Cheap Electricity and Robotics

While robots can in some cases reduce energy intensity, the promise of automation depends on an abundance of electricity.

Source: Robotworx

People talk a great deal about robotics and automation. Less discussed is the demand it will create for electricity. Would higher levels of automation require more electricity or less? First, it is worth getting a rough estimate of global electricity demands from robots today and in the future.

Calculations

Based on a 2017 paper from Applied Energy Group, an average industrial robot operating for 20 hours a day across a whole year has an annual usage of 22 Megawatt hours (MWh). There are around 3.5 million industrial robots installed worldwide. Based on this, the yearly consumption of all industrial robots annually is about 77 terawatt hours (TWh). This is comparable to the electricity consumption of Algeria or Finland. It represents 0.28% of the electricity demand for the world in 2022.

That is for industrial robots, which are the most energy-intensive systems from an annual electricity standpoint. They have heavy payloads, are generally very heavy themselves, and are operating near constantly. Now, using the same paper’s numbers, the total load of professional service robots is 28% of industrial robots, while for domestic robots it is 15%. So, by a rough estimate, all robot electricity generation in 2022 amounts globally to about 110 TWh. This is slightly more than all electricity demand for the Philippines and slightly less than all demand for Kazakhstan. It is less than the electricity demand for crypto mining in 2021 (140 TWh) and far less than the 320 TWh used by data centers that year. Data center electricity demand globally is set to expand to 2,000 Twh by 2030. Whatever demand robots bring, it will be tiny compared to this market.

So what will robot electricity demand be in 10 years? Well, from 2012 to 2022, the compound average growth rate (CAGR) for industrial robots was 10.09%. Let us assume optimistically that this holds for the next 10 years up to 2032. This would mean an additional 9.5 million industrial robots, which have a lifespan of about 20 years. Let us add to the current installed base of 3.5 million and provide a 20% deduction to account for retirements and the phasing out of old equipment. This leaves us with 10,425,000 industrial robots. Use the same average annual usage per system, and the yearly load of industrial robots would be 228 TWh. Of course, forecasts are never perfect, and electricity efficiency could reduce this, but it works as a respectable prediction.

Assuming the load required to power professional service robots and domestic robots is roughly the same relative to industrial robots as it is now, the total load for all robots worldwide would be 326 TWh. This would be more than the 306 Twh produced by the entire British electricity grid in 2021. In fact, domestic and professional service robots are likely to increase in market share relative to industrial robots, so their contribution to electricity demand could be greater relative to now. Let us assume their shares grow by 25% in 10 years. In this scenario, the final prediction for electricity demand for industrial, service, and domestic robots worldwide by 2032 will be 351 TWh. 

Figure 1: Global robotics electricity demand, Source: Based on 2017 Applied Energy Group Paper and International Federation of Robotics (IFR)

From these back-of-napkin calculations, we can reasonably speculate that total robotics electricity demand will exceed 500 TWh mid-century, having broadly the same impact on demand that Germany has today. Of course, this would be heavily localized in manufacturing and commerce and would be more present in rich countries. The expected expansion of automation would therefore have a noticeable impact on energy consumption, but far less than the projected increases from data centers and network use.

Based on other figures, global electricity demand will be around 32,000 TWh by 2032, and 41,000 TWh by 2050. Assuming that robot generation will be 351 TWh in 2032 and 500 TWh in 2050, it will have increased from 0.28% of electricity demand today to 1.1% of electricity demand in 2032 and 1.2% in 2050. In essence, robots are going to be a bigger share of overall energy consumption. 

Will robots improve energy efficiency?

But of course, these robots are doing tasks that might otherwise be done by humans or other machines. Whether automation will reduce electricity demand is very sensitive to individual cases. Applying drones to kill weeds on mega-farms could massively reduce the need for herbicides and pesticides. Productivity-enhancing measures like this could reduce the need for energy-intensive inputs, especially those which require the use of fossil fuels. This application for drones has not been scaled up significantly at this point. 

When talking to roboticists, I got to the question of electricity use, and it was not considered to be a big problem. The roboticists in question were selling automated forklifts to big equipment manufacturers in the automotive industry. Such customers are already used to being energy-intensive, so onboarding a few robots was likely not a great challenge when it came to electricity costs. For larger deployments, analysis is tricky. Amazon is the largest deployer of automated guided vehicles (AGVs) in the world, with over 500,000 robots deployed in its fulfillment centers. It currently does not disclose the effect of this on its electricity bill.

It is also worth factoring in the efficiencies that may come as a result of automation. Automated vehicles move slower and more predictably than manual vehicles in workplace environments. Robots can operate in lights-out facilities, reducing the use of other machines and people. Assuming output does not increase by an order of magnitude, automation could well reduce overall electricity demand. 

But this is far from certain for many case studies. Vertical farming, a highly popularized case of automation, has recently been in the news for the wrong reasons. The proposition flourished in the era of zero-interest rates and cheap energy. Present circumstances have seen investments dry up and companies go to the wall. Vertical farms have a higher average energy use at 38.8 kWh per kg of produce compared to traditional greenhouses, which average 5.4 kWh per kg. There is also a question about whether vertical farming could be justified on land-use efficiency. 

The Energy Theory of Everything

While we might expect automation to improve electricity efficiency in the long term, actual robot deployment is being driven by energy-intensive manufacturers and energy-hungry logistics giants like Amazon. These companies are high-CAPEX, large consumers of energy. Robots improve productivity in the long term, but they are not simple investment decisions. While modern 6-axis industrial robots have price tags around $65,000, their overall cost to deploy is usually triple that, so closer to $200,000 for one system. 

Countries that excel in robot adoption rely on big corporations that have the spending power to invest in this technology. Such corporations want a stable electricity market. It is unsurprising then that the countries with high robot density are also the ones with low and stable industrial electricity prices. 

Figure 2: Robot density correlates with low electricity prices. Sources: UK BEIS and IFR

While correlation is not necessarily causation, it is also no surprise that Britain, the exceptional laggard in robot adoption, also has exceptionally unappealing industrial electricity prices (see above). Now the UK does maintain an impressive record in energy-intensive data centers. London specifically is the third largest city-wide market based on data center power capacity, behind only Beijing and the Northern Virginia area. But this position is increasingly only maintained by costly sacrifices elsewhere, such as placing moratoriums in homebuilding in West London.

So while a long-term effect of automation in certain domains will be greater energy efficiency, its promise is far more likely to arise with the help of cheap electricity. Many discrete policy prescriptions are being developed to turn the British economy around and increase productivity. But bringing down electricity costs precedes any success in these measures.

It might be time to pen “The Energy Theory of Everything”.