Having led a Bitcoin mining firm for over two years, I’ve come to realize the importance of computing power. Computing power connects the real (chip energy) and virtual (algorithm) dimensions of our world. Under the condition that the ownership of the assets remains unchanged, computing power is an intangible asset that can be used and circulated. It is a commercialized technical service and a consumption investment. This is a remarkable innovation for mankind, and it is an upgrade for the digital economy.
2020 marks the birth year of the computing power infrastructure. Our world is at the beginning of a new economic and technological cycle. We have entered the digital economic civilization. This wave of technology is driven by the combination of AI, 5G, quantum computing, big data and blockchain. People have started realizing that in the age of the digital economy, computing power is the most important and innovative form of productivity.
Computing power is not just technical but also economic innovation. It’s a small breakthrough at the fundamental level with impact that will be immeasurable. And people have finally seen the value of the bottom layer through the 10 years of crypto mining evolution.
However, there are two major problems faced by the entire technological landscape: First is insufficient computing power. Second is the dominance of centralized computing power, which creates a monopoly and gives rise to manipulation problems and poor data security.
How does more computing power help?
Mining Bitcoin has allowed my company to build the foundation of computing infrastructure, so we are planning to eventually expand into AI computing. This experience has further shown me the importance of working toward developing more computing power if tech leaders want to continue creating innovative technologies.
Consider this: For an AI system to recognize someone’s voice or identify an animal or a human being, it first needs to process millions of audio, video or image samples. It then learns to differentiate between two different pitches of voices or to differentiate faces based on various facial features. To reach that level of precision, an AI model needs to be fed a tremendous amount of data.
It is only possible to do that if we have powerful computers that can process millions of data points every single second. The more the computing power, the faster we can feed the data to train the AI system, resulting in a shorter span for the AI to reach near-perfection, i.e., human-level intelligence.
The computing power required by AI has been doubling roughly every three and a half months since 2012. The need to build better AI has made it mandatory to keep up with this requirement for more computing power. Tech companies are leaving no stone unturned to rise to this demand.
It is almost as if computing power is now an asset into which investors and organizations are pouring millions of dollars. They are constantly testing and modifying their best chips to produce more productive versions of them. The results of this investment are regularly seen in the form of advanced, more compact chips capable of producing higher computing power while consuming lesser energy.
For new technological breakthroughs, computing power itself has become the new “production material” and “energy.” Computing power is the fuel of our technologically advanced society. I’ve observed it is driving the development in various technological landscapes, such as AI, graphics computing, 5G and cryptocurrency.
Similar to AI, the decentralized digital economy sector also relies on high computing power. Transactions of cryptocurrencies, such as Bitcoin, are validated through a decentralized process called “mining.” Mining requires miners across the world to deploy powerful computers to find the solution — or the hash — to a cryptographic puzzle that proves the legitimacy of each transaction requested on the blockchain.
The bad news, however, is that the reward to mine Bitcoin is halved almost every four years. This means that following May 20, 2020 — the next halving date — miners who mine Bitcoin would receive half the reward per block compared to what they do now. Two primary factors that compensate for the halving of rewards are an increase in the price of Bitcoin and advanced chips with high computing power.
Miners run not one but multiple high-end graphics processing units to mine Bitcoin, which is an electricity-intensive process. The only way to keep mining profitably is to invest in better chips that produce more computing power with lower electricity consumption. This helps miners process more hashes per second (i.e., the hashrate) to get to the right hash and attain the mining reward.
So far, mining chip producers have delivered the promise of more efficient chips leading to an increase in the mining hashrate from 50 exahashes per second to 90 exahashes per second in the past six months. Per the reports, the efficiency of the latest chips combined with increased Bitcoin prices has helped keep the mining business highly profitable since the previous halving.
High computing power has become an addiction we humans are not getting rid of in the foreseeable future. With our growing fondness for faster computer applications and more humanlike AI, it’s likely that we demand faster and more perfect versions of the systems we use today. A viable way to fulfill this would be to produce more computing power.
The two biggest challenges that lie in our way are producing clean electricity at lower costs and developing chips that have a lower electricity-consumption-to-computing-power-production ratio. The core of industrial production competition today lies in the cost of producing electricity. Low energy prices enable us to provide stable services. For example, there is an abundance of hydro-electric power in southwest China, and cooperative data centers are located there so they can harness the hydropower.
If we could make low-cost, clean energy available everywhere, we’d cut the cost of producing computing power. When this energy is used by power-efficient computing chips, the total cost drops even more and high computing power becomes highly affordable.