Back to my tech hobbyhorse

[Suff]

I was talking about Tesla and their Self Driving computer recently. They are full on in Beta mode with customers testing the latest software on the road. It is going well and will probably expand beyond the US and to most owners, by January.

At the same time Tesla have just about completed their plant extension in China and are exporting cars to the EU from Shanghai.

Also at the same time, they are very rapidly building new plants at Berlin and Texas. Completion of these plants and at full capacity, it will give Tesla the same car manufacturing capacity as Ford. At around 4.5m vehicles, in the 2022 time frame.

Tesla's goal is to take 20% of the global new car market by 2030, at close to 20m vehicles per year. Or twice what Toyota (the largest vehicle manufacturer in the world today), produce.

I would not bet against Tesla on that.

However this begs a question.

The UK New car market is around 2m vehicles per year and 20% of that is 400k. Until the middle of 2019 that was the total world capacity of Tesla from their Fremont plant. So it is a fairly good bet that Tesla would want to build a plant in the UK, rather than import from Europe or Asia.

Musk has already been in contact with the Government on tracts of land, so this may come to pass in the next 1-2 years.

I think that is a good thing. I would prefer it was a UK company that did this but, as has been seen by Tesla, you need a razor sharp focus, massive capital markets and a leader right on the spectrum, to break into the vehicle market.

I shall continue to watch it unfold. Tesla is getting a name for making car manufacturing facilities 2-4 times faster than their competitors.

[Workingman]

Do you have shares in Tesla?

To the end of 2020 Tesla deliveries are expected to be between 477,750 to 514,500 cars in total, worldwide. Its Gigafactory 3 will eventually approach 250,000 vehicle by 2021. It is going to have to do some massive ramping up the get anywhere near 4m units by 2022.

Yes, its percentage increases are very good, but they are from a low starting point and sooner rather than later the Pereto Principle will kick in. And don't forget that all other manufacturers will be moving from ICE / CIE production to EV - the market is only so big.

And Elon needs to be careful not to waste too much cash on the sinkhole that is the useless Hyperloop.

[Suff]

Do you have shares in Tesla?

I wish, I was out of work in 2018 when they dropped to under $200 and could not buy any. The shares entered 2020 at just under $400 and, after a 5 way split, are now just over $400. Over $2,000 per share equivalent.

Current run rate for Shanghai is 800 per day on a 7 day week. They run a 48 week year, so that is 268,800 per year, Model 3 only. They are in the completion stages of the Model Y plant on phases 2/3 of the Shanghai build. Initial Model Y volime for 2021 is sized at 300,000. However Model 3 was originally sized at 250,000 and has already exceed that.

Expected final manufacturing capacity at Shanghai is 1m vehicles around 2022

Giga Berlin is 3x the land area of Shanghai and has a phase 1 expected volume (Model Y initially), of 500,000 vehicles. However it will not go into operation until May/June 2021, so will only hit that cadence around December.

Final manufacturing capacity of Berlin is expected to be 2m vehicles, models 3/Y and a new €25,000 smaller vehicle.

Giga Texas is 3 times the land area of Berlin, scoped for 500k vehicles phase 1 and no quoted figures for final volume. It will manufacture the Y and the Cybertruck initially.

Fremont is already producing at 550k vehicles per year run rate, models S/X/Y/3 and will continue to increase in 2021. It has two sprung structures which will eventually become hard build production lines.

The next milestone for Tesla is expected to be 500k vehicles produced in 2020 with Shanghai phase1 and Fremont only. That will be over 180k in one month, or significantly more than half the total 2018 deliveries.

By the end of Q3 2021, Tesla will be producing more vehicles per quarter than it did in the entire year of 2018.

This is a story worth watching.

Even if Tesla do not hit 500k this year, which was their target set in Jan, it will be no huge deal given both Fremont and Shanghai were shut down by covid for 6 weeks this year.

If they do hit it, then it will be a significant over deliver.

[Suff]

When I set out to write the post initially, I had not read this article from the Guardian.

[Workingman]

But I thought that in the future we would stop owning cars and just call up a driverless EV pod using an app - no buses, no cars, no congestion and no emissions, anywhere, ever.

Or perhaps we will all be using flying drone taxis? Then again, maybe not.

[Suff]

We can't all use driverless taxi's and there will always be a place for the family car in most situations.

However, there are plenty of places for the driverless taxi. The commute to the train/bus station is one case in point. Others are people who live in cities who buy cars for infrequent use because there is no real choice.

The interesting point here is that a company which has succeeded in breaking into the mainstream automotive market (the first time in the US in 100 years), who has designed and engineered hardware that beats Nvidia at it's own game in the automotive control hardware and has created software with a neural net which thrashes anything Waymo can produce; is determined to give us that choice.

So let's see. People who live in cities and own vehicles for certain shopping trips and a once or twice a year holiday trip are going to say "no I won't use a driverless taxi because......."

