Apple's silicon revolution: From mobile to AI data centres

Apple's journey into custom silicon has been driven by a singular pursuit—efficiency, power optimisation, and seamless integration across its devices. Starting with the A-series chips in the iPhone and iPad, Apple showcased how control over its silicon could deliver unparalleled performance. Now, with the M-series powering its Macs, Apple's move into high-performance hardware has solidified its technological dominance. 

But this story extends beyond mobile and desktop devices—Apple is now positioning itself as a key player in the future of Artificial Intelligence (AI) and edge computing infrastructure, leveraging its custom silicon to drive the next generation of data centres.

A bold bet on silicon

Each year, Apple's chip innovation redefines not just its own devices but the entire tech industry. Their chips are found everywhere, from Ultra-Wideband technology to the complex silicon architecture in the iPhone. If you open up your iPhone, you will see several PCB boards filled with silicon chips, many of which are already Apple Silicon. However, the journey toward silicon dominance wasn't instantaneous. To understand Apple's journey from designing mobile chips to data centers we have to go over some crucial moments in the history of Apple Silicon. 

The first iPhones Apple featured Samsung chips, and interestingly, some of them even had an Apple logo. However, this dependency soon gave way to a strategic vision of full control. A pivotal moment came in 2008 with the acquisition of P.A. Semi (Palo Alto Semiconductor), a chip design company, marking Apple's first steps toward designing its own chips. 

The first Apple Silicon was the A4 application processor in 2010, which went into the iPhone 4 and the first iPad. Built on ARM architecture, this 45nm chip signified Apple's entry into the world of custom silicon, manufactured at Samsung's fabs. It was followed by the A5, A6, and finally, the A11 Bionic chip, which introduced the neural engine which was powering some initial AI features like the first FaceID, signalling Apple's long-term strategy of AI acceleration through silicon.

Apple's latest unveiling of the A18 and A18 Pro at its latest 'It's Glowtime' event showcases this silicon evolution. Manufactured using TSMC's advanced 3 nm process, these chips, with their advanced ARM extensions, offer a 30% performance boost while integrating a 16-core neural engine for faster AI processing. AI is no longer a feature—it's a core element of Apple's hardware strategy.

Strategic acquisitions: shaping the future

Apple silicon's story is much deeper and more interesting than just their A-series chips. Their next strategic move was to build an in-house power management system. In 2018, Apple acquired Dialog Semiconductor, a brilliant move to secure control over power management technology. By owning this key aspect of the hardware stack, Apple optimised energy efficiency across its devices, ensuring longer battery life without sacrificing performance.

Apple's relentless pursuit of vertical integration extends beyond mobile devices. In 2020, Apple shocked the industry by announcing the shift to M1 chips in its Mac lineup, abandoning Intel's legendary x86 architecture. The switch to ARM-based chips allowed Apple to unify its ecosystem under a single architecture. This was a win for Apple but a significant loss for Intel, which is still struggling to recover from Apple's departure.

For nearly every product they create, Apple now builds its own chip. Take AirPods for example, which feature W-chips that evolved into the H-chip series. Over the past few years, Apple has acquired more than 100 companies in areas such as memory controllers, AI, AR, and AI music. At this point, Apple acquires a new company roughly once a month on average.

The acquisition of Intel's modem business in 2019 is especially worth mentioning. This strategic move was aimed at replacing Qualcomm's expensive 5G chips. Although technical hurdles have delayed the rollout of Apple's own 5G modems, the goal is clear: to design a fully integrated system-on-chip (SoC), consolidating more components onto the PCB, ultimately reducing space and power consumption. As we know, the form factor plays a huge role in the mobile market. Apple has already managed to do Wi-Fi and Bluetooth chips, it's only a matter of time before we see an Apple 5G chip.

These incremental innovations, driven by strategic acquisitions and in-house development, reveal Apple's broader game plan: total hardware and software integration. 

The edge and beyond

While Apple is integrating AI at the edge, not all processing will happen at the edge, as there are significant cloud-based communication involved in the process. Using Apple Intelligence, when you ask something to Siri, it will send your prompt to the cloud, which is running on Apple Silicon. This hybrid edge-cloud model allows Apple to scale AI features and deliver them seamlessly across its ecosystem. This is why Apple is buying these very fast and very expensive 5G chips from Qualcomm.

In hindsight, Apple's strategy is simple: analyse user behaviour, identify the most-used AI features, and build hardware acceleration for these features in the next generation. The best way to accelerate an operation and make it run fast on low-power is to build an ASIC (Application-Specific Integrated Circuits). Just as Advanced Media in the A18 Pro accelerates specific media tasks, future Apple devices like AirPods and Apple Watch will follow suit with AI-enhanced features, powered by Apple silicon.

Project ACDC: Apple's next frontier

During its latest keynote, Apple referenced its data centres twice, emphasising its focus on security and privacy. As Apple ventures further into artificial intelligence and cloud infrastructure, it is developing chips specifically designed to handle massive computational demands that exceed the capabilities of edge devices.

The Wall Street Journal previously reported that Apple's custom silicon development for data centres—under the umbrella of Project ACDC (AI Chips for Data Centres)—aims to ensure top-tier security and privacy. Reports suggest that Apple is producing large volumes of M2 Ultra chips and the low order volume of Nvidia GPU further hints at a broader plan: to make its custom silicon the backbone of its AI infrastructure.

The transition from Nvidia GPUs to Apple's M2 and, eventually, more advanced M4 chips in the future for generative AI processing signals a massive shift in how Apple plans to dominate cloud AI. Apple's expertise in AI accelerator packaging, particularly with TSMC's Integrated Fan-Out (InFO) technology, positions the company to lead in the AI and cloud computing space. There's even speculation that Apple could acquire Ampere Computing, which specialises in ARM-based chips for cloud environments, further expanding its cloud silicon capabilities.

Strategic investments and partnerships

As AI becomes a critical battleground, Apple is not just focusing on its own AI models but is also exploring partnerships with leaders like OpenAI, Google, and Claude, potentially to accelerate its AI ambitions. Despite these partnerships, Apple's strategy remains clear: build custom chips to maintain control over both the hardware and software elements of its AI ecosystem—a move that could yield new revenue streams, including advertising and AI-powered services.

While developing AI-specific chips requires a substantial investment, the cost of training AI models is relatively lower. For Apple, this investment is about more than just performance; it's about securing control over its infrastructure and building an ecosystem where AI enhances every device and service it offers.

The beauty of a modern smartphone is that there are many more chips inside than you can even imagine. For example, an iPhone contains a LiDAR chip from Sony, gyroscopes and accelerometers from Bosch, and other chips from Broadcom, Renesas, NXP, and memory from the Korean company SYNNEX. With each new iPhone release, there are countless memes suggesting nothing has changed. However, when you realise all the hardware design updates and the immense effort that goes into creating this piece of hardware, you kind of appreciate your phone a bit more!