Computing & Energy – The Fourth Technology Revolution Driven by Computing

I. God’s Perspective on Bitcoin’s Revelation and Driving Role

Bitcoin has been in existence for more than ten years. In a completely free market competition, it is undoubtedly the most successful case so far. Instead of discussing the price of Bitcoin, we’ll look at what Bitcoin brings to the table and what it drives from a more fundamental perspective.

1.Evaluating BTC by monetary standards – the globalization case for technology cost reduction

Bitcoin’s rapid growth is tied to the fact that it solves crucial core problems with technical means.

Issuance rules, BTC uses code to lock in the fact that the total number of 21 million cannot be increased. In contrast to the hyper-inflationary situation of fiat currencies in many weak sovereign countries, BTC uses technology to support the consensus that no additional issues can be issued.

Interchangeability, compared to traditional currencies, BTC’s interchangeability is undoubtedly superior. No third-party permission is required, and mastering the private key holds true ownership, independent of any third-party influence.

Trust costs, Bitcoin uses technology to ensure the security and verifiability of transactions, greatly reducing the cost of trust for parties who do not know each other.

Circulation costs, as a digital asset, Bitcoin has a higher flow rate and lower circulation costs than any physical currency, which creates a positive feedback mechanism that makes more transfers willing to choose Bitcoin as the path.

2.Analyze BTC mining with computing industry ideas – leading the chip computing industry

We all know that Bitcoin is mainly obtained and recorded through the act of “mining”, mining is more like a traditional industry, requiring a lot of hardware investment and construction costs and a long period of capital return expectations.

From the perspective of chips, mining is inseparable from mining machines. Let’s look back at the development history of Bitcoin mining machines. From the earliest personal computer CPU era, to the later GPU and FPGA era, to the ASIC era, the power consumption of mining chips With rapid iteration, the overall computing power is also showing an exponential increase. From the perspective of chip development, Bitcoin mining machines have promoted the large-scale application of ASIC chips and the rapid progress of the process. In the future, it is believed that Bitcoin mining machines are likely to adopt more advanced nano-processes on a large scale earlier than mobile phone manufacturers. Following the pace of Bitcoin mining machine chips, we have seen the rapid development of AI, and artificial intelligence chips have also begun to transform to ASIC, beginning to use 28nm or even 16nm for small-scale applications. I believe that with the rapid development of AI and marginal computing needs, we will also see the mass production, application and development of AI chips.

From the dimension of stand-alone power consumption, over the years we have seen the mining machine from a single several hundred watts to about 1.5kw in the 16nm era and then to 2-3kw in the latest generation, the stand-alone power consumption is getting bigger and bigger. To the current 2-3kw stage, we will see many hardware technology bottlenecks: for example, the production of a single 2-3kw large power supply stability and yield rate has been a problem, 2-3kw single power consumption of power distribution / cable have put forward a new level of requirements, cooling problems are more serious, the original air-cooled can no longer meet the current higher heat flow density of cooling needs, some manufacturers even need two sets of violent fans to ensure the heat dissipation effect. Energy costs have become more prominent, both in terms of energy consumed by computing and energy consumed by cooling, reaching a new order of magnitude, with both individual energy consumption and total industry energy consumption entering the next phase of rapid growth. All of these changes are driven by the demand for computing power, all of which has freely and rapidly evolved to where we are today. Regardless of the Bitcoin mining machine, we will find that the upcoming 5G era, 5g base stations are also developing in this direction and facing the same problem. 4g base stations have a server stand-alone power consumption very close to that of 16nm era miners, while 5g base stations now have a standalone power consumption of 3-4kw and need 3-4 times the number of 5g base stations to achieve good signal coverage compared to 4g, which means that telecom operators face huge energy consumption costs. At the same time, 5g base stations with such large power consumption pose a challenge to power distribution, many scenarios do not have the corresponding power distribution standards, changing power distribution facilities lead to higher cost investment, so we see that Huawei launched 5g base stations with a battery, which is to solve the problem of power distribution, the same large power consumption will also face serious heat dissipation problems, especially in outdoor conditions in the four seasons of the year, the ambient temperature difference may reach more than 60 degrees Celsius.

