apple unveiled its work in the field of artificial intelligence on Monday at the annual Worldwide Developers Conference in Cupertino, California.
WWDC showed off a demo of the redesigned Siri. Siri can talk to you and is a huge improvement over previous versions of the assistant. In the demo, Siri was also able to check concert dates, set reminders to buy tickets, and even get directions to pick up a friend on the way to the concert venue.
But the announcement also highlighted that Apple is adopting a different strategy than many of its Silicon Valley rivals, choosing not to spend billions on infrastructure or its biggest, most cutting-edge models, instead focusing its message to potential customers on privacy benefits and convenience.
Apple executives on Monday emphasized the difference in their statements.
“Some companies seem to be pursuing AI for AI’s sake without explicitly considering the people AI should ultimately serve: all of us,” Craig Federighi, Apple’s senior vice president of software, said in the announcement.
But two of the traditional AI leaders: google and Nvidiais supporting Apple with its most advanced model, called the Apple Foundation Model Cloud Pro, Apple executives told the media during a speech at its headquarters on Monday.
Apple and Google announced their partnership on Apple Intelligence in January, but this is the first time the companies have officially acknowledged that some Apple Intelligence features will run on Nvidia chips.
Apple AI executive Amar Subramanya said AFM Cloud Pro is comparable to Google’s Gemini frontier model. Apple officials say it will run in the cloud on Nvidia GPUs, which are part of Apple’s private cloud computing infrastructure.
“We are working with both Google and Nvidia to extend our private cloud computing infrastructure to Nvidia GPUs in the Google cloud, while maintaining Apple’s unparalleled privacy guarantees,” said Subramanya.
Sebastian Marineau-Mes, vice president of software, said Apple wanted to use Nvidia’s latest chip, but Apple wanted to configure the chip in a more private way that wouldn’t allow it to read what was on the server.
Marineau-Mes said Nvidia’s recent improvements, including a technology called “fuzzy sensitive computing,” have enabled Apple and Google to build systems that meet the standards.
“We wanted to take advantage of Nvidia’s latest technology, so we set out to extend our private cloud computing to third-party clouds,” Marineau-Mes said.
Apple is trying to differentiate itself from companies investing heavily in AI by emphasizing that its software is more private, meaning it doesn’t collect as much data as web-based AIs like OpenAI’s ChatGPT or Anthropic’s Claude, and that it uses access to locally stored user information like calendars and text messages to personalize AI features.
Craig Federighi, senior vice president of software engineering, at Apple’s WWDC at Apple Park in Cupertino, California, on June 8, 2026.
Kif Leswing | CNBC
Monday’s tech talk was held for Federighi and his lieutenants to discuss how Apple built Siri AI and its Apple Intelligence layer, and how it differs from the AI that consumers may already be familiar with.
Apple executives outlined the software’s architecture, which powers Apple’s operating system and software with software called a system orchestrator that routes AI queries to the appropriate model on the device or in the cloud, depending on the amount of computing power and personal data needed.
The system orchestrator is “key to the privacy architecture of the entire system,” Federighi said.
The talk also provided more details about Apple and Google’s partnership.
Federighi said the AI software built into the company’s devices, called Apple Intelligence, uses Apple’s own model, rather than the same model as the publicly available Google Gemini, as many in the tech industry expected when the two companies announced their partnership in January. He also said that he is not using Google’s off-the-shelf cloud infrastructure.
Federighi said Google’s technology is being used to build Apple’s own models, specifically the “3rd generation” AFM model for the cloud announced Monday, which is designed to run on Apple’s chips.
“These four models I talked about earlier, AFM Core, Core Advanced Cloud, and Cloud Image, are all custom built for Apple Silicon, trained using our own data with reinforcement learning, and refined using output from the Gemini frontier model,” Subramanya said.
