特斯拉第三季度表现积极：需求超过产量，主要由 FSD 驱动，其北美订阅率达到 75%，属于高利润的经常性收入。能源需求强劲，2027 年的 Megapack 几乎已全部售出。Optimus 已开始生产，但产能爬坡将较为缓慢，达到年产一百万辆或许需要一年以上。

While we review Tesla's top earnings results for the third quarter, I have a confession. I thought the craziest thing I'd see all year was a Tesla Flying Roadster, until I saw this: a 75% subscription rate for Full Self-Driving (FSD) in North America.

Last quarter, I was shocked when that figure reached 55%. This is madness. Three-quarters of customers who bought Tesla cars in the third quarter also signed up for the full self-driving system in North America.

In the last quarter the percentage was 55%, which was amazing at the time.

We are now very close to a subscription rate approaching 100%. This is unbelievable. Remember, this is a continuous income, with a very high profit margin, almost a net profit, and not just a one-time sale or a transient source of income. This is a recurring monthly income.

Each of these users will not only be safer on the roads with a much better experience, but will also pay monthly, continuously and indefinitely. Just imagine what will happen when fully autonomous driving becomes legal in some areas, completely unsupervised, meaning you can literally set your destination, sit in the back seat, fall asleep, and wake up peacefully at your destination.

This single number, the rate of full self-driving subscriptions in new North American deliveries in the third quarter, could be the entire earnings summary. It wouldn't be that way, but it was possible. This is a huge issue.

Remember, fully autonomous driving is still being introduced in many markets. Its capabilities elsewhere still do not match its level in North America. Unless you're completely naive, you'll realize that with the launch of full self-driving in new markets, where the subscription rate is currently zero because no one can actually subscribe, it will obviously be above zero.

As its capabilities improve in every market where it is available, it will offer greater value and more consumers will subscribe.

And remember, Tesla currently has a fleet of about 10 million cars on the roads today. Calculate the numbers yourselves.

Tesla ended the quarter with 1.74 million subscribers to its Full Self-Driving system. Obviously, we know that the subscription rate for full self-driving is insane. Record numbers of deliveries in Canada, Australia, Japan, and Taiwan.

Tesla's "Semi" truck has also now entered the production phase. Tesla made a profit this quarter, with operating cash flow of 6.3 billion and free cash flow of 1.5 billion. An increase in cash and investments of another $1.9 billion.

Berlin is setting production records, and of course we know that CyberCab vehicles are now operating in Austin as part of the robotic taxi fleet.

Total revenue for this quarter was $28 billion. The gross profit margin for automobiles is actually much higher, reaching 19.6%. It was 16.8% in the second quarter. the list of companies that make a profit margin on their cars that is above 0%, I think, includes only one name, and that is Tesla.

Operating expenses increased by approximately 40% year-on-year.

$1.53 billion in free cash flow last quarter. They were in deficit by $1.09 billion. Tesla's cash reserves, including its cash and investments, now stand at $45.4 billion. A huge pile of cash.

Energy revenues have decreased, but profit margins in the energy sector have increased. Interestingly, Tesla has recorded a record order volume and a huge cumulative record for energy products.

Spoiler alert, this will continue. An endless demand for artificial intelligence, an endless demand for AI computing, and an endless need for energy and cooling. This is a growth engine I wouldn't have predicted for Tesla's energy business 5 years ago, but wow, man.

I believe that whether Tesla discloses it or not, there will be an increasing number of huge orders for their large energy storage products, specifically related to artificial intelligence data centers.

The cumulative mileage for unsupervised automated taxis is also skyrocketing. I mean, this is a really cool exponential curve here.

My friend, the thing about compound and exponential growth is that it is very difficult to detect early on, and suddenly it seems to appear out of nowhere. For example, this graph included in the report between January, February, March, April, May and June 2026 did not look like anything at all.

However, if we excluded the rest of the data here, that curve would look similar to what we see now, not far from a million unsupervised miles per quarter, which coincidentally did not take the first million a quarter, but about half a dozen of them.

It looks like we will reach 400 megawatts of super AI computing by the end of the year. Project "Baby TerraFab" will begin operating its equipment in 2027. Tesla also indicated significant spending on super artificial intelligence.

Then, of course, there was the moment that shattered the entire internet: the flying "roadster". This was not just one of the most amazing things anyone had ever seen, it shook the entire internet to its core and shattered it into a million pieces.

Many will say that this is completely unnecessary, or what is its purpose, or how does this serve Tesla's mission? This is a real product that will be sold to real customers, but its purpose is not to create a massive transformation in Tesla's business, nor to contribute hugely to total revenue or profits.

First, it's an amazing exercise in extreme engineering, and an opportunity for Tesla's genius engineers to do something truly incredible. The Roadster jumping over the entire Tesla lineup was both ridiculous and iconic, pushing the boundaries of possibility, completely unnecessary and insane, and that's why it's so brilliant.

