World at the End of Moore's Law

By GrabBag | CrackpotPhysicist | 8 Oct 2020


I was born in 1991, so continuous technology improvement has always been part of my life. Everything gets better over time: TVs, computers, smartphones, cars, airplanes... at this rate, it seems like there's no problem in the world that won't be solved by technology in the near future. But something interesting happened recently: the rate of transistor density improvements in electronics has hit a wall.

Transistor density is the biggest factor in determining the improvement of electronics. Take TVs for example: we started with black and white, then had color, standard definition, then HD, and 4k UHD. Pixel density in TVs has been an indicator of technological progress. But this progress wouldn't be possible without all the transistors pushing pixels to the screen, or all the transistors in the routers that deliver those pixels over a network, or all the transistors in the render farms that generate the special effects in your favorite movies. So while we notice the side-effects of improving transistor density, it's easy to forget the underlying cause.

So if transistor density ceases to improve, this would mean the end of an age, and the beginning of a new one. Well, it appears as though we are in the midst of such a paradigm shift.

Microchips are getting small. Like, super small. The transistors are so tiny that they are approaching the limits of physics. Experts are unsure if they will get any smaller than 3nm, which we should see hit the market as soon as next year. So, it's not over yet. But it's definitely slowing down.

All this makes me wonder: what will technology be like when processors can't get any faster? Let's speculate.

First, processors are going to get cheaper. With technology no longer improving, manufacturing will become refined and simplified. The return-on-investment for a chip fabrication facility would be higher because it won't become obsolete in 5 years. Transistors-per-inch would be a commodity, in the same way steel became a commodity during the industrial revolution.

Processor "flavors" will get more diverse. Similar to how ASICs were created to make Bitcoin mining more profitable, we're going to see processors that are domain-specific for things like AI and simulations. There will be less risk for manufacturers in making these kinds of products because they will never be surpassed by consumer-grade hardware. And the general purpose chips would still exist in the future - in either a high speed (CPU) or high parallel (GPU) form factor, and even hybrids of both, depending on the application.

And what if you could operate a processor at a higher frequency? High-end Intel i9 processors currently operate at around 5 GHz. It's difficult to get a CPU any faster than this due to limitations of the underlying material: silicon. But it's totally possible to make chips out of different materials such as gallium arsenide. Transistors made of this material can operate at frequencies in excess of 250 GHz. It has disadvantages over silicon though, which is currently preventing it from being used for general purpose. But this could change in the future, possibly with the addition of other exotic materials, when companies race to develop the next step of progress.

Then there's quantum computers. The first quantum computer prototype was manufactured in 1998. Scientists have made quantum computers bigger, but haven't done much to solve the underlying problems with the technology, like how quantum computers just give wrong answers most of the time. But if scientists could manufacture a perfect quantum computer, it would have severe impacts on the world.

Many forms of digital security in use today rely on certain math problems being impractical to solve. Take TLS for example, the protocol that gives you a green lock icon in your address bar when you are browsing a secure website. This would no longer be secure in a world with powerful quantum computers, so new protocols would need to be adopted. Bitcoin and other cryptocurrencies would become insecure, as quantum computers will be able to crack the security of wallet keys. Advancements to quantum computer implementation (alternatives to Shor's algorithm) could further extend its crypto-cracking abilities. We rely on some form of cryptography at nearly every level of modern electronics from hashing functions to electronic signatures; it's hard to imagine a future where this technology becomes obsolete.

Then there's the "wetware computer," or a computer made from organic matter such as harvested or grown neurons. This kind of technology could lead to organic robots that can think for themselves. Artificial neurons could be injected into a person to make them smarter, especially in circumstances where an injury, birth defect, or genetic problem has caused mental impairment.

There will also be advancements to technology that we can't comprehend right now, either because it doesn't fit our current understanding of physics or it sounds ridiculous. The ancient Egyptians built pyramids and other structures using techniques that modern science can't explain: what was their secret? We still don't know why ceramics are high-temperature superconductors. If scientists discovered superconductors at higher temperatures, it would be a complete game-changer.

So what's the world like at the end of Moore's Law? Well, I'd like to think it's a bright future full of even more technology advancement. Don't you?

How do you rate this article?

7



CrackpotPhysicist
CrackpotPhysicist

Interesting stuff, from anywhere in the universe.

Send a $0.01 microtip in crypto to the author, and earn yourself as you read!

20% to author / 80% to me.
We pay the tips from our rewards pool.