From the perspective of the data center, although the Bitcoin mining farm is currently very extensive compared to the traditional IDC computer room, which is mainly constrained by cost and policy, the total load of the Bitcoin mining farm is close to 10 million kilowatts (according to the full Bitcoin Network 100E computing power computing, considering the average power consumption of the new generation of mining machines 60w/T, the average power consumption of the whole network mining machine can be roughly regarded as 100w/T), the growth rate of this scale exceeds that of traditional IDC data centers and will grow rapidly. Similarly, with the rapid development of AI, 5G, and IoT, there will be exponential growth in computing demand and corresponding data center power load. At the same time, Bitcoin mining as a highly sensitive industry to electricity costs, the free hand of the market allows miners and mine owners to take the initiative to discover high quality and cheap power resources in the country and around the world, accordingly bringing additional revenue for owners of idle power. When a large amount of data will generate demand for edge computing and distributed computing in the future, energy consumption cost will also become a common problem for them.

In summary, we see many similarities between Bitcoin mining and traditional data centers and future trends in AI and edge computing, which is no coincidence as these can be categorized as high-performance chip computing industries and are rapidly iterating with the rapid growth of computing demand.

3.Energy perspective on the POW mechanism – energy-based liquidity premium

POW (Proof of work), the consensus mechanism of Bitcoin, plays a crucial role for Bitcoin. The above analysis of Bitcoin mining is also a category in the computing industry, let’s extend one more step to think about the nature of mining. Miners know that mining has two main costs, one is the cost of the mining machine which can also be considered the cost of the chip, and the other is the cost of electricity, which as an ongoing cost input tends to take up the lion’s share. But if we look at the cost of electricity in a different way, the cost is the essence, so the essence of mining is the production of electricity-based activities, and the role of miners is like a globally distributed workforce that uses electricity as a raw material to produce the product Bitcoin. Treating Bitcoin as a commodity, then its cost is critical, and we all know that businessmen must have an incentive to find and use lower production costs, so miners will actively seek out cheap electricity. Further, if a commodity like Bitcoin is produced with electricity as the main cost, then Bitcoin can also be seen as a high value added product of raw materials like electricity. At this point for the owner of the electricity or the corresponding energy source (coal, natural gas, etc.), is it better to choose to sell the electricity and other energy sources as cheap raw materials or to produce Bitcoin on site for higher value added?

At this point we can see Bitcoin as a high value-added product of electricity (energy). Combined with Bitcoin’s own high circulation rate and low circulation cost, it is equivalent to providing global liquidity for globally dispersed energy resources such as coal, natural gas, and electricity. In the context of globalization, energy owners will have another revenue option besides selling through trade channels.

II. High-performance Computing is The Core Driver of The Fourth Technological Revolution

1.Computing is the bridge between the physical world and the virtual world

We analyze the flow path of energy in computing activities from the point of view of physics: (To simplify our analysis, we isolate the focal part of computing activities which is the chip) 100% of electrical energy passes through the chip and is converted into more than 99% of heating energy (and a small part of electromagnetic waves, which can be neglected), while completing the task of computing. Computing, whether processing data and information or completing computation, is a way to bring order to the means of production in the virtual world, so it can be seen as an entropy-reducing activity with electrical energy as input and heating energy as output. As long as it is computing, it will definitely consume the corresponding energy. Through computing, the production materials such as energy of the physical world and the production materials such as data and information of the virtual world achieve a real correlation, and we can even macroscopically think that the result of each computing is the mapping of the energy of the physical world in the virtual world.

2.Chip is the engine of high performance computing

From the third technological revolution to today, the accumulation dimension of production materials has been extended from the previous physical world to the virtual world, data, information and other digital production materials began to accumulate in large quantities, and the next 5G, IoT, AI and blockchain development will bring exponential growth of production materials in the virtual world, corresponding to such huge production materials, need very powerful computing power to complete processing tasks, and accordingly The same amount of energy is needed as input, and high-performance computing chips are the core role.

With the exponential growth of computing tasks, the chip from the performance and number of two dimensions of rapid iterative evolution, the performance of a single chip more and more powerful, power consumption is getting lower and lower, the total number of chips is getting larger and higher total power consumption. It is foreseeable that the future demand for a large number of computing, will bring the explosive growth of the chip industry, will also lead to the explosive growth of related industries providing energy services for computing. The chip is compared to the engine, electricity is like fuel, how much fuel can run how far, to how much electricity can complete the corresponding number of computing tasks, which is consistent with the reality of our perception of the computing industry.

