Pushing the limits: The nanometer plateau and the future of phone chipsets،
The silicon chips that power our phones and many other devices are, for the most part, the primary driving force behind their evolution. Of course, there are other innovations that make up the typical user experience you get with a modern phone today, like its cameras or its screen, but at its core is a little powerhouse that makes everything possible. you do with your phone. , the best method for increasing the processing power of a chip has been to increase the number of transistors contained inside it, which inherently means making said transistors smaller. The first transistors were invented in the 1950s and measured in millimeters. But today we have improved this process to the point where transistors are measured in nanometers. This is why we sometimes hear phone manufacturers bragging about their latest model having, for example, a chip made using the 4nm process (like the Galaxy S24 series) or even using the 3nm process (like with the iPhone 15 Pro and iPhone 15 Pro Max). It is important to remember that it is not the chip itself that is described when we talk about nanometric measurement, but the width of the gates of the transistors that make up said chip. The smaller the gates of the transistor, the more advanced, powerful and energy efficient the chip (check the graph for a visual representation).
Source – Nikkei Asia.
Major players in the chip industry and the journey so far
There are currently three major manufacturers in the chip industry: Intel (United States), TSMC (Taiwan), and Samsung (South Korea). There is also the Chinese company SMIC, but it is behind the rest of the pack and is currently about two generations behind.
The three major manufacturers mentioned above have all made strong efforts to advance chips, with the most significant advancements beginning in the 2000s. Here's a quick look back from 2013 to today, the year we're in we currently find, and what we can hope for in the near future.
The top row shows the year, with the corresponding transistor gate width measured in nanometers below. The blank spaces do not represent any newly released generations for that year.
“*” means “expected” because the information has not yet become official.
“*” means “expected” because the information has not yet become official.
You can probably see for yourself that over the years, the size of nanometers has steadily decreased to about half the size of the previous generation. In other words, the number of transistors included in a chip has doubled, and closer inspection reveals that it has always taken the industry about two years to take the next step.
Moore's Law: a misleading term
The consistently similar amount of time it takes for manufacturers to move from one generation to the next in a chip's evolution is no coincidence. In fact, Intel co-founder Gordon Moore himself stated as early as 1965 that every eighteen months the number of transistors inside chips would double, which today is known as name Moore's law.
This trend in flea evolution has held true for so long that it almost seems like we are talking about a natural law, one that does not change regardless of circumstances. But the truth is different. Moore presented his thoughts on the subject as mere observation and not as “law.” In fact, he did not anticipate that his prediction would continue beyond a single decade.
It's simply impressive that Moore's observation has held true throughout all these years, but there are physical limits to how long this can last.
Current transistors are made from silicon and a silicon atom measures 0.2 nm. In other words, even if we manage to keep reducing the size of transistors, we can't make them smaller than that.
Visual aid for nanometric measurement
Before we continue talking about nanometer this and nanometer that, it's probably best that we first understand how big a nanometer actually is. Just as it's difficult for a human brain to grasp the size of the universe, it's also difficult to understand the size of a nanometer, but the following examples should put things into perspective (and will probably blow your mind).
Are we near the end of chipset evolution?
So the natural question to ask is whether chip innovation will stop or whether there are other avenues the chip industry can turn to. Well, luckily there is always a way to develop and improve a product. Here are some glimpses into the future of computing.
Advanced chip packaging
Advanced chip packaging refers to the stacking of multiple chips onto wafers, which can help achieve performance levels considered comparable to chips made using the 5 or even 3nm process. This method is also less expensive, as it does not require the same amount of chips. precision and delicate equipment needed for shrinking transistors.
Popular companies such as Intel and Nvidia have already implemented this method in their latest chipsets.
Make transistors from a different element
So, as we have already established, transistors are made of silicon. There's a reason we have what's called Silicon Valley in California. But as we said earlier, silicon can only take us so far as we reach its limits.
An example is gallium. However, the atomic radius of gallium is not smaller than that of silicon, so you cannot install more transistors there. Gallium nitride, however, produces less heat, meaning less cooling is required to keep them at peak performance.
However, this is just an example. There are other materials that manufacturers can experiment with to get better results from a chip.
Quantum computing
Quantum computing is another path we can take to continue to increase our technological power. Quantum computers have the potential to help us explore completely new territories, but the chances of them powering our phones anytime in the future are extremely slim because they require extreme cooling (15 millikelvin to be exact, which is colder than space). which in itself requires very complex and bulky components. In other words, quantum computing is an option, but not one that would significantly affect mobile technology. At least not in the foreseeable future.