Gurman on Apple, OpenAI’s Deal Terms, Forbes vs. Perplexity

Episode Summary

Executive Summary: The episode centers on Apple’s WWDC and its AI strategy, especially the zero-cash OpenAI integration and Apple’s likely longer-term pivot to Google Gemini. The hosts argue Apple’s AI is more about polished integration than breakthrough innovation, but still a meaningful catch-up moment. They also debate Perplexity’s use of Forbes content and Elon Musk’s reinstated pay package, framing both as flashpoints in the ongoing fight over AI economics, IP, and corporate control.

Main Topics: Apple’s WWDC and Apple Intelligence (Priority: 5/5): The hosts assess Apple’s AI announcement as comprehensive and well-integrated, but not a leapfrog moment. Gurman argues Apple had to ship an AI story to avoid lagging behind, even if most features are catch-up rather than novel. OpenAI–Apple deal economics (Priority: 5/5): A major theme is that Apple is not paying OpenAI and OpenAI is not paying Apple; the arrangement is portrayed as a barter of distribution for model access and future subscription upside. The discussion emphasizes how unusual and strategically limited the deal is. Google Gemini as the likely future default (Priority: 4/5): Gurman argues OpenAI is likely a placeholder and that Google Gemini could eventually become the default chatbot/search AI on iPhone, mirroring the existing Google search deal and preserving Apple’s search revenue. Perplexity vs. Forbes and AI scraping (Priority: 4/5): The conversation explores whether AI products can summarize and repurpose paywalled journalism without undermining the economic incentives for original reporting. The hosts note the tension between value-added aggregation and content appropriation. Elon Musk’s Tesla pay package approval (Priority: 4/5): Ranjan frames the reinstated pay package as both a victory for Musk and a troubling precedent for governance, highlighting dilution concerns, board oversight issues, and the stock’s dependence on Elon’s persona. AI, product integration, and future devices (Priority: 3/5): Gurman suggests Apple’s current AI work is foundational for future platforms like robotics, smarter wearables, and AR devices, with AI enabling a new generation of device categories beyond the iPhone.

Key Arguments: Apple’s AI rollout matters because it meets market expectations and gives Apple a credible AI story, even if it does not innovate ahead of rivals. The OpenAI integration is strategically more valuable to Apple than to OpenAI in the short term because Apple gains a chatbot without paying cash while OpenAI gains distribution to hundreds of millions of iPhone users. Google is still the more likely long-term default AI partner for Apple because it already pays Apple enormous search revenue and has the business model to support a new rev-share deal. Apple Intelligence may not create an immediate iPhone supercycle, but it could help future hardware cycles if Apple keeps expanding AI features over the next several years. Perplexity’s use of Forbes material illustrates a broader conflict: AI tools need content to function, but widespread summarization of paywalled work could weaken incentives for original reporting. Tesla shareholders approving Elon’s package reflects the market’s dependence on Musk, but the hosts criticize the governance process and the dilution it imposes on other shareholders.

Data Points: Apple's Google search revenue deal: $8 billion to $15 billion per year - Gurman cites Apple’s current search revenue stream and compares it to future AI rev-share possibilities. Google payment to Apple for search default: About $20 billion per year - Used to illustrate why Google has strong incentive to preserve or replace its iPhone default position. OpenAI subscription price: $20 per month - Discussed as the paid ChatGPT tier Apple users could subscribe to through the iPhone. iPhone AI hardware requirement: 8 GB RAM - Gurman says Apple Intelligence is limited to devices with 8 GB of RAM and newer chips. Apple stock reaction to WWDC: Down about 2% initially, then up about 5% and 3% in the following days - The hosts cite a strong post-event rally after an underwhelming first-day market reaction. OpenAI users: 100 million within two months; whisper number of 200 million today - Ranjan uses these figures to argue that OpenAI still has limited consumer reach relative to its valuation narrative. Tesla pay package size: $56 billion at peak valuation; about $46 billion at the time discussed - The reinstated 2018 compensation package for Elon Musk was described as one of the largest in corporate history. Tesla shares in pay package: 304 million shares - The package grants Musk options amounting to about 12% of outstanding shares. OpenAI reported revenue run rate: $1.6 billion in late 2023 - Ranjan questions whether this old run rate supports current optimism around OpenAI’s business model. OpenAI projected revenue: $3.4 billion - Mentioned as a widely cited figure that may be extrapolated from outdated data. Forbes referral traffic share: 3% of total audience/referral traffic - The hosts discuss Forbes’ claim that Perplexity sends only a small portion of its traffic. Apple AI rollout timing: Several features delayed into September and beyond - The hosts note that many AI features were not immediately available at WWDC and will arrive gradually.

Pivotal Quotes: "Apple is not paying OpenAI." — Mark Gurman: Gurman explains the no-cash structure of the Apple-OpenAI partnership. "Apple Intelligence probably won't be bringing a super cycle to the iPhone 16 this fall." — Mark Gurman (via his Twitter/referenced report): Used to frame the idea that Apple’s AI announcements may not immediately drive a major upgrade cycle. "This was the most Apple-y vision of generative AI imaginable." — Ranjan: His positive reaction to Apple’s framing of AI as practical, integrated, and consumer-friendly.

Implications: Apple is repositioning AI around usability and distribution rather than raw model leadership. The episode suggests AI’s next battles will center on defaults, revenue share, content rights, and device integration—not just model quality.