Renewable energy can’t be ignored
– David Wanis, April 2019
We may be at the cusp of an economic disruption that rivals the microprocessor or the Internet. Driven by three mega trends: Renewable energy, electrification of the transportation fleet and automation / artificial intelligence.
The productivity drivers are similar to the microprocessor and Internet disruptions over the past 50 years – significant improvement in output per unit of cost, compounding over long periods of time, to great economic benefit.
In microprocessors, we saw this in Moore’s Law – a doubling in computing power per dollar of cost every two years. The Internet emerged, supported by a 50% increase in bandwidth per dollar of cost every year since 1983.
For many years, the key technologies that underpin the energy, transportation and automation disruption of the incumbent industries, have been on similar productivity curves to those we saw in microprocessors and Internet enabling technologies. The key technologies here are wind turbines, solar photovoltaic (PV), battery storage and matrix computing (GPU, TensorFlow etc.). Over 10 years, cost per unit has declined at compound annual rates of 21%, 20% and 49% for Solar PV, battery storage and GPU processing respectively. Most forecasts have these productivity rates continuing for the next decade.
10 years ago, we needed to estimate when the crossover point may happen and when significant economic advantages of renewable technologies would drive market adoption. This point has now arrived. We can move from prediction to observation.
10 years ago, the role of renewables was not considered feasible without government subsidies. This market is now evolving purely on economic merit. It is this merit that makes the case against renewables increasingly harder to make.
This is not academic. This is not just opinion. This is what we can all observe based upon investments being made today and the economic returns being generated. Unsubsidised renewable generation (wind and Solar PV), today have total costs lower than the marginal cost of coal power generation. Existing coal plants, even in developing countries such as India, are being mothballed as they cannot compete against renewable capacity coming online. Load balancing gas peaking plants are being de-commissioned as on-grid battery storage generates higher economic returns for owners. Continued productivity improvement of the underlying technologies means these renewable advantages will only grow in the future.
Auto manufacturers are shifting to electric cars. Not because ‘Tesla forced them,’ but rather because in a very short time, an electric vehicle will be functionally equivalent and cheaper to buy, run and maintain, than an internal combustion engine (ICE) based vehicle. And not by a little, by a lot. These bets are no longer being hedged – auto makers’ future investment in capital and capability is now dominated by electrification and autonomy. Because the economics are so compelling, if VW, Ford and Toyota were not serving this market, it would be Tesla, Rivian and BYD.
The role of artificial intelligence and full self-driving cars could be as disruptive as the above two trends. But given the general difficulty in understanding just how far advances in the science, software and hardware have come, it is easy for investors to be skeptical. The increase in technology capability over the next decade could be anywhere from 10x to 100x what is available today. Having high conviction this will never work, is as bold a call as Elon Musk saying it will be here next year.
It is now pure economics driving change across these industries. The environmental benefits are a (much needed) bonus.
For this change not to happen and for investors to be comfortable backing the incumbents, we would need to see these long-term technology productivity trends stop and reverse. We would need investments being made in renewable generation and automotive electrification to stop and reverse. For the whole field of AI to be a dead end (despite ongoing advances in computing productivity and software research), useful only in the small number of tasks in which it has currently succeeded.
Incumbents may step up their lobbying efforts and convince governments to act against the short, medium and long term, economic interest of their own electorates. However, this is likely to slow down rather than to halt progress.