If for no other reason WM, I would think you would be interested in how Tesla is creating a driving intelligence using 8 cameras radar and sonar, without Lidar. I liken this to Waymo going around Lidar mapping zones then putting a smart Circus chimp in the driving seat whereas Tesla has, today, a 15 year old novice driver behind the wheel. The Circus chimp can't go anywhere that doesn't fit with the extremely tight parameters set and the up front information provided by the mapping. Eventually the Tesla will be able to go anywhere.

In order to do this, Tesla had to create hardware that is 4 years ahead of its competitors. They did this by hiring Jim Keller (AMD K8 architecture and Apple A4/A5), then gave him a team and 3 years. He produced silicon well ahead of the competitors. Nvidia's offering burns 5 times as much power for a theoretical boost of about 30% over Tesla. Always remembering that a Tesla drives on the battery pack it carries around with it and has a limited space for the computer with limited cooling. Tesla's 4th generation hardware is slated to ship Q3 2021 and will be another 3x boost in speed. Tesla owners with older vehicles, who buy full self driving, get the computer for free as a plug in replacement.

Tesla has 8 cameras running at 30fps, 240 fps output. But the hardware can handle over 2,000 fps of full motion video. The logic engine analyses 4d information (3d plus time, otherwise known as full motion video) and creates multiple projected 4d paths. It then chooses the best path out of the simulated. All in real time.

Then Tesla realised that training such an AI driver would take a team of tens of thousands. Tesla vehicles drive billions of miles each year and tesla gets data on the simulations and projections on a large chunk of them. So they created a training net and a completely new style of labelling so that labelling could be done initially then automatically and the training net could consume it.

Google created a game playing AI. It took 3 years to write the software. It took 3 days to train it with 3 years worth of recorded gameplay!

So we have a company that is knocking on the door of being one of the largest vehicle manufacturers in the world (if not the largest), which is producing the most advanced self driving software in the world and the most advanced neural network self driving software trainer in the world.

They have applied their engineering skills to building factories. 2-4 years the competition said as they laughed at Tesla. When Tesla came in with a China factory, fully functional, in just one year, they stopped laughing. They have just released, on their battery day, information on how they have reduced the footprint of a battery manufacturing plant by 90%, are able to build larger battery cells faster and will go into vehicle battery manufacturing over the next decade.

https://www.youtube.com/embed/l6T9xIeZTds

It is worth watching for the sheer scale alone. Those quoted figures would, by 2030, power around 50 million new Tesla vehicles per year. Or 20m plus a few hundred thousand 38 ton articulated electric trucks with a 500 mile range.

This is an industrial revolution change in how we get around the world and, because Musk is such a controversial figure, the press prefer to sneer at him.

Never mind the fact that they are also producing solar products including a whole roof with integrated solar, Home power storage (powerwall), grid scale battery packs and software to manage your grid and your megapacks so that your megapacks can remove coal and gas peaker plants and reduce the cost of electricity (Australia).

Personally I think it is worth utilising a few braincells over.

Oh yes, I forgot to mention. Tesla does monthly updates on the vehicle systems. Literally, after the 8 year warranty, your tesla is more capable than it was when you bought it. Spotify being the latest addition to the in car entertainment system. What other vehicle manufacturer does this? Today you want more features, you buy a new car. Don't even think about VW and their offering (crapware). Their testing increased the bugmark by 250 every drop they tested to remove the prior thousand bugs.

Have a review of the competition.

https://youtu.be/2hn31vLKgho

Finally, after this long brain dump. It is worth remembering that Musk was the CEO and largest single shareholder of PayPal, set up SpaceX and ramrodded a company that has produced reusable spacecraft (thought to be too impractical to use) and is on a path to become the largest internet ISP in the world.

[Workingman]

I am not against all the new hardware and software and the science and engineering making it all work, but it is for the few and not the many.

When the news broke about the 2030 ban on ICE & CIE the news channels had plenty of slots to fill with opinion. When asked about the electricity supply there was lots of talk about the modern efficiency of our electrical appliances and that we were building in new provisions for it. It went little further than that. When the infrastructure for charging was mentioned it was admitted that more public charging points were needed but that there was an ongoing installation plan. Thank you.

Then I looked round my leafy suburb in a large city and it didn't look so simple.

My area has one road in and one road out. It has 64 units each with four flats, 256 homes, and there is not one off road parking bay - not one. So I got to wondering how we would charge our wonderful EVs. Would it be with extension leads from a window across the paths to the cars? And what about when we cannot park outside our properties, we do not have dedicated parking slots, and have to park away or across the road?

Then I looked at the 340 unit housing association estate up the road. It has 60+ semi-detached bungalows surrounded by three storey flats. There are roads down ether side and car parks with ginnels through to the staircases and bungalows. No electricity provision at the car parks.