3.High performance computing is all about energy-based POW

Putting aside Bitcoin mining and taking a global view of the high performance computing industry, this is an industry that uses electricity as a means of production and a major cost of production, so it is also an energy-based industry. Here we look at the Proof of work mechanism, where work is defined as “workload” in the Bitcoin mining mechanism, but work itself is the concept of physics doing work. Therefore, we can completely define computing as a class of behavioral activities in which electricity does work, so from this perspective all high-performance computing can be considered as proof of work mechanisms.

4. Energy is the source of all scientific and technological revolutions

Looking back at the successive technological revolutions, the first industrial revolution of steam, the second industrial revolution of electricity, the third technological revolution of atomic energy and information technology, all achieved rapid progress in productivity after the energy change. The fourth technological revolution will have a sky-high amount of digital production materials that need to be processed by high-performance computing, and perhaps this is the core of the fourth technological revolution, that is, energy-based high-performance computing to bring about the large-scale value of digital assets production materials. In this way, it seems that energy remains the source power of the fourth technological revolution, which is consistent with the objective law.

If each technological revolution is regarded as a cycle, then the production activities in the cycle is more of a demand-determined supply, that is, energy and productivity has achieved a breakthrough, by the market demand to determine how much productivity supply is needed, and constantly optimize the cost of productivity (i.e., energy costs). In the phase of cycle change, due to the rapid breakthroughs in productivity and the rapid increase in the amount of energy, greatly liberated the original productivity constraints reduce the cost of production, so this phase is the supply determines demand, for example, in the past to deal with a certain scale of data requires several supercomputers to spend a lot of electricity to deal with a month or even a few years, while the future to deal with the same amount of data only requires a chip to consume very little electricity with less than 1 second to complete the time.

III. The Great Integration of Energy, Finance and Technology

1.Energy & Computing

As mentioned earlier, both Bitcoin mining and high performance computing integrate the two industries of energy and computing through chips, and the computing industry can be considered as a production manufacturing industry with energy as the main means of production and production cost.

Let’s analyze the energy industry separately. At present, 8%-10% of global electricity consumption is used for computing. With the rapid development of the computing industry in the future, since electricity is currently the only direct energy source for the computing industry, this proportion will rise rapidly. Related research predicts that in the next 5 years, the power consumption for computing will account for 15-20% of the total global electricity consumption. We think this is a relatively conservative forecast. We analyze from the power consumption side that the demand market for manufacturing and living purposes (lighting, heating) is an industry that is closely related to population and national development. In the future, there will be no exponential growth trend, and the computing demand market is due to 5G The popularity of technologies such as AI and AI will produce an exponential growth trend, so the proportion of power consumption in the computing market will increase rapidly. In the last round of the popularization of the Internet and mobile Internet, a large amount of storage and computing demand has pushed up the overall power consumption of the computing industry. The rapid popularization of this round of new technologies requires both the volume of data and computing tasks and the speed of generation. It is much faster than the previous round, so the growth of power consumption in this round of computing market should also be much faster than before. If we look further, then in the future virtual world as a parallel world mapped by the real world, the total scale of its digital production materials should not be smaller than the existing physical world, and the energy consumed by these production materials should be processed by computing. It should be very large, so we might as well make a bold assumption: the computing industry will reach 50% or more of the global electricity consumption in the future.

At the same time, the rapid growth of electricity demand in the computing market will also drive changes on the supply side of electricity, and how to meet the demand for such rapid electricity growth will be a critical issue. Here regardless of the need to consider the growth space of power scale, but also need to consider the cost of electricity, because the cost of electricity will directly affect the cost of computing, and eventually return to the end-user cost of computing power. On the other hand, changes in the supply side of power generation will also extend to other related energy fields, nuclear power, thermal power, hydropower and other corresponding power generation methods will participate in this fast-growing market, the upstream energy supply (including oil, natural gas, coal, etc.) will have a linkage effect, and the global energy consumption market pattern will produce a new round of changes.

If we make further guesses, the cost of computing power, i.e. energy cost, will keep falling as technology advances, but the overall demand will rise rapidly, then will it reach a tipping point where the return on energy flow to other uses starts to be lower than the flow to computing uses, then this is when more energy resources start to actively flow to the computing demand market. This situation is already reflected in the global market for Bitcoin mining, and I believe it will show up in more high performance computing areas in the future.

