Back in the 1990s, a group of scientists started to worry about the future of computer circuits. Over the prior thirty years, circuits had been getting smaller and smaller. Costs had been halving on a regular 18-month clip, and computer power was going exponential. But now circuits were getting so small that scientists worried that they were running up against the law of physics. At a certain point, quantum uncertainty would apply. At this scale, engineers would no longer be able to dictate the flow of electrons. Thus, scientists would not be able to make circuits smaller.
The trend of ever-shrinking transistors was originally proposed by IBM engineer Gordon Moore. Moore observed that the number of transistors you could fit on a circuit doubled every 18 months or so. His law has held constant, even until today. But now things are changing. We weren’t close to bumping up against that quantum wall back in the 1990s. But now, circuits are so small that we are.
This presents a fundamental problem for the entire field. The explosion in technology we’ve seen over the last few years has all been made possible by making circuits smaller. If scientists can’t make them smaller, the growth of the industry will grind to a halt.
So what does this mean for investors? Is there still a bright future in technology?.
The Growing Cost Of Traditional Methods
At the moment, chip makers have to go through a complicated process to make components. They have to design them using advanced software. Then they have to develop processes to print circuitry on such as small scale. Then they have to use the right facilities, like clean rooms at http://laserlight.com/capabilities/clean-room-processing-packaging/.
So far all of this has been great value for money. Computer components have gotten cheaper every year. But the cost of the processes to make them is now going exponential. Clearly, the traditional manufacture of chips will come to an end sometime before the end of the next decade.
The Search For New Opportunities
The impending doom of traditional methods doesn’t mean an end to the industry. As the prices of developing smaller processes go up, that will create research pressure for the next chip paradigm. When the costs of retooling get too high, chip makers will simply find new ways to get more done on a single chip.
All of this is rather exciting. It means that we’re likely to see entirely new chip technologies emerge. Some of these technologies might not even use silicon.
We’re also likely to see more specialized chips for different applications. For instance, http://www.dailymail.co.uk/ reported this year about an IBM chip that mimics the brain. Other companies, like Nvidia, are developing chips that are optimized for autonomous cars.
Going forwards, there are still opportunities for investors. But these opportunities won’t be in general computing. Instead, they’ll be in the development of more specific applications. Expect investors to be investing big in chips that are designed for AI systems. These will be in demand in the future.