a full year delay from the original spring 2025 timeline. The company needs several more weeks to decide between internal and external AI approaches before finalizing development and testing schedules.
Maria Garcia
Marcus Schuler
Meta chose licensing over acquisition with Midjourney, tapping the profitable AI lab's aesthetic technology while preserving its independence. The deal signals a new model for successful AI startups to monetize expertise without surrendering control.
OpenAI quietly scrapes Google Search results to power ChatGPT while positioning itself as Google's rival. The practice reveals messy dependencies behind AI's search ambitions as user behavior shifts toward longer, conversational queries.
Google breaks AI energy silence with first granular data: 0.24 watt-hours per Gemini prompt. But critics challenge what's excluded from the math, setting up a new fight over measuring AI's true environmental cost.
Court filings reveal Elon Musk approached Mark Zuckerberg about financing a $97.4B bid to acquire OpenAI earlier this year. Meta declined, but the approach signals how AI competition drives unlikely alliances in Silicon Valley.
Get the 5-minute Silicon Valley AI briefing, every weekday morning — free.
