Apple is in early talks with Google to use Google’s Gemini AI model for a new version of Siri.

The discussions are part of Apple’s effort to advance its generative AI capabilities after delays in upgrading Siri.

Apple has also considered working with Anthropic and OpenAI but has not made a final decision.

Executives are assessing whether to use Apple’s own AI models or rely on external technology.

Two Siri versions are reportedly in development: one using internal models and another powered by outside partners.

No formal deal has been made, and the talks with Google remain exploratory.

If Apple chooses an external model, it would run on its Private Cloud Compute servers, not directly on user devices.

Apple recently postponed a planned Siri upgrade by a year due to engineering setbacks, and its AI team has seen key departures.

🔗 Source: Bloomberg

🧠 Food for thought

1️⃣ AI talent wars are forcing even tech giants to reconsider their development strategies

Apple’s internal AI struggles reveal how difficult it has become for companies to build competitive AI systems in-house, even with massive resources.

The company’s AI models chief architect Ruoming Pang left for Meta in July with a $200 million compensation package, and several colleagues soon followed him1.

Many remaining Apple AI employees are now interviewing elsewhere, either due to the potential shift to third-party technology or to capitalize on multimillion-dollar job offers from competitors1.

This talent exodus comes as Apple’s Siri overhaul faced engineering setbacks that delayed the project by a full year, forcing the company to sideline AI chief John Giannandrea from Siri development entirely1.

The technical gap is also evident in Apple’s AI capabilities. While the company recently began testing its first trillion-parameter model, it remains well behind OpenAI’s several-trillion-parameter systems1.

Recent Apple developments