I know most people don’t believe this, but it is likely that electric cars have a good chance of becoming far cheaper than ICE cars over the coming 5-10 years.
EVs are complex, but in ways that are much more subject to economies of scale.
For example, much of their value is in the software, which replicates for nearly zero marginal cost. Electric motor assembly is highly automatable and a fraction of the cost of assembling engines. There are no emission control systems, fuel systems, exhaust systems.
That leaves battery materials and construction, and these are dropping fast. Goldman Sachs just predicted pack-level prices will be below $100 KWh by 2025 and continue dropping by 11% per year throughout the decade.
In fact, BYD already sells a perfectly-reasonable EV in China for around $11,000 US equivalent.
There will, of course, always be premium priced EVs in the market, but I believe the lower bound of the market will be much lower than the Corollas, Tercels and Honda Fits of today.
I'm in the US and I don't see any current manufacturer lowering their price over time. They'll either have higher margins or they'll just add more features to raise the price.
Average car price in 2002 was ~$19k.
Adjusted for inflation is ~$33k.
Average car price in 2024 is $48k.
If a company was run in a more sane way where they want to provide cars at a reasonable price where the company can pay it's employees a fair wage then maybe that could happen. Instead, most US corporations pay employees as low as possible, take as much money from the consumer as possible, and have the excess go to executives and shareholders.
No, but after having checked it looks like prices have dropped but also go up and down so word is still out if they keep dropping on scale or if it's just stock price shenanigans.
Watch Australia. I feel the same way about the incumbent manufacturers and their pricing and chasing the higher end market. But BYD has come and captured a huge chunk of the EV market, and that is before releasing their low cost vehicles.
They could really only do that as long as zero interest rates and relative wages permitted. Frankly I think they made luxury EVs to justify the battery costs early in the production.
The automotive industry is known for bad margins. Interest rates go up, willingness to pay for large ticket items goes down. We won’t see anything like 0 percent rates for possibly 100 years.
One interesting outcome: near shoring of production drives local labor wages up for the long, permanent term. Maybe more in India/Mexico than the US, but still significant. And better for all than the past.
I'm curious whether the increasing popularity of SUVs is skewing the average cost upward. There appear to be numerous sedans available in the market priced between the mid-$20,000s and low $30,000s, so the average figure of $48,000 seems somewhat inflated.
US vehicle margins are in steep decline, and competition is fierce. Sorry but your comment is in opposition to reality. Any company who could do what you say, produce good cars cheaply, would be an instant hit.
I'm not disagreeing with what you're saying but your comment is the one that's not in reality. Yours is a hypothetical since it's not happening, and maybe not possible.
> For example, much of their value is in the software, which replicates for nearly zero marginal cost.
People have been saying this for decades now and conveniently forget that replacing CD cases with Internet downloads doesn't eliminate the fact that software engineering is expensive in itself.
Crazy how TurboTax isn't $5 or Tesla FSD doesn't come standard. So much for zero marginal cost.
There isn't really a reason to expect a significant effect, in practice, Tesla COGS/unit plateaued a few 100K ago. It's a bit too abstract and high-level.
Batteries specifically does make sense, ex. Model X has 100 kWh battery. if I price out 100 kWh at $200 right now (high) and its $100 a decade from now, about $20K.
Not everything has the same unit economics. The self-driving software is one part of the costs that is heavy on the fixed costs side and will therefore scale better than other costs. If you produce a million cars vs one car, the cost to make that software isn't a million times more expensive.
There are examples of this, yes [0]. The phrase "run . . . on Tesla" might have different meanings. OpenPilot is not installed directly on the Tesla hardware (as far as I'm aware). OpenPilot runs on separate hardware that communicates over the car's network, typically CAN bus [1]. An analogy might be attaching USB devices to a laptop, which has its own video, sound, keyboard, and pointer; but for which a user might desire some other feedback and control method.
