Tesla’s “Deadly Dilemma”: Does Underwhelming Car Production Still Matter Amid the Boundless Promise of AI?

$Tesla(TSLA.US) released its Q2 2025 earnings report after the U.S. stock market closed on July 23, Beijing time. In summary, the Q2 performance was decent, but Tesla is a stock priced for the future, especially at the current relatively high price of 330. Here are the core details:

1. Total revenue performed well: This quarter's total revenue was $22.5 billion, exceeding market expectations by about $200-300 million. The energy storage business showed flat performance due to tariff impacts, which was already priced in, while the service business continued to rise quarter-on-quarter, performing well. However, the real surprise was in the core automotive business, which finally began to recover!

2. Car sales revenue and gross margin both exceeded market expectations, emerging from the Q1 slump: Q2 car sales were $16.7 billion, finally exceeding expectations (market expectations were around $16.0-16.2 billion). Carbon credit revenue was below expectations due to policy factors, which was anticipated.

In the core automotive revenue (excluding carbon credits), this quarter was $15.8 billion, exceeding market expectations of around $15.0-15.2 billion, mainly driven by the price increase from the launch of the new Model Y Juniper.

3. Car sales gross margin also began to rise quarter-on-quarter due to the price increase of Juniper and improved economies of scale: Q2 automotive sales gross margin (excluding carbon credits and leasing) was 15%, improving by 2.5 percentage points quarter-on-quarter, mainly due to the higher pricing of the Model Y Juniper launched in Q2, leading to higher car sales prices. Meanwhile, the amortized cost this quarter, due to improved economies of scale, offset the adverse effects of tariffs.

4. R&D expenses and capital expenditures are still heavily invested in AI's vast business: This quarter's R&D expenses were $1.59 billion, exceeding market expectations by $200 million, but due to the quarter-on-quarter improvement in gross margin (fundamentals improved in Q2), operating profit and profit margin both rose quarter-on-quarter.

Free cash flow this quarter saw an increase in operating cash flow due to improved fundamentals, with capital expenditures at $2.4 billion, continuing to rise by $900 million quarter-on-quarter! Still heavily investing in AI business, Tesla added 16,000 H200 GPUs in Texas, so free cash flow still fell by $500 million quarter-on-quarter to just $150 million this quarter.

Dolphin Research's overall view:

The Q2 performance was decent, but since the Q1 report, the stock price reaching 330 is largely due to the normal progress of the Robotaxi business and pricing in most of the vast business of Robotaxi and Optimus, somewhat decoupled from the performance of the automotive fundamentals.

Given the already relatively high stock price, the market will be relatively harsh, so compared to the Q2 performance itself, it is more necessary to look from a future perspective:

1. Car sales fundamentals are still under pressure:

a. The U.S. IRA $7,500 subsidy phase-out is the biggest negative factor: One Big Beautiful Bill Act's $7,500 IRA subsidy will end on September 30. Although U.S. demand in Q3 may be released in advance to some extent, the decline in sales in the U.S. region after the subsidy phase-out in Q4 is inevitable, so it is extremely dependent on the mass production and launch progress of the cheap Model 2.5.

b. Model 2.5 mass production is delayed again to Q4: For Tesla car sales, the new cheap model "Model 2.5" is still the most critical variable, and its production and delivery progress is crucial to Tesla's car sales fundamentals this year. Although repeatedly rumored to be canceled, this earnings call confirmed that this model will continue to be launched, but the mass production progress is delayed by about a quarter, launching in Q4.

In this conference call, Tesla mentioned that due to prioritizing deliveries before the IRA subsidy expires (prioritizing the North American market to maximize production of existing models before the IRA policy expires), Model 2.5 mass production is again postponed to the next quarter, launching in Q4.

This delay in the mass production and launch plan of Model 2.5 may still lead to a further downward revision of the overall 2025 car sales expectations (currently market expectations are 1.6-1.65 million units), and the specific mass production ramp-up progress of this model still needs to be closely monitored.
(Note: The market previously expected Model 2.5 to be launched first in the U.S., in Europe in August, and in China at the end of 2025, making it difficult to achieve large-scale delivery before Q4. This time, Model 2.5 delivery is delayed again).

