# Why the world needs 12 times today’s clean electricity generation

Last month I had the honor of giving this TED Talk on why the world needs 5 times as much total electricity generation — 12 times as much clean generation — as we have today.

Here I lay out the methodology behind this estimate, so those who have started citing the talk as a reference point have the scientific backing. This isn’t a peer-reviewed academic analysis; it’s one independent scholar’s best guess at what we should be aiming for. And, crucially, it’s not a prediction of what we’ll have: it’s an estimate of what we need to aim for given the realities of the world. To the extent that global development proceeds faster than expected (which would be great!) we will need even more. To the extent that political action is stronger than expected, we could need less. But **if we want to be confident of achieving net-negative emissions by around 2050**, assuming imperfect policy in at least some major countries, **this is what we must aim for**.

The world generates about 25 PWh (petawatt-hours, which are trillion kWh) today (source). I rounded down from 27 for the talk.

Over 1/3 is already clean (same source). To be exact, it seems to be 38% in 2020 according to this report’s numbers.

Clean sources are projected to generate roughly 25 PWh in 2050 (source). I downloaded the IEA scenario data and summed the TWh projected for 2050 for each clean source in the baseline scenario, which came out to ~24 PWh.

**Addition #1: Electrification**

Baseline for total electricity generation in 2050 is around 40 PWh (same IEA source). That report has 145PWh total final energy demand (excluding non-energy use), which means 28% electrification (ignoring grid losses for this order-of-magnitude estimate).

60% electrification is ambitious (source, see bottom right corner of very last chart, second to last page in pdf).

Electrification lowers total final energy demand because electrified equipment is more efficient; in my book I looked up various comparative electric vs fuel equipment (examples here and here) and used the conversion that electric equipment uses about 30% of the total energy for a given process, so for every 1 PWh of electricity added at this step, total final energy goes down 2.4 PWh.

Math: using that 2.4 number this time, I found the amount of electricity to add to the baseline such that (when you reduce the total final demand by 2.4x) you get 60% electrification. But the IEA baseline of 40 PWh already includes my Reason #2, developing countries’ growth, so I used the % electrification in their baseline (28%) with a total final energy demand of 100 PWh (rounding down from current levels of 110–120, assuming that non-electrification efficiency would lower demand in richer countries, so when ignoring global growth, 100 seems reasonable). Based on those, you have a baseline of 28 PWh of electricity and you calculate how much to add to that (subtracting 2.4x that addition from the total final energy number) such that you get 60% electrification, which comes out to 41 PWh of electricity and 69 total final energy, which I’ve rounded to 40 of electricity.

**Addition #2: Global Development**

To do my reason #2 you just go back to the IEA baseline scenario for 2050, which includes projected economic development, and you make that 60% electrified in the same manner, which comes out to 60 PWh electricity and 100 PWh total, roughly rounding.

**Addition #3: Synthesized Fuels**

To calculate the number with synthesized fuels, you subtract the 60 PWh electricity from the 100 total to see that there are 40 PWh of non-electric final energy demand in this scenario we’ve created. Since we’re talking about a net-negative emissions world, pretty much all of that has to be emissions-free. 19 PWh of sustainable bioenergy could be used in 2050 (source, page 25 of pdf) to supply some of that 40. If fossil fuels with CCS plays a very limited role due to imperfect policy (it is always more expensive so will only happen where mandated — and only applies to stationary factories, not vehicles), then most of the rest of the 40 has to be synthesized carbon-neutral (or carbon-free) fuels. I round to 20 PWh of synthesized fuels, and assume an average synthesis efficiency of 65%: hydrogen itself can be produced with up to 80% efficiency (sources here and here) but some of the synthesized fuel will be ammonia (hydrogen combined with nitrogen from the air) or artificial jet fuel or diesel (hydrogen combined with carbon split from CO2 in the air) which means an additional step and much lower overall efficiency — this would change depending what portion of each fuel you assume, and I don’t know exactly how efficient the secondary processes are, but 65% is my best guess for a rough overall average. That means that to get 20 PWh of fuels, you need about 30 PWh of input electricity.

Hence adding 30 to the 60 for 90 PWh in step 3.

**Addition #4: Sequestration**

For sequestration, I used the PWh per gigaton of sequestered CO2 from this article, which is 2.8 PWh/Gt. And I assumed 10 Gt per year of direct air capture in 2050, which is a semi-arbitrary number — it’s about the amount of remaining emissions (mostly from agriculture) that we can expect in 2050, and so you need 10 Gt/yr of total sequestration just to reach net-zero. Some of that will be growing forests, soil management, maybe rock weathering, etc. Some will be DAC. You want all methods to add up to considerably more than 10 Gt/yr so that you have meaningful net-negative emissions. Given our current rate of addition of ~60 Gt/yr CO2 equivalent, 10 Gt/y net-negative emissions seems the minimum to feel comfortable of pushing temperatures back down to manageable levels over a reasonable timeframe (many decades, not many centuries). So I’m assuming here something like 20 Gt/yr total sequestration, half of which is DAC (or other methods that require some electricity), which is reasonable because other methods are more physically limited (ex: growing forests). There is a bit more detail and some sources on this in my book, but this number is definitely arbitrary and ambitious.

With 10 Gt/yr DAC at 2.8 PWh/Gt, you add 28 (I rounded to 30) PWh to the total to get the final total of 120 PWh.