It's not just about security, it's also about scale and efficiency. Credit: Apple The rush to build global cloud infrastructure to support artificial intelligence (AI) has turned Nvidia into a top three tech stock and will likely nurture innovation in processor design and energy creation — but Apple may need to serve itself. Apple needs to bring more for less The IEA says global electricity demand to drive data centers will more than double by 2026. That’s even as consumption of AI services such as OpenAI spikes when devices (including new iPhones) gain easy access to them. To satisfy demand, tech firms must tweak more computational performance and additional energy efficiency from the chips they put inside both servers and edge devices. But this push for efficiency and low power reflects what Apple has been trying to do with its own silicon development for years. Apple Silicon chips were way ahead of the wider industry on both metrics (at least for consumer devices) long before generative AI (genAI) climbed through the Overton Window. While attention on server processors rests with Nvidia, the M4 processor inside the iPad Pro could be seen as harbinger of what is to come. At 38 trillion operations per second, we know it has the fastest Neural Engine you can get. Ever modest, Apple has described the M4 iPad as the best AI PC money can buy, and those same processors are also coming to servers. Apple is now in the server market The company’s Private Cloud Compute (PCC) system consists of proprietary Apple servers running Apple Silicon chips to provide responses to Apple Intelligence queries. While these are only being situated in US data centers at this time (as that’s the only place Apple Intelligence will be made available on launch), it’s easy to expect the company will deploy these highly secure systems globally *except in the EU in the coming months. It must. As Apple Intelligence launches globally, Apple will find itself needing to ramp up its international server infrastructure to meet the demands for AI its billion or so customers might make. But one data center at a time, server by server, Apple is already in the server market. Motivated by privacy, these Apple servers also meet wider industry needs around energy consumption and performance requirements. It makes sense for Apple to expand this provision, perhaps even to offer highly secure, low energy server services to enterprise users, but it’s more likely to drive its streaming services while lowering energy consumption. There’s another chip coming It is worth noting that Apple hasn’t yet hit a performance ceiling. Very likely to be tumbling off production lines right now, the next iPhone chip is expected to be a 3nm processor. This might deliver even faster Neural Engine performance than you get inside the M4, which means the next iPhone will be capable of handling even bigger calculations at higher speeds for less energy than Apple’s best available current chip delivers now. Apple also has a road map toward 2nm chips, which will maintain that pattern of performance and energy efficiency. All of this means the company already has a road map toward processor performance that it can now apply to the server market it abandoned in 2011, when it discontinued the Xserve. This is true corporate social responsibility Apple already knows it isn’t good enough to just put high-performance chips inside servers and edge devices if they consume vast quantities of energy. To mitigate this, the company has already invested hundreds of millions of dollars in reducing energy consumption across its entire ecosystem, including major investments in renewable energy supply. It will not turn back time. With that in mind, it will want to widen its ecosystem of low power, high performance iCloud servers, and when doing so it makes sense for it adopt those servers across its other online services over time. After all, if it can build servers and services that can be delivered at lower energy requirements without compromise on performance, why wouldn’t it do so? For Apple, adoption of these may be an easy win in terms of its environmental performance data. But at what point will these systems become a service offering in their own right? What value could they unlock for the company? However the eventual story ends, it’s interesting that by focusing on energy efficiency and computational performance for iPhone chips (and the PA Semi purchase), Apple put itself in a good position to meet the then-unseen challenges of server-based AI — and the space between the lines suggests we’re not near the end of that particular story just yet…. Please follow me on Mastodon, or join me in the AppleHolic’s bar & grill and Apple Discussions groups on MeWe. SUBSCRIBE TO OUR NEWSLETTER From our editors straight to your inbox Get started by entering your email address below. Please enter a valid email address Subscribe