In terms of gross margin, regulatory credit adjustments will cause regulatory credits to continue to decline in subsequent quarters, and the impact of tariff costs on car sales will also be fully reflected in subsequent quarters, which may lead to continued pressure on gross margin in the short term.

Therefore, it can be seen that Tesla's car sales fundamentals are expected to remain under pressure until the official launch of Model 2.5 in Q4, and the energy storage business is also affected by tariffs, which will be fully reflected in subsequent quarters.

2. In the vast business of Robotaxi and Optimus, the current progress is basically consistent with the Q1 plan, and the arrival of this technology and progress inflection point is expected at the end of Q4 and early 2026:

① In the Robotaxi/FSD business:
a. In Robotaxi: Continuing as planned, no incremental information
Currently still proceeding as planned, officially launched in Texas, USA in June, with the first batch of 20 unmodified Model Ys put into small-scale trial operation.
Tesla's previous plan was to expand to multiple cities in the U.S. by the end of the year (generalized end-to-end algorithm supports rapid expansion to other cities), aiming to reach 1,000 Robotaxis within a few months, while expanding geographically to San Francisco, Los Angeles, and San Antonio in the U.S., and according to Tesla's previous plan,
it is expected to start contributing significantly to the financial statements in the second half of next year.
This earnings call's plan for this part of the business remains unchanged, with Tesla advancing regulatory approvals in multiple locations such as San Francisco and Nevada, aiming to cover about half of the U.S. population by the end of the year, with service areas and models showing exponential growth.

b. In the FSD business: The time inflection point is expected to achieve another major breakthrough in Q4 this year and early next year
Currently, Tesla's FSD (intelligent driving for passenger cars) is still based on the V13 software version and HW4.0 hardware, and the market is still eagerly anticipating Tesla's next-generation intelligent driving version's major update, which is another major leap in the intelligent driving end (referencing Tesla's V12/V13 version directly switching to end-to-end routes, bringing a significant performance leap):
The market is still anticipating the FSD 14.0 version on the software side and the HW5.0 version on the hardware side, and the HW 5.0 (latest called AI 5) version's hardware computing power is expected to reach over 2000+ TOPS, and compared to the existing model, the model parameters supported by the FSD 14.0 version will reach several times the current level, and its performance is expected to leap to another level, thereby driving the penetration rate of FSD and the continued expansion of Robotaxi.
In terms of progress, according to research, the current plan is to start internal testing of the V14 alpha version in Q4 2025, and officially launch the V14 official version in Q1 2026, and Musk also stated at the 2025 shareholder meeting that "HW 5.0 + V14 is the last piece of the puzzle to L5, and Tesla owners can sleep in the car in 2026."

② In the Optimus business: Gen 3 is expected to start mass production for the first time at the end of 2025 and early 2026
According to Tesla's plan in the Q1 conference call, by the end of 2025, thousands of robots will be put into trial use in Tesla factories, with confidence to achieve an annual production of 1 million units within 4-5 years, expected to reach this goal in 2029 or 2030.
This earnings call mentioned that the Gen 3.0 version design is complete, with a prototype to be launched within 3 months, mass production next year, and a target of achieving an annual production of 1 million units in 4 years, with progress basically consistent with previous plans.
According to research, Optimus is expected to launch the Gen-3 complete function at AI DAY in October 2025, with the Gen-3 version achieving significant performance improvements and continued cost reduction (cost reduction expected to reach $16,500), and in terms of production planning, small-scale trial production is expected in Q1 2026, and large-scale mass production is expected to start in Q3 2026, so this mass production and release inflection point may also arrive in Q4.

Therefore, Dolphin Research's view on Tesla's stock price is that the current price of $330 is still in an overvalued state, especially due to the delay in Model 2.5 mass production and the continued phase-out of the U.S. IRA $7,500 subsidy, the car sales fundamentals are likely to remain under pressure before the launch of Model 2.5.