The 60s / 70s open-plan estate of about 800 units a bit along has six winding cul-de-sacs meandering in to the estate, It was designed as a pedestrianised zone with parking bays and, again, no dedicated parking. Then there is the large council estate of the post war years designed when cars were a rarity. Very few of the semis have drives and it is communal parking only for the many maisonette blocks.

A visit to the core of the city is a landscape of terraced housing from the Georgian, Victorian, Edwardian and post WWI periods. Many cities are similar in layout, even London, we must not forget London.

Then there are the chocolate box towns and villages of the Downs, Cotwolds, Cheviots and Dales etc.. Many are now by-passed because of the congestion created by cars parked on the roads as they had nowhere else go made then un-navigable, and don't forget places such as York, Bath, Chester, Lincoln, Warwick......

To hear the proponents of EVs you would be forgiven for thinking that we all have double driveways and our own charging points. If only that were true for many millions of us.

Now, today, we hear that Sunak has a tiny bit of a problem with all this EV stuff - a £40bn problem. When we all swap to EVs (it won't happen but play the game) he loses all that fuel duty, the VAT on the fuel and the fuel duty, and VED. Now he can change VED so that all those smug EV drivers chip in, but he still has a £28bn hole to fill. He could, of course, tax electricity, but that would be political and electoral suicide. So, what does that leave. Well, according to the press he has "pay-as-you-drive" or "pay-per-mile" as his best bets. Tough call.....

It's a real sod in life when reality impinges on the faery tales and dreams.

Sorry for the length.

[Suff]

NP on the length, mine was too long also.

Yes it is a challenge. But unless we challenge ourselves, we will never do it.

The interesting fact is that the average journey in the UK is less than 20 miles per day. With a Telsa equivalent vehicle with a 240 mile range, discharging to 50% and recharging can be done once every 4 days. 80% charging from completely empty can now be done in around 40 minutes at fast chargers which means that when, not if, public fast chargers become pervasive, charging at shopping trips becomes viable. Or even getting a coffee in the morning/evening if charging from 80%.

For the power infrastructure, that was a surprise to me too. We use less than our available capacity by night and renewables are often wasted due to the lack of power draw when solar/wind sources are high. In reality, a small increase in our generating power, spread over a charging window of time, can accommodate most of our electric vehicles. Remembering that Electric vehicles average about 120mpge in fuel usage. Current UK MPG for the fleet average is around 55mpg for new vehicles. So whilst vehicles burn a little more than the entire electricity grid generation, electric vehicles will burn a maximum of half of that.

With heavy trucks (articluated), this has also proven to be true. The comparative MPG are, UK Arctic 8mpg and Semi 17mpg. Nobody expects heavy trucks to transition in a decade.

At 2m cars and vans per year year it will take an absolute minimum of 15 years to refresh the existing vehicles on the road. Even longer as older vehicles will stay on the road longer. If we cannot transition our grid and our charging in 25 years, then we are incompetent in the extreme. Given that the additional load will be around 1/8th current generating capacity.

On top of this, battery and charging tech is changing at a rapid rate. By 2030 it is likely that cars will be able to charge faster with less load on the system with less heat than today. We don't have time to say "Oh it's too difficult, lets leave it", just because Norway is about the only other country doing this now.

One of the problems with visualising how we will power these vehicles is that nobody ever recognises that when you plug them in is not necessarily when they will charge. Changes will have to happen, but let us say I plug my car into a public charging point whilst at work, it sits there all day, all the vehicles that need to be charged negotiate with the charging network and tell the network how much charge they need and how long they have. The network balances the power to all of the vehicles. Equally at night.

Power delivery can be priced and targeted for peak utilisation at peak excess power.

There is then V2G. When it becomes advanced enough, you will be able to say "OK I've got 80% in my battery and I'll allow you to drop that to 60% during the day and you can recharge me at night to 80% or 100%. Times even 30 million vehicles with an average 50kw/h battery and you suddenly have enough power to start balancing the variability of wind and solar.

The main problem is that it's all too difficult looking from the outside. Until we force ourselves to make the transition and do the work, it will always be too difficult. Just like it was too difficult to stop putting lead in petrol.

Granted my view is different because I talk to people, who know far more than I do about this, on a daily basis. Including one of the founders of the UK V2G initiative which has government support and funding. He gives me a totally different view of where the UK will be 15 years from now than you will see in any press articles.

[Workingman]

It's hard work with you Evangelicals.

To overcome the practicalities of reality all you have to do is throw down some magic numbers and it's job done.

[Suff]

It's hard work with you Evangelicals.

To overcome the practicalities of reality all you have to do is throw down some magic numbers and it's job done.

Not at all, it requires a HUGE amount of work and a lot of money.

But the longest road is the one never started and we have been avoiding starting it for way too long now....