For obvious reasons, a high-tech facility is not the place a journalist is often invited to visit, although there are occasionally opportunities to catch a glimpse of what lies beyond the threshold. In the heart of Europe, four office buildings, without external indicators that allow us to guess what is inside, make up the Apple Silicon chip design center in Munich. This research complex will involve a total investment of 2,000 million euros from 2021 to 2029. The Californian company uses three points in the world to design its chips, which today are the most advanced in the world in performance per watt of power: Cupertino (California), Germany and Israel.
In 2020, Apple announced the transition of its computers towards a more efficient processor architecture, the same ones that had equipped the iPhone and iPad for years. It was a new concept that nobody had approached until then from the ARM architecture, the same one used by mobile phones. The strategic decision meant goodbye to Intel, which had built Macs since 2006. Almost the entire range of computers now works with its own chips. The history of this technological leap, with very powerful chips but with very little energy consumption, began in 2010, with the first iPad.
Apple’s first tablet had a chip called the A4, which in the fall of 2010 was also used for the iPhone 4. The A4 was called a SoC (system on a chip). It doesn’t have just a central processor (CPU), but it brings together more elements that the device needs to function. In this case, it also had an integrated graphics processor (GPU).
Traditional PCs, including Macs that have been powered by Intel chips, include a CPU and GPU as separate chips on the motherboard. Each one has its own memory in addition to going to the system memory. In order for the two chips to communicate, they have to move data back and forth in a process that consumes more power and time. With Apple’s scheme, the memory is included on the same chip, so data input and output to the processors is more efficient. Each chip also carries a powerful neural motor for artificial intelligence tasks.
The Apple Silicon for Mac are baptized with the letter M. There is already a generation M1 and an M2. When Apple began developing the A4, it was made with 65-nanometer transistors – a nanometer (nm) is one billionth of a meter. The A16 Bionic launched last year uses a 4 nanometer technology. The smaller the transistor, the less resistance it offers to the flow of electric current, so it heats up less and more transistors fit in the same space. In short: more speed and less power consumption, which means more battery. Today, a Mac mini M2 consumes 60% less than its predecessors with Intel.
Few companies in the world could make such a change, because you have to have your own operating system so that hardware and software fit together perfectly. Right now, company sources explain, each chip is designed for each product. One of the revolutions in the design of Apple Silicon is that they are scalable. From each basic development of the current M1 and M2, three additional versions have come out so far (Pro, Max and Ultra –only in the M1)–) that each greatly multiply the capabilities of its predecessor.
In the lab you can see engineers measuring the efficiency and power scaling of different cores on a chip based on workloads. Those are the key factors that can get you a full day of laptop battery life. In Munich they designed two small black chips located on each side of an M2 chip in a MacBook Air that are the integrated circuits that manage power.
In an adjoining secluded room, the simpler motherboard architecture of a newer MacBook Pro is shown compared to the last model carried by an Intel. The result is that the leftover space from several separate chips has been used to design a new computer that brings back the magnetic MagSafe charging connector, an HDMI port, larger speakers and a memory card reader, something that many users requested.
One of the most striking installations in the center of Munich is an anechoic chamber. They have several. It is a room that can be closed tightly in which a series of cones made of a material that absorbs sound waves is distributed everywhere on the walls and ceiling. There it is verified that, once the new chips are installed in a device, they do not cause interference with other devices. This is a type of test that is very difficult to simulate theoretically and much easier to check physically before the product reaches the market.
Having a chip design center of this size in Europe is also a strategic move that allows Apple to hoard talent just as the United States and China find themselves locked in a geostrategic competition for chip hegemony. It may seem strange that the design depends on three centers at the same time, but from a certain point of view it can be said that Apple is designing its chips 24 hours a day.