Apple’s A17 processor was the first chip to use TSMC’s N3 (3nm class) technology, and this week the company expanded its N3 series with its PC-focused M3 family of chips for desktops and laptops. Analyst Jay Goldberg of Digits to Dollars estimates that the company spent about $1 billion on the development and production of the M3 alone.

“We must assume that only abatement costs [серії M3] for three [SoC] It should be around $1 billion, Goldberg wrote. “Few companies can afford such a large undertaking.”

Currently, Apple’s M3 family consists of three rather complex processors: the 25-billion-transistor M3, designed for entry-level desktops and high-end mass computers, laptops and tablets; M3 Pro with 37 billion transistors for productive batch machines; and the M3 Max with 92 billion transistors for high-end laptops and entry-level workstations. Each chip is designed to meet different computing needs, from everyday tasks to professional coding, heavy engineering modeling and video production.

Apple’s flashy M3, which comes with eight general-purpose cores and a new integrated GPU, is just as complex as AMD’s top-rated Phoenix processor (25 billion vs. 25.4 billion), while the M3 Pro and M3 Max are much more complex.
In fact, with 92 billion MOSFETs inside, the M3 Max is the most complex single-chip processor ever released (although based on what we know about some future AI processors, that won’t be around for long).

Apple used TSMC’s N3 manufacturing process to make the M3 family more affordable; This is a risky move since the technology is relatively new, but it seems to have worked. As chip detective @Frederic_Orange points out, Apple can probably fit up to 415 M3 dies on a single 300mm wafer; This implies a die size of approximately 146 mm^2.

In comparison, AMD’s Phoenix (of similar complexity) has a die size of 178 mm^2. We can only guess if Apple’s M3 is cheaper to produce than AMD’s Phoenix based on TSMC’s rumor, but smaller chips are generally easier to produce and produce.

Apple spent $26.251 billion on research and development in 2022, with a significant portion of this spending going towards chip development. The scale of investment in silicon in general and M3 series SoCs in particular suggests that Apple is one of the few companies with the economic capacity to embark on such developments.

Development of complex PC-oriented processors requires long development cycles and intensive capital investment, often lasting years. When it comes to a completely new platform – like Apple’s M3 family – the cost of development is staggering, especially in Apple’s case as the company tries to develop as much intellectual property as possible in-house. With the M3, Apple is not only using its own general-purpose cores based on the Arm instruction set architecture, but also an entirely new GPU architecture that supports hardware acceleration of ray tracing and mesh shaders, a new AI NPU, and a new multimedia interface. engine. Source