I really hope automakers don't start embracing touchscreens even more. Tactile feedback / muscle memory is important for safer operation. At least Mazda seems to get that[1], if only others would get over their minimalism obsession. Would embrace regulations requiring physical controls for most common tasks.
I want the year 2010 car experience with some of the EV benefits. So electric motor/battery/charging/steer and break by wire/classic dials for speed/classic climate control controls/classic signalling and lighting controls/... Maybe Carplay/Android Auto, but no tracking or operating controls that are touch screens. Also, a way to disable the display with a button.
I don't think anything is brake by wire, even that Lexus has a failsafe mode that ultimately is a direct physical connection. No saving parts there yet.
Like the other commentor said, the brake is still connected to the brake pedal and the car mechanically assists for one-pedal mode. You can always still use the brake pedal even with any automation or assistance off.
<thing>-by-wire means there is a control with no physical connection to <thing>.
for example, older tesla cars can steer themselves, but the steering wheel still connects to the front wheels mechanically.
The newer cybertruck has a steer-by-wire. The wheels are turned side to side by motors, but the steering wheel is not mechanically hooked to it, it's just an electronic controller.
The one-pedal though is only a mechanical assistance to the existing connection between the brake pedal and the brake. Maybe we're splitting hairs, but its more akin to power steering.
With [traditional] power-assisted steering, there is always a direct mechanical connection between the wheel and the steering box/rack. It's assisted, usually hydraulically, but it is not isolated by a wire; that direct mechanical connection is always there.
With [traditional] power-assisted brakes, there is always a direct connection between the pedal doing the braking and the wheel cylinders. It's assisted (usually, but not always, by vacuum), but that direct mechanical connection is always there.
With one-pedal driving that only uses exactly one pedal, there is never a direct mechanical connection between that singular pedal and the braking system. That connection is only electrical; there is no direct mechanical (or hydraulic, to split hairs) connection at all. It is entirely brake-by-wire, unless one chooses to place a foot on the brake pedal and thereby do something other than one-pedal driving.
Thus: Unless there's hydraulics connected to the accelerator pedal, then: As long as mechanical braking (pads-on-rotors friction) can be performed in one-pedal mode (and it can be), then one-pedal mode must perform braking-by-wire.
We're already there, I think; people use brake-by-wire every day. (That we also have a functional hydraulic brake pedal as a backup does not mean that brake-by-wire is not a thing that one-pedal-mode provides.)
(If that sounds aggressive, then I apologize. I'm still recovering from a vacation with a 30-hour drive, and I may be up too late, and may also be drinking a bit much tonight.)
You skipped the complex heat pump system needed to regulate battery pack temperatures (and used for cabin temps as well).
BMW i4,i5,i7,iX built in mid to late 2023 use an outsourced system that widely failed in cold temps. Tesla has a famously ingenious simpler but efficient custom in house design.
This kind of comment is perplexing to me. I'm always really confused why people try to claim heat pumps are complicated because they're not really (especially compared to many of the systems in ICE cars). Most cars have even already contained a heat pump for many decades (an air conditioner is exactly the same, just moving heat the opposite direction. It literally takes one extra valve to make it bi-directional).
There are videos of hobbyists making heat pumps from compressors salvaged out of fridges or ice makers, and other scrounged parts, maybe an Arduino controlling it! They're amazingly simple when you see what they're made up of.
Sure, it sounds like there were problems with this system used in BMWs, but that's likely just a design issue. Pumping a coolant around a battery and then through a heat exchanger (that is part of the heat pump) is not that difficult compared to the cooling of an extremely complex internal combustion engine!
The complexity and ingenuity in the Tesla heat pump system is the octovalve, not the heat pump. 8 different places to source/sink heat, with different set points. Keep the battery, motor, interior each at different optimal temperatures with a single heat pump.
Right: it’s not conceptually complex and Tesla, as i mentioned, shows such.
The parent comment to which i replied was listing significant systems in an EV, and hadn’t mentioned thermal regulation.