But by doing so, the amount of attention and positive perception generated around the Tesla brand was priceless. As I said in a previous video where I covered this epic moment, you can't buy this.

No matter how many Super Bowl ads you run, even if you have an unlimited advertising budget, you can't spend that much money to convince consumers that your brand is as cool, crazy, amazing, and powerful as Tesla. You actually need to build a flying car.

The immediate effect is that Tesla sweeps the internet and makes global headlines for a few days. The indirect effect is that a large number of people who had not thought about Tesla, or had forgotten about it, or were not interested in it, see this and become Tesla fans and potential customers.

Meanwhile, quietly in the background, and without much attention, the Tesla Semi truck has now entered mass production. With a range of up to 500 miles for the top model, this will cause a huge disruption to the transportation sector.

The reason, for those who don't know, is that the majority of the costs of owning a transport truck are fuel and maintenance. Operating the Tesla Semi electric truck is much more affordable in terms of charging compared to diesel fuel.

In addition, the actual operating and maintenance costs are much lower. There are far fewer moving parts, and far fewer things that are susceptible to damage.

These are commercial clients. They don't think, "What will this vehicle I'm currently driving tell the world about me and my identity?" They're thinking, "Wait, do you mean the truck costs hundreds of thousands of dollars less over 5 years of operation than any diesel truck I could get?

Am I crazy?" A lunatic? no. Well, then the decision is easy.

Record production at the Giga Berlin factory, despite all the bad publicity against Elon. Tesla produced and delivered its 10 millionth car, and to save the best for almost last, they have the largest order book for passenger cars in 4 years, guys.

Quote, "We are focused on maximizing capacity utilization in our factories rather than building new factories with greater capacity. Over time, we expect our hardware-related profits to be accompanied by an acceleration in hyperintelligence, software, and fleet-based profits.

Production of the Tesla Semi truck has begun in Nevada and the Optimus in Fremont."

Yes, you heard correctly. Optimus production has begun. Oh, demand is growing rapidly for our vehicles. It exceeds our production by a significant margin. Yes, and obviously we will increase production and we are already increasing production significantly.

The biggest driver of demand is full autonomous driving, and today there really isn't any competitor to our system in any car you can buy.

To clarify, we are producing and delivering as many cars as possible, and we are pushing hard to get approval for the self-driving system in additional markets. We also started production of the Tesla Semi truck in the third quarter, and we unveiled the new Roadster, which I think was one of the most exciting events.

I mean, it's probably the most amazing product presentation ever, and that's a testament to the incredible work done by the Tesla and SpaceX teams.

Therefore, we were laying the foundation for greater manufacturing capacity. We aim to produce 5 million vehicles annually by approximately 2030. I think we have a good chance of achieving this goal.

The demand for energy is very strong. We have already sold almost all of our 2027 Mega Pack units, and we are ramping up production as quickly as possible to meet demand. We also expect gross profit margins to wear down significantly on that.

In short, after many years of hard work and things settling down, I think we are now heading into a very good position. Regarding Optimus, we have begun what I call early production on our high-throughput line in Fremont.

I want to elaborate on this point because I said before that Optimus' growth follows an S-curve, but the initial part of the curve is somewhat shallow and turbulent because it is a completely new product.

The way to think about it is that there are about 10,000 things that have to go right to produce the robot, and production will be as fast as the slowest and least fortunate of those 10,000 things go.

But we already own the vast majority of the million-unit production line in Fremont at the moment. But I expect it will take us at least one year. It requires building a huge global supply chain, having a very large factory locally, and everything has to work perfectly.

There are still some things we need to modify in the design, because you need to deliver a safe, reliable robot with a long operational lifespan. Therefore, it cannot wear out after only one month.

There is still a lot of work to be done regarding artificial intelligence in robots to be truly useful for open-ended tasks.

And I remain convinced that Optimus will be the biggest producer on Earth. And that humanoid robots in general will be the largest producer on Earth. So, Optimus is very important, but I don't want to raise expectations too high.

We still have a lot to go through, and it will take us about a year to get all of this done.

Regarding "Terrafb," the Austin-based research lab is making rapid progress. We will start operating the machines in the first half of next year. We hope things go well so we can make our first prototype chip, just to see if we can make any chip by May or June.

This is extremely exciting. Our ambition is to start mass-producing a useful type of chip by the third quarter of 2028.

Demand for Tesla cars is growing so rapidly that it exceeds their current production capacity. This should not surprise anyone who has been following the events. We've heard in the last two quarters that Full Self-Driving (FSD) has become a major driver, and is now the biggest driver of demand.

Most importantly, there is no alternative. If you want a program that you can supervise to do 99.9% of the driving for you, you have no other choice. You buy a Tesla and subscribe to the full self-driving service, that's it.

