Why Apple’s Slow AI Pace Looks Like a Strategic Chess Move

Apple’s Capital Spending vs. Industry AI Investment

TechCrunch author Lucas Ropek notes that Apple plans roughly $14 billion in total capital expenditure this year, compared with about $9 trillion in AI‑related commitments from other tech giants. The figure is Apple’s overall capex, not an AI‑only budget, yet the contrast suggests Apple is not racing to burn cash on AI.

Ropek also points out that Apple’s record iPhone sales this quarter generate cash while the company continues to collect a 30%/15% App Store cut from AI‑focused apps such as those from OpenAI and Anthropic, effectively “taxing” the AI ecosystem.

Renting Gemini for Siri’s Knowledge Layer

Apple has disclosed a collaboration with Google and the Gemini model series to power the next‑generation Apple Foundation Models. In practice, Gemini runs the web‑knowledge and near‑real‑time information component of Siri, marking a shift from Apple’s historic rivalry with Google Maps.

The author speculates that as large models become commoditized, hardware companies prefer renting models rather than spending billions to train their own. Renting allows Apple to swap providers—e.g., from Gemini to Claude—while retaining control over the data pipeline and privacy guarantees.

Federighi emphasized that “privacy in AI is non‑negotiable,” meaning Apple wraps rented models in its own data‑handling policies, keeping the “moat” in the privacy layer rather than the model itself.

OS‑Level AI Distribution as a Competitive Advantage

Siri AI is embedded directly into iOS, macOS, and iPadOS, offering contextual suggestions across Mail, Messages, phone calls, and on‑screen awareness. Ropek argues this OS‑level integration threatens third‑party AI apps like ChatGPT, Perplexity, and Claude because users get AI functionality without downloading separate apps.

The article notes that Siri AI is still in beta and real‑world performance remains untested; Ropek cautions that a final verdict must wait for broader consumer access.

Sarah Perez’s Counterpoint

TechCrunch’s Sarah Perez offers a contrasting view, asserting that Apple has been quietly fixing core software issues—search, file sharing, Health app—over the past two years. She interprets the WWDC keynote as “leading with fixes before features,” suggesting that Apple’s narrative of steady progress may be overstated.

Implications for Developers and Hardware Companies

For AI product builders, Apple’s OS‑level AI could be a double‑edged sword: it may limit distribution channels for general‑purpose AI tools while opening opportunities for vertical, domain‑specific copilots that can differentiate beyond generic Q&A.

For hardware manufacturers, Apple’s apparent preference for renting models rather than training their own signals a strategic posture: conserve capital for channel stability, buybacks, and privacy‑focused services while leveraging the best available models.

Strategic Questions

If Apple’s “slow” approach proves successful, should other hardware firms continue investing in in‑house model training, or follow Apple’s rent‑instead‑buy model to allocate resources elsewhere?

The author remains skeptical until Siri AI is fully released and Apple’s promised software fixes are validated.