BMW’s outsourced Vitesco (part of Continental Power Trains originally) design differs in engineering (materials, components integration), with more disjoint components and more components therefore needing interconnections, and had cars (service bulletins applied to thousands) across Finland, Germany, Canada, US, UK, failing and leaking in a juncture not present on the Tesla design. (“Ask me how I know…twice.”)
Look at the work of Tony Seba, who has analyzed the properties of a possible fully-renewable grid.
In his models, you over-provision PV and wind (doable because it is so cheap, and getting cheaper) so it can cover peak loads. Then during the 90+ percent of non-peak time there is super-cheap excess power available which he dubs “super power”.
Using pricing signals, this can be used to enable activities which would normally be cost-prohibitive under current models. When power is cheap, crank up the desalinators, electric steel and aluminum mills, pumped storage, hydrogen converters, bitcoin miners, and anything else smart people can devise to do with inexhaustible, inexpensive clean power on a somewhat intermittent basis.
His model also shows that you can choose trade offs between how much battery you invest in and how much generation you build, allowing you to optimize costs depending on how much or little of this “super power” you want to generate. More batteries = less generation and less excess power, and vice versa.
I think many find this unintuitive because they formed their opinions of renewable energy during the early, high cost parts of its ramp and have not adjusted their thinking to the current part of the S-curve.
But renewable energy is already the least expensive source of electricity in most markets. It is our amazing good fortune that it is also environmentally fairly benign, extremely safe, has very low maintenance costs, can be geographically distributed to reduce grid distribution costs, and is able to scale with few limits aside from simple land availability. And its Wright’s law cost declines appear unconstrained for the foreseeable future.
This is the key part that most people are missing.
Standard thermal electricity generation is a very mature technology, no matter the heat source. Turbine-based combined cycle natural gas generation was a bit of a step up, but it used already mature tech: jet turbines.
Solar and wind and batteries are fundamentally different, and more like integrated circuit technology or DNA sequencing: falling in cost at a tremendous rate.
Looking at current prices or current installed bases of solar or wind or batteries is not very informative for assessing the technologies. You really need to see the history of prices to project their future, then you need to start seeing that for each application, adoption is non-linear once the tech becomes the cheapest option for that application.
What I'm most surprised about on HN is how few commenters here seem to understand tech curves. The HN audience, of all audiences, should have the clearest view of how traditional energy industry assumptions are about to be disrupted in the coming decades. We don't know the full extent of the disruption, but views like Seba's of huge amounts of surplus energy, seem inevitable. Arbitrage in time and geography will be the money making opportunities.
Working for one of the largest energy grid providers in Europe, I can tell you: This is the future we‘re currently building. We already have many areas with a huge energy surplus completely generated by renewables.
There is no one here that doubts the renewable future and we‘re quite sure we can finish building the necessary grid until the end of the decade (with around 6 times the capacity of the current European grid).
That is very encouraging to hear! In the US we have a few forward thinking utilities, thou go perhaps not as forward thinking as imagining a 6x capacity grid. But it seems that most are stuck focusing on the disadvantages of change as opposed to the immense opportunity this transition provides.
Wow, so much willful misinformation. Let’s cover a few!
- Steering: indeed the CT uses an innovative, triple-redundant steer by wire system. The steering wheel, which is a squared circle, not a yoke, turns 380 degrees lock to lock, so you never need to reposition your hands. It adjusts steering ratios dynamically according to speed and other factors.
- Crash safety: The 4 other Tesla models all had record-breaking safety scores, each the highest ever tested at the time of release on both European and U.S. standards. CT will undoubtedly continue and likely exceed that record. In any case, it would not be on the road if it had not passed U.S. safety standards.
- Pedestrian safety: This one is Bizarre. The CT front profile is about 2 feet lower than every other full size pickup sold in the U.S., yet not a peep heard about any of them being dangerous to pedestrians. “But no crumple zones!!” —- pedestrian safety is not primarily achieved through crumple zones, rather by the shape of the vehicle and how impact is likely to move someone who was hit. The CT is infinitely better on this metric.