2. Finance & Computing

Modern finance and computing are closely related, both in the global trading markets and in the independent data centers of financial organizations, which perform a large number of computing tasks every day. With further globalization, we believe that the financial industry will rely more and more on the computing industry: the rapid growth of demand for mobile payments, trading markets, customer data, market analysis, etc. will also drive the rapid growth of the computing market.

When we look at energy through the lens of finance, we find more relevance. If we were to give finance a key word, I think “liquidity” would be one of the most appropriate. As an asset, the price of energy (roughly, price = value) can be considered to be composed of two parts, one is the use value, which is the value of the use properties as fuel, power and other usage scenarios; the other part is the subsidiary value, which includes the value of a commodity in the secondary market influenced by supply and demand and liquidity. The aforementioned Bitcoin is more like a liquidity premium for energy, which is actually a reflection of the way in which use value and subsidiary value are combined. In addition to Bitcoin, the entire computing industry is actually combining the use value of energy with the collateral value to a fuller extent, bringing out the greater value of energy.

Therefore, looking at the above relationship between energy and computing, finance and computing, we can summarize: computing brings the two industries of energy and finance together closely, and in the future development, the rapid growth of computing and finance will drive the rapid growth of energy consumption.

3.Technology & Computing

Back to computing itself, the main carrier of computing is the chip. The chip industry is currently one of the core of the global cutting-edge technology industry, almost all intelligent devices and modern services are inseparable from the chip, for example, cell phones, computers and other intelligent terminals will be directly built-in chip, large data centers, edge computing centers are through the large-scale high-density chip to provide services to end users. The surrounding industrial ecology around the chip also basically concentrates all the most advanced technologies and enterprises, from the production of wafer flow to the production of PCB version, involving hundreds of disciplines such as materials, materials, mathematics and chemistry.

At the same time, the development of computing industry will also promote and attract the evolution and gathering of the whole technology industry. More advanced materials and technologies will flow into the chip manufacturing, more and more professional talents and enterprises will invest in the research and development of the chip computing industry, and more powerful services and resources will gather in the chip computing industry.

In summary, the computing industry will promote the integration of energy, finance and technology industries, energy, finance and technology itself is the three pillar industries of modern society, which are inextricably linked to each other, while computing is more like the intersection of energy, finance and technology, and in the demand for the promotion of this intersection of the three industries in the proportion of expanding.

IV. The Great Transformation of Energy, Finance, and Technology

1.The core of the computing industry – computing and energy

In physics, the relationship between force and work can be simplified with E=a*F*t (where E represents energy, that is, the amount of work done, F is the size of the force, t is the time, a is a parameter and has different values for different forces and energies), for example, the formula for doing work in Newtonian mechanics is W=FS=F*v*t, W=P*t in electricity, the force and energy in these scenarios are can be distilled using the simplified formulas we described above.

So how do we represent computing power? A straightforward way to represent it is to define the number of calculations per second or the ability to complete them, for example, the current Bitcoin miner’s computing power is in the tens of T/s, but such a representation does not see a direct relationship between computing power and energy. As we analyzed earlier, computing is an entropy-decreasing process that must consume energy and require energy input, while computing power is a momentary state that does not represent the result of the entire entropy-decreasing process of computing. For instance, the computing power of a mining machine is determined, but the results it can produce in one second of work and in one year of work are completely different, and the energy consumed is also completely different, just like the energy consumed in real life to move different weight of bricks to different floors is also different.

What is the best way to evaluate the results of computation? Similar to the relationship and definition of power and kinetic energy in physics, we define the energy of computing to complete the task over a period of time as computing and energy. We have analyzed earlier that computing is highly correlated with energy consumption, and if we approximate the matter of computing as a process in which the chip consumes electrical energy to complete the computing, from the perspective of the first law of thermodynamics of energy conservation, then we can use the consumption of electrical energy to indirectly represent the computing and energy, so the computing and energy is equal to the total amount of electrical energy consumed by the computing.

In the actual process, a device used for computing is studied as a unit, in which the vast majority of electrical energy is used for computing of chips and other electrical components, but there is also a part used for peripheral services such as heat dissipation to maintain the stable working condition of the chip. Therefore, we know that in order to improve the utilization of electrical energy to improve the calculation of energy, we must minimize the consumption of non-computational work on electricity.