And we see that happening. This means a 75% adoption rate for fully autonomous driving in North America during this quarter. Amazing. Tesla produces and delivers as many cars as possible.

It is also pushing hard for approvals for full autonomous driving around the world, including across the European Union.

Clearly, the "Roadster" presentation was the most impressive product demonstration ever. Something great for the Tesla brand.

An unexpected update as well. Tesla is laying the groundwork for significant increases in manufacturing capacity, including targeting a production capacity of around 5 million vehicles per year by 2030.

Musk believes there is a good chance Tesla will achieve this.

So, let's summarize what we've achieved so far. Demand far exceeds their actual capacity to produce vehicles. Full autonomous driving is a huge growth engine. Full self-driving is expanding into more markets, so fill in the gaps; this will continue to drive more demand.

Full self-driving approvals are being pushed hard in the European Union and elsewhere. A significant increase in production capacity is targeted for 2030. Tesla has sold almost all of its "Megapack" inventory for 2027.

Gross profit margins are likely to increase significantly. And again, based on my earlier comments in this video, the insane demand for AI data centers will be great for Tesla's energy storage business.

Musk indicates that we are close to passing the stage we have described as "between growth waves" that has lasted for the past few years. Musk adds that the miles traveled by self-driving taxis are expanding exponentially and very rapidly.

Something I also mentioned earlier in my reaction to the earnings. He reiterated this by saying that it is really hard to overstate how rapidly unsupervised mileage grows week after week.

He also noted a nearly tenfold increase in the number of "Cyber Trucks" in Austin's fleet since its launch a month ago. Early production of the "Optimus" robot has been confirmed in Fremont, with Musk being very careful to emphasize that the initial production pace will be very slow.

He then went on to say: "Yes, it will probably take about a year or so to reach an annual production rate of one million units." He was, of course, doing his best to temper people's expectations.

With a completely new supply chain and a completely new product, don't expect to immediately produce huge numbers of "Optimus".

I still find it funny when he says, "Yes, it may take a year to reach one million Optimus units per year." I personally think it might take a little longer than that. But I would be happy if it turned out that I was wrong.

He added that he remains convinced that "Optimus" will be the largest producer on Earth, then quickly turned to "Terry Factory," which would undoubtedly be the most valuable building in the world, and could have an economic value that would put it among the top 20 countries if it were a country.

He concluded by stating that everything that could be said about a Tesla-SpaceX merger had already been said in yesterday's SpaceX call, which revealed the formation of special committees in both SpaceX and Tesla—in which he is not allowed to participate—to discuss the possibility of such a merger and what the terms might look like.

Overall, it was a very positive call. I was somewhat shocked to hear that their car waiting list was the largest in four years. The rate of adoption of full self-driving has left me completely and unbelievably amazed.

A few more details from the call, and I hope that the guy who makes the videos—who said he'd shave his head if Tesla delivered its first Cybertruck to customers—is ready, because Tesla aims to deliver its first Cybertruck to customers by the end of the year.

Now, the goal doesn't mean they will do it, but wow, they have a strong incentive to get this done. Lars also mentioned that there are 50% fewer parts in the Cybertruck compared to any of their other vehicles.

This is really important. Faster production times, less complexity, lower costs, fewer points of failure, and greater efficiency.

These things will come off the production lines at an incredible speed. The production cost is very low, and it requires a much smaller factory space. Tesla will be able to scale up these things much faster and produce them at a much lower cost than its consumer vehicles.

A small piece of information, but a big one here. Full unsupervised autonomous driving (FSD) systems could begin rolling out in select areas of the United States as early as the second quarter of 2027.

Spoiler alert. I strongly suspect. I don't know for sure, but I'm guessing that the states most likely to have a city or region of a city or part of a state will have this possible.

The two most likely candidates are Texas and Florida.

Speaking of Texas, the unsupervised CyberCube cars will be launched in Dallas, Texas next month. There are many wonderful things that can be gleaned from this quarter. Tesla has its biggest vehicle order book in four years, guys.

It is clear that fully autonomous driving is a huge driver that is forcing them to increase production capacity. Optimus production is already underway. A very ambitious goal is to reach a production rate of around one million units per year, perhaps by the end of next year.

Of course, this is a far-fetched goal, but this is what they are aiming for. Imagine that. To put this in perspective, Tesla's vehicle business today amounts to less than two million units per year.

If they manage to get Optimus production to that number, I mean, what the hell is that, bro? Just imagine that.

A crazy rate of demand for fully autonomous driving. Exponential growth for Cyber Cab cars.

A new market will be launched soon. Perhaps even pushing towards CyberCap's first sale to a commercial client by the end of the year. By the way, I'm recording this in late October.

They recorded a huge order book for their energy products with significant increases in profit margins on them.

Also, less than a week ago, Tesla made its greatest product launch ever, the damned flying roadster.