- No crumple zones: See the crash test videos of the CT to see what an absurd assumption this is. The front deforms beautifully and the passenger compartment is fully intact. Same for the 30 MPH side impact test.
- Not mentioned: the first full-car 48 volt architecture in history, which paves the way for this much-needed innovation for the rest of the industry.
- Further extension of Tesla’s astonishing electrical architecture, where all functions in the car are performed by a few high-density PC boards running industrial PowerPC processors rather than having 100+ “ECUs” bought off the shelf. This architecture is what makes it possible to control (and fix and refine) nearly all aspects of the vehicle THROUGH SOFTWARE (and explains why no other car makers can do this).
- The CT pushes this forward by making nearly every electrical device a peer on a redundant gigabit ethernet bus (greatly enhanced CANBus). Each device then has a single power lead—-as short as possible from any convenient location—and activates itself by commands on the bus. 48 volts plus this reduced the weight of the wiring harness by 80% over a typical vehicle.
- Looks: I thought this site audience was made up of engineers or those who appreciate engineering. The look of the CT follows its function. Using castings, a structural battery pack and durable, paint-free stainless steel, it makes one of the toughest, most durable and most structurally rigid vehicles ever made. Yet, compared to an F150 Lightning, it has more interior space, a bigger bed while weighing significantly less.
- Economics: Ford in recent quarters is losing around $36,000 per EV they sell, despite the prices charged being quite high. CT is designed for manufacturing simplicity and architectural efficiency. This difference means that Ford is stuck at high prices and big losses while Tesla has ample price flexibility. As they have with their other vehicles, they can bring down prices as market conditions warrant and as materials and economies of scale make possible. No other EV maker has ever been profitable, and none have ever dropped prices as aggressively as Tesla has this last year or so. There is every reason to assume this will continue with the CT, once the first early-adopter wave subsides.
I get that Elon Musk infuriates many people (as he frequently does me). But it’s just dumb to disparage the achievements of the most talented and innovative team of automotive engineers since Henry Ford just because you don’t like the boss.
There’s a bunch of things wrong with your comment.
> Steering: indeed the CT uses an innovative, triple-redundant steer by wire system. The steering wheel, which is a squared circle, not a yoke, turns 380 degrees lock to lock, so you never need to reposition your hands. It adjusts steering ratios dynamically according to speed and other factors.
I’ll take most of what you’re saying here at face value because I haven’t looked into the steering that much. That said, can you elaborate on how one would turn the wheel 380 degrees without ever needing to reposition their hands?
Other than that, I’m not sure innovation for the sake of innovation is necessarily a good thing.
The blinker buttons on the wheel, placed above each other, already are causing issues in day to day driving[0]
> Crash safety: The 4 other Tesla models all had record-breaking safety scores, each the highest ever tested at the time of release on both European and U.S. standards. CT will undoubtedly continue and likely exceed that record. In any case, it would not be on the road if it had not passed U.S. safety standards.
This is a non-sequitur for a couple of reasons.
For starters in the US car manufacturers self-certify, there is no pre-approval before cars can be sold.
It’s one of the many reasons why the NHTSA and US regulations are a joke.
The CT hasn’t been rated by the NHTSA yet[1] and isn’t scheduled to be rated in 2024[2].
As for using prior achieved scores as an indicator for future scores; that’s silly in the best of times, but outright ridiculous when you’re talking about a model with a design that’s nothing like the other models you refer to.
> Pedestrian safety: This one is Bizarre. The CT front profile is about 2 feet lower than every other full size pickup sold in the U.S., yet not a peep heard about any of them being dangerous to pedestrians. “But no crumple zones!!” —- pedestrian safety is not primarily achieved through crumple zones, rather by the shape of the vehicle and how impact is likely to move someone who was hit. The CT is infinitely better on this metric.
Here you’re rolling multiple things into one.