At the same time, in addition to improving the utilization of electricity, improving the efficiency of energy calculation itself is also a very worthy direction to think about. The so-called efficiency is just two parameters, namely cost and revenue, so to improve the efficiency of computing and energy should start from reducing the cost of computing power and improving the revenue of computing power. And we look more macro, want to improve the efficiency of computing and energy, we need to reduce the cost of electricity and chip power consumption ratio, improve the processing power of the chip. This thing is happening from the first day of the computing industry, chips continue to use more advanced technologies and processes to reduce power consumption, data centers are constantly looking for cheaper electricity and reduce the total power consumption of the way.

2.Computing power drives energy reform

Continuing with the concept of computing and energy, the development of the computing industry will bring three main changes to the energy industry because of its close relationship with electricity consumption: 1. the total amount and share of energy used for computing will grow rapidly; 2. lower cost and more stable energy sources will account for an increasing share of the energy required by the computing industry; and 3. more types of energy resources will be converted to electricity.

As we said before, the current electricity consumption of computing accounts for 5%-8% of the total global electricity consumption, we should not underestimate this figure, because the demand for computing will be exponential growth, so the growth of computing electricity consumption will also show exponential trend, so this figure will grow rapidly in the next 10 years. From the perspective of electricity supply, the surplus of electricity production capacity can meet the growth of computing demand, but after a certain scale, we should start to consider the redeployment of electricity demand market, and the electricity from the original low output value of electricity-consuming industries will be shifted to the high output value of computing industries. After that, the deployment of the stock power market can not meet the global demand for electricity, more primary energy will be put into power generation to drive more rapid growth of the incremental power market.

Energy cost has always been the core proposition of human society development, so the computing industry can’t escape the constraint of energy cost. From a macro perspective, the growth of computing and energy demand will bring exponential growth of electricity demand, so more cheap or even unused electricity will be utilized, while more primary energy such as natural gas and coal will be concentrated into electricity to serve the computing industry. As the operation and maintenance service provider of computing power, it will have more incentive to adopt more advanced technologies to reduce the PUE of the whole data center, such as adopting more energy-efficient heat dissipation methods to reduce heat dissipation energy consumption, and adopting better infrastructure to reduce the loss of power transmission.

To add another point here, according to the above, people should first worry about the energy cost of quantum computing while worrying about the security threat brought by quantum computing.

3.Computing drives financial reform

Due to the substantial increase in computing power, the processing capacity of digital production materials has increased significantly, and the combination of the financial industry with the computing industry has led to a significant increase in efficiency and a significant decrease in costs, so the financial industry will also evolve in this direction: through the Internet and advanced technology to reduce the cost of trust, circulation costs, and transaction costs in the financial industry, while also driving the continuous growth of computing demand in reverse. Therefore, mobile payment, e-commerce, digital finance, global transactions will remain a long time in the future rapid development of the industry, computing power will penetrate the traditional financial industry from all dimensions, which is more like a process of empowerment, not simply Internet finance, but computing and energy finance, smart finance.

4.Computing drives technology reform

The computing power itself is also relying on technological advances continue to evolve, 5nm or even 3nm process will bring more low-power high-performance chips, the development of new materials will bring better heat dissipation. At the same time, with the rapid spread of 5G, AI, the IoT and edge computing rely on the computing power of the empowerment began to highlight the value of technology will promote each traditional products will have data access and computing power, these computing power will further promote the development and evolution of computing power.

Computing power may surpass atomic energy weapons to become the the country’s core competitiveness in the future. By then, all advanced science and technology will flood into this industry. The increase in computing power demand will bring about rapid growth in related industries, which has recently been favored by capital. The chip industry, as the country’s core strategy, will get more resource tilt and rapid development opportunities for a long time, and will drive the rapid development of chip peripheral industries and computing and energy service industries.

Just as the Internet has driven the development and growth of e-commerce, giving rise to giant enterprises including Alibaba and Amazon, which serve the needs of e-commerce, the rapid growth of demand for energy and the rapid development of the chip industry will also lay a solid foundation for the possible birth of future energy service giants.

 

Subscribe to SAI

By subscribing, you consent to Fluence storing your information and receiving our blog posts in your inbox. You may unsubscribe from these communications at anytime. For more information, check out our Privacy Policy.