Apple Wants AI to Run on Your iPhone, Not Just in the Cloud
In recent years, it has seemed that every conversation about artificial intelligence leads to the same place, vast server farms, thousands of graphics chips and data centers consuming enormous amounts of electricity. Technology giants are investing huge sums in expanding their cloud infrastructure to support larger and more complex AI models. Apple is no exception, and it is investing billions of dollars in building and expanding data centers and operating the cloud infrastructure that underpins some Apple Intelligence services. ● Siri upgrade and cooperation with Google, all the announcements from Apple’s flagship event ● After Anthropic, OpenAI filed a confidential prospectus ahead of an IPO
However, a behind-the-scenes session at the WWDC developers conference painted a somewhat different picture. Alongside recognition of the cloud’s importance, Apple is promoting the idea that a significant share of AI tasks should be carried out directly on users’ own devices. From the company’s perspective, the combination of Apple Silicon chips, dedicated AI engines and unified memory in ever-growing capacities makes it possible to move more and more computing capabilities from where they are performed today, in remote server farms, to the computer, phone or tablet in the user’s hands.
To illustrate what this vision could look like in practice, Apple devoted a substantial part of the tours for WWDC attendees to meetings with software companies that are already building products around the idea of local AI. Draw Things, for example, showed how images and videos can be created using artificial intelligence directly on a MacBook Pro, without sending the information to external servers. A user can start from an idea or a text description, make repeated changes to an image, edit specific details and even generate a short video, all while the entire process is carried out on the computer itself.
Perplexity presented a system called Personal Computer, which runs on a Mac mini and serves as an AI agent capable of performing complex tasks with a single request. During the demonstration, the system built a website, created charts and automatically made changes to a presentation.
At the same time, LM Studio presented a computing cluster made up of four Mac Studio computers connected together, with 2 terabytes of unified memory and the ability to run especially large language models. The company also demonstrated how these models could be accessed from a laptop or from an iPhone.
Although the examples covered different areas, from content creation to AI agents and language models, they all reflected the same strategic direction, shifting an increasing share of processing power to edge devices.
Will users’ privacy improve? For Apple, this is much more than a specific technological improvement. The company is trying to offer a different operating model from the one that has become dominant in the industry over the past two years. Instead of every action being automatically sent to the cloud and processed in a remote data center, Apple wants to perform as many actions as possible locally, and rely on servers only when especially large computing power is required.
This approach has several clear advantages for the company. It can improve user privacy because more information remains on the device, it can reduce waiting times and ease the burden on expensive cloud infrastructure, and it can also give Apple a competitive advantage thanks to its tight control over both hardware and software. While this is not an attempt to replace data centers or cloud services, but rather to create a new balance between them and edge devices, the demonstrations showed that Apple sees Mac computers as much more than ordinary personal computers. In its view, they may become a major AI platform in their own right in the coming years, one capable of performing tasks that once required access to massive computing infrastructure.
*** Full disclosure: The reporter was a guest of Apple
"Innovation is an inseparable part of the DNA of the Lubinski Group"
Please note: Globes editorial staff strives for diverse, substantive and respectful discourse in accordance with the code of ethics set out in the trust report by which we operate. Expressions of violence, racism, incitement or any other inappropriate discourse are automatically filtered and will not be published on the site.