For starters there are plenty of “peeps” about trucks and SUVs being dangerous for pedestrians. Both from “full time dissidents”[3], mainstream media[4] and studies[5] alike.
With regards to the CT, I hope I don’t have to waste time explaining how stainless steel v. pedestrians is more likely to cause injuries.
> No crumple zones: See the crash test videos of the CT to see what an absurd assumption this is. The front deforms beautifully and the passenger compartment is fully intact. Same for the 30 MPH side impact test.
The debate isn’t just crumple zones v. no crumple zones and keeping the cabin intact.
If that were the case then we shouldn’t even bother testing because almost all cars crumple to one degree or another.
The purpose of a crumple zone is to absorb as much of the forces as possible.
When I look at the Tesla provided test video I see a very short crumple zone with a lot of the forces ending up being absorbed by the dummies.
In particular the dummy in the back without airbags is being flung into the well in front of that seat (to the point I’m wondering if those seatbelts were malfunctioning, but let’s not get distracted).
This is far from “beautiful” as you describe it.
> Not mentioned: the first full-car 48 volt architecture in history, which paves the way for this much-needed innovation for the rest of the industry.
Cool. Normally those are things you focus on when everything else is up to snuff.
> Further extension of Tesla’s astonishing electrical architecture, where all functions in the car are performed by a few high-density PC boards running industrial PowerPC processors rather than having 100+ “ECUs” bought off the shelf. This architecture is what makes it possible to control (and fix and refine) nearly all aspects of the vehicle THROUGH SOFTWARE (and explains why no other car makers can do this).
All I’m hearing is more points of failure. I don’t think this is a matter of other manufacturers not being able to do, rather a matter of them not willing to do it because they understand the risks of failure better than Tesla does.
> Looks: I thought this site audience was made up of engineers or those who appreciate engineering. The look of the CT follows its function. Using castings, a structural battery pack and durable, paint-free stainless steel, it makes one of the toughest, most durable and most structurally rigid vehicles ever made. Yet, compared to an F150 Lightning, it has more interior space, a bigger bed while weighing significantly less.
Looks are in the eye of the beholder, so far most beholders seem to have a different eye than Tesla does.
> Economics: Ford in recent quarters is losing around $36,000 per EV they sell, despite the prices charged being quite high. CT is designed for manufacturing simplicity and architectural efficiency. This difference means that Ford is stuck at high prices and big losses while Tesla has ample price flexibility. As they have with their other vehicles, they can bring down prices as market conditions warrant and as materials and economies of scale make possible. No other EV maker has ever been profitable, and none have ever dropped prices as aggressively as Tesla has this last year or so. There is every reason to assume this will continue with the CT, once the first early-adopter wave subsides.
Respectfully, this just sounds like copium.
Even if Tesla would cut the price of the CT in line with prior price cuts, it will still be nowhere near the price he originally stated.
Unless somehow you see a big brain move behind announcing a low price and then launching it with a price that’s not even in the ballpark of the original promise, I can’t take this seriously.
I do generally agree with your point here that this isn't very revolutionary. However it's worth pointing out that Apple chips have lots of on-package memory but relatively little on-die and they are quite different from this IBM chip because of that.
I know basically zero about chip fabrication but I remember reading somewhere a while (15÷ years) ago that processor-in-memory was always a desirable design objective for obvious reasons, but that there are fundamental differences in the process for fabbing memory versus logic (different regions of Si doping not possible in same wafer, something like that? See this is where I should stay out of these discussions) that haven't been resolved, so the next best thing is on package pairing.
Roughly speaking, CPU wants smaller space and RAM wants bigger space. At a high-level take:
* CPU design is the most expensive space due to it having the greatest quality and capability requirements. RAM is mostly just a very, very large repetitive structure, so more space better.
* A CPU fault can be potentially corrected by microcode changes to route around the damage (ie part binning). RAM cannot generally take faults.
* DRAM is simpler to make, "just" a capacitor, but capacitance leaks over time; which, means generating resistive heat in an area that we want as little heat as possible. You could use SRAM (two transistors) but now you have substantially more complex part to fail.
