It’s not like people run intense ML / data workloads on their Apple laptops. These are all run in the cloud on commodity hardware, not M1 / M2 chips.
AI regulations. Insecured politicians who are anti-tech can put pressure to regulations which can decide what you can and can't train on cloud. With your laptop it's hard to regulate.
Idea is eventually with the data restrictions you need to have solid isolation, hence the direction.
The compatibility of machine learning frameworks with Apple silicon is significant for performance optimization and efficiency on Apple devices. As Apple transitions to its custom silicon, having machine learning frameworks that are optimized for this architecture ensures better utilization of hardware capabilities, leading to faster and more energy-efficient operations. This is particularly crucial for tasks like AI processing on devices such as iPhones and Macs, enhancing user experience and overall system performance.
This sounds like something out of chatgpt.
💯 this is gpt
I wouldn't be surprised if few years later Apple Silicon finds its way into datacenters. Power efficiency is most important thing for DC, and if Apple Silicon does it 5x better than Intel/AMD, it's inevitable it will be in DC
It's the arm architecture that's behind energy efficiency, it means a lot of new companies will take up that space, including Qualcom, Mediatek(I hope)
Never. Gonna. Happen. Aside from power efficiency, you know what else is important to DCs? Flexibility. Upgradability. Compatibility. All of which the M series sucks at. Is Apple going to open up their silicon to run any Linux distribution? Configure with any third party memory? Run with *gasp* Nvidia GPUs? The Mac Pro can't even support discrete GPUs, despite having plenty of PCIe bandwidth. What makes you think this is going to be viable for AI/ML? The M-series is efficient in the low power budgets of a laptop or tablet. It's not designed for server use-cases, where you want hundreds of efficient cores. Plenty of companies are already chasing (and succeeding) there. Apple has zero advantage there. And that's before we get to the non-technical issues. Apple doesn't have a sales network or support network that enterprises need. Microsoft and Amazon - the two biggest players - are already working on their own custom chips. It would be an ironic twist of fate if Apple starts supplying chips to IBM, whom they once openly mocked. The M series is going to remain in laptops, where it belongs.
Some apple direction gets a chance to do empire building by proposing a new project. This is the easier project to do then improving Siri quality using LLM
ICloud of commodity M1/M2 chips? The point is to have your customers subsidize the investment.