Tesla EV Pricing Strategy

Tesla’s profit margins plunged last quarter after the EV maker made aggressive price cuts.

On an earnings call, CEO Elon Musk said he’d push ahead with the strategy.

Musk could sacrifice short-term profits to grow Tesla’s market share.

But analysts have warned that Tesla may have to sacrifice its short-term financials to boost its market share — and it might be too early to say whether the price war will help or backfire.

We knew Musk eventually would have to cut prices.

In Wednesday’s earnings call, Musk told investors that the company will put sales growth ahead of profit in a weak economy.

“We’ve taken a view that pushing for higher volumes and a larger fleet is the right choice here versus a lower volume and a higher margin,” he said.

The primary questions in my mind are: 1)Could the price cuts have waited for a later date?
2)Are the price cuts too severe?
Musk has proven to be right in the past. We’ll see if the “magic” continues.

The entry-level Model 3 now costs less than $40,000, down from $62,990 at the start of the year. The Model S and Model X are also 20% cheaper than they were at the beginning of 2023, even after Tesla raised US prices Thursday.

1 Like

I doubt that info is correct. Maybe the entry level Model 3 is now $40K. But the $63K number in the linked article was for the long range dual motor version.



Tesla is expanding production capacity to reach their goal of 20 million cars by 2030. High prices that keep cars in inventory are of no help, cash flow from sales is helpful. My general answer to your two questions is that price cuts should be inventory clearing but no lower. As far as I know Tesla has the lowest number of Days Sales of Inventory. That should be an indication that pricing is OK.

The Captain

1 Like

Thanks for the catch Mike.

January 23, 2023 article:

The 2023 Model 3 starts at $46,990 for the rear-wheel-drive base model. The all-wheel-drive Performance trim is priced from $62,990.

The Model 3 Standard Range RWD has been reduced from $41,990 to $39,990.


Tesla’s always maintained a pretty low level of inventory. I think you kind of have to give them the benefit of the doubt that these are, in fact, the prices they need in order to clear all of their expected production on a going-forward basis. They’ve shown they’re pretty skilled at that.

Musk also mentioned, a few times IIRC, that he views future revenues downstream from the initial sale (full self-driving and charging, I believe) as factoring into the decision. The bigger the fleet of cars, the broader the market for those services. In particular, he seems a big believer in the potential of their FSD product to have a big impact. Seems a bit unlikely to me, but he’s generally stressed the importance of autonomy to Tesla.


Thanks Captain for the quick answer. There you have it. Musk is on track to achieve his goal.


An example is from the most recent earnings call:

“We’re the only ones making cars that technically we could sell for zero profit for now, and then yield actually tremendous economics in the future through autonomy,” Musk said April 19. “I’m not sure how many of you will appreciate the profundity of what I’ve just said, but it is extremely significant.”

1 Like

Not that it’s necessarily comparable, but the most successful corporation in history, Apple, gets 20% of its revenue from services. The remainder comes from hardware production of the iPhone (50%), iPad (11%), wearables (8%), and Mac computers (10%). Miscellaneous includes other hardware, like Beats and payments for things like setting Google as default search, etc.

Musk’s brag seems unlikely to me, although there may be continuing revenue streams in things automakers already do like pre-installing satellite radio, on-star, and seat-warmers :wink: Self driving might be one, but that is far enough away that no responsible accountant would include it in future projections.

But when is he going to deliver on what was promised 4, 5 years ago?


This year, he said (though with a bit of a wry chuckle, so perhaps not).

Selling cars for zero margin, but making it up on the back end with services, would imply a pretty hefty price tag and a high take rate for autonomy (electric “fueling” can’t generate that much revenue). Hard to see that happening unless Tesla is the first and only to autonomous driving for a while…which seems pretty unlikely, given that you’ve already got a few companies operating limited robotaxi services today.

The great thing about Musk is he makes all these wild predictions…most of which never pan out. Yet, he still gives back to back interviews on Fox and everyone reports on the wild predictions he made.

I believe Elon has mentioned a few times that roughly 6 billion miles is the number needed for a true autonomous system. Taking what we know from transformer based LLM systems, massive scales of data truly do continuously improve performance beyond what most experts would have predicted even a couple of years ago. I’m not well-versed in this field, but I imagine there is an analogy between the neural nets learning the nuanced long-range language relationships as data scales compared to the image neural nets for self-driving learning nuances of edge cases in the physical world.

Having said that, maintaining production and sales at the expense of today’s margins will only continue to add to Tesla’s FSD dataset, which I think is critically important to FSD becoming a reality. In the most recent earnings deck, they showed a cumulative FSD miles plot that’s near exponential–I roughly hand-fitted an exp(0.19x) function to the data they showed. Extrapolating out that function would give over 500 million miles by the end of the year and around 6 billion by the end of 2024.

I know Elon has said FSD is coming for years, but he may become the broken clock that’s right in the next year or two assuming they can continue exponentially scaling their data.