* DRAM quality requirements are much less stringent if you make just bigger cells.
I think something missed in this discussion is the fact that an EV is so much simpler than an ICE vehicle* that costs will inevitably come down.
Many people believe that decent models will be considerably cheaper than today’s Tercels, Civics, Corollas, etc. in as little as 2-3 years.
Today you can buy a brand new Tesla Model 3 for $39,000 and with the U.S. tax credits that drops to $31,500. If you are in states like NY with additional credits, $26,500.
So it is already around the price levels of Accords, Camrys, etc. on the existing Tesla platform.
Tesla’s goal in their next platform is to cut the price to manufacture in HALF, and there is good reason to believe they will achieve that. The aspirational goals Toyota is claiming in this article, if achieved, will help them get closer to Tesla’s current architecture, which will be a great step forward.
As manufacturing becomes more efficient and prices drop, it will indeed make less and less sense to do major repairs, which are only getting more expensive.
* EVs are simpler when they have a proper hardware/software integration and over-the-air updates. Continuing the current industry practice of stringing together 150 off-the-shelf ECUs is one reason why most companies are losing money on their EVs and will never keep up with those who understand the software-defined car.
SpaceX is well underway building their second orbital launch platform and tower at Cape Canaveral in Florida.
My understanding is that Boca Chica will be a factory and development/testing site and Florida the primary launch location. They are also building a rocket factory at Cape Canaveral.
Nearly all of the breathless doom mongering has been from one or two small environmental groups who opposed the plant from the start and who keep inventing objections to attempt to support their foregone conclusion. They use every possible opportunity to slow down progress irrespective of the merits of their complaints, and judges have rejected most of them.
For example, water use. Tesla has built a complete onsite water reclamation plant that will allow it to build twice as many cars using the same water allocation they are already granted.
Also “they are killing the forests!”. What is not said is that this so-called forest is actual one of many vast tree farms located throughout Germany which are planted with quick-growing monocultures of identical trees and harvested regularly. Tesla is planting an identical size plot of new trees elsewhere, and that effort was 62% complete as of May 2021
Let’s please not have our future depend on building giant, fragile things that are obsolete technologically and financially before they are even completed.
Wind and PV have some challenges, but they get online fast. Plus they are both riding down a steep cost S-curve, so (for the foreseeable future) the more you build, the cheaper the next one gets.
No one could say that about Nuclear with a straight face…
Yes. Please tell me the alternative solar or wind solution which can provide us with a carbon free grid in 13 years. There is none. You have to hand wave and pretend that maybe the next 13 years of battery technology will be fundamentally different than the past 30 years since the invention of the Lithium ion battery. Even if we had the battery solution today we don't have the global manufacturing capacity.
Solar and wind go online fast and turn stable grids into pricy unstable ones as you increase their share of production. Meanwhile Nuclear is ready, today. And if you cut down on some of the regulatory burden it could be built sooner.
You are comparing apples to oranges. We don't have the global industrial capacity to deploy nuclear at scale either. And the intrinsic technology challenge is more difficult with nuclear.
One thing I was hoping to see—-remove MS Word induced junk like “Smart quotes” and “smart apostrophes” and whatever else tends to blow up web text when importing or displaying.
EVs are complex, but in ways that are much more subject to economies of scale.
For example, much of their value is in the software, which replicates for nearly zero marginal cost. Electric motor assembly is highly automatable and a fraction of the cost of assembling engines. There are no emission control systems, fuel systems, exhaust systems.
That leaves battery materials and construction, and these are dropping fast. Goldman Sachs just predicted pack-level prices will be below $100 KWh by 2025 and continue dropping by 11% per year throughout the decade.
In fact, BYD already sells a perfectly-reasonable EV in China for around $11,000 US equivalent.
There will, of course, always be premium priced EVs in the market, but I believe the lower bound of the market will be much lower than the Corollas, Tercels and Honda Fits of today.