Makes me question his IQ. The hardware is not good enough now in the older models for FSD. Why would that change if he updates the software? Has he no clue?

I get his IQ is higher than mine. But having a high IQ and getting things correct are two different and in Musk’s case expensive things.

At any rate the net profit margin diving is good news as usual for Tesla. It will always work out.

God bless what AI can do. Sarcasm…that is forcing AI to do anything at all.

Actually the fleet of passenger cars in the US drive over 3 trillion miles per year. Some basic math using smaller figures to be conservative. About .2% of our yearly driving can make this work?

The problem…does that assume the experiment is only done with completely FSD vehicles? Not much of anything currently is completely FSD.

Enter Mercedes!

Tesla should increase its focus on service and parts - so cars with body damage aren’t in the shop for months - to make even more people comfortable with buying the cars.

19 M3 SR+


Rant alert!

I’m not a scientist but I was an avid and curious reader. As a programmer I had to watch for the smallest detail such as the semicolon at the end of a line of code and the macro, what my client’s business was all about. Most important, I had fun doing it. Maybe the most interesting subject I discovered was the Science of Complexity and my favorite Complexity scientist, Stuart Kaufman, the author of At Home in the Universe: The Search for the Laws of Self-Organization and Complexity available at Amazon

Increasing entropy is one of the fundamental laws of the Universe, it means increasing disorder, increasing randomness. The only way for a closed system to become organized is to acquire energy like the Earth gets energy from the Sun. After a few billion years random molecules get organized into life and humans. Atoms like to get together provided there is enough energy to get them to do it. Stuart Kaufman calls it “Order for free”

Kauffman contends that complexity itself triggers self-organization, or what he calls “order for free,” that if enough different molecules pass a certain threshold of complexity, they begin to self-organize into a new entity–a living cell. Kauffman uses the analogy of a thousand buttons on a rug–join two buttons randomly with thread, then another two, and so on. At first, you have isolated pairs; later, small clusters; but suddenly at around the 500th repetition, a remarkable transformation occurs–much like the phase transition when water abruptly turns to ice–and the buttons link up in one giant network.

At Home in the Universe - The Search for the Laws of Self-Organization and Complexity - NASA/ADS.

The brain is no exception. In my most unscientific way I came to the idea (long before I discovered Complexity ) that the way the brain works is by pattern matching. It stores huge amounts of patterns, sounds, images, memories, faces, voices, smells, and it has the ability to match incoming data with stored data. That’s how one recognizes a face among millions almost instantly. The deeper the memories (more data) the faster the recognition. I don’t have a clue of how it’s done but it probably is not much different from how ants and bees organize themselves. Anything that does not work dies and we get evolution of what works.

Punch Line

That’s probably how neural networks and machine learning works. When I started as a programmer not all was code, some computing machines were programmed using plugboards


Most everything that can be done by code can be done by hardware. Back when Alan Turing was building the machine to break the Enigma code they had discussions about using vacuum tubes (valves in British) – now replaced by transistors – (my first computer had 2000 triodes). Vacuum tubes won out.

tri·ode | ˈtrīōd |
a vacuum tube having three electrodes.
• a semiconductor rectifier having three connections.

The Dictionary


It’s the data, Stupid!

The Captain


OMG - My manager at P&G had a screensaver on his laptop that scrolled that message back in 2002. Thanks for the memory!!



The world’s car industry will shrink to only 10 companies over the coming decade, a Chinese rival to Elon Musk’s Tesla has said, as intense competition in China’s electric vehicle market spills on to the global stage. Brian Gu, vice-chair of Guangzhou-headquartered Xpeng, said for Chinese companies to be among the last carmakers standing, they would need to have annual sales of at least 3mn vehicles, underpinned by global exports. The world’s largest carmaker Toyota sold 10.5mn cars in 2022, while Tesla sold 1.3mn.

The warning comes at a historic juncture for the global car industry. China is on the cusp of overtaking Japan as the world’s biggest exporter of cars by volume after passing Germany last year. At the same time, slowing growth and an intense price war is pushing low-cost carmakers to the brink of collapse in China, the world’s biggest car market.

Xpeng is ranked 12th by sales amongst Chinese EV maker in the first quarter of 2023.
Xpeng is backed by AliBaba and has heavily invested in autonomous driving and is expanding EV sales in Europe.


I don’t work on neural nets, but do build production ML systems and it’s true for my “traditional” models and seemingly even more so for massive transformer neutral nets.

Also, this article is from when GPT3 was launched back in 2020, but is still worth a read: https://gwern.net/scaling-hypothesis.

It is the data, but this means more than just quantity. (i.e. making a target of so many billions of miles driven is a silly target). Data quality is important, such as variety of data. Example: most Tesla’s are driven by reasonably affluent people in certain parts of the country. There is likely little training data provided by those cars for driving in the snow, gravel roads, etc. This is why some companies rely on synthetic training data to add to the “real world” data from car sensors.

Still skeptical of Tesla FSD. Will believe when it actually ships. Not a day before.

1 Like