However, in the vast business of FSD's next major iteration update and the imminent launch and mass production of Optimus Gen 3, this mass production and technology inflection point is very close (around the end of Q4 and early 2026), the story and logic are still very strong, belonging to a short-term unverifiable but long-term grand narrative, so if the stock price can adjust due to expectations of car sales fundamentals (gradually increasing positions below 250-270), it is instead a very good entry opportunity.

Below is a detailed analysis of the earnings report:

I. Tesla: Finally delivered a decent report card

1.1 Automotive Revenue: The launch and price increase of the new Model Y Juniper helped the automotive business climb out of the slump

This quarter's total revenue was $22.5 billion, performing well, exceeding Dolphin Research's expectation of around $22.3 billion from major institutions. Although the energy storage business performed generally, the tariff impact was largely priced in (BBG's expectations are relatively outdated), and the real highlight was in car sales revenue, finally exceeding market expectations after several quarters of misses, due to the price increase from the launch of the new Model Y Juniper.

Specifically:

In the automotive business, this quarter's total revenue was $16.7 billion, finally exceeding expectations (market expectations were around $16.0-16.2 billion), and carbon credit revenue was below expectations due to policy factors, which was anticipated (due to the withdrawal of CARB state's exemption in May, and Tesla's 50% regulatory credit revenue comes from CARB state's ZEV credits).

But in the most concerned market of carbon credit and leasing revenue, this quarter's real car sales were $15.8 billion, finally climbing out of the Q1 car sales slump, also exceeding market expectations of around $15.0-15.2 billion, mainly driven by the price increase from the launch of the new Model Y Juniper.

In the energy business, this quarter's energy business revenue was $2.8 billion, below expectations, still mainly dragged down by tariffs, but this impact was largely priced in, and in the service business, this quarter performed well, with Q2 service revenue of $3.05 billion, rising $400 million quarter-on-quarter, mainly due to the continued expansion of supercharging stations bringing incremental revenue.

1.2 Car sales gross margin finally begins to recover!

Every earnings report, more important than revenue, and the real incremental information during the earnings report is always the performance of automotive gross margin.

Automotive gross margin finally climbed out of the worst car sales period last quarter, with overall automotive gross margin rising from 16.2% last quarter to 17.2% this quarter, and the most core carbon credit-free car sales gross margin was in a state of exceeding expectations (actual 15.5% vs. market expectation 13.3%)!

In other businesses, the energy storage business gross margin was in a state of exceeding expectations, because Tesla's energy business mainly uses LFP batteries from China, which are significantly affected by tariffs, but the actual gross margin performance still exceeded expectations, possibly due to Tesla's own stockpiling of LFP batteries, coupled with an increase in the proportion of Megapack shipments, but this business will still be dragged down by tariff uncertainties in the future, so Tesla also stated that it will start producing LFP batteries domestically in the U.S. this year to resist tariff impacts.

In the service business, this quarter's gross margin was 5.4%, rising quarter-on-quarter, also performing well, mainly due to the continued expansion of Tesla's supercharging network driving gross margin upward, offsetting the negative drag of the used car business.

II. Car sales fundamentals began to improve in Q2

As the most important observation indicator every quarter, automotive gross margin is of utmost importance, especially in the current situation where Tesla's existing models are aging and competition is intensifying. To clearly see the real situation of automotive gross margin, Dolphin Research separately broke down the automotive sales gross margin excluding carbon credits, automotive leasing gross margin, and overall automotive business gross margin.

Since the automotive leasing business is small in scale and has stable gross margin, the overall automotive gross margin is a combination of the two, and breaking it down so carefully is mainly to observe the automotive sales gross margin excluding carbon credits.

Q2 automotive sales gross margin (excluding carbon credits and leasing) was 15%, exceeding market expectations of 13.3%, finally emerging from the worst car sales trough in Q1, improving by 2.5 percentage points quarter-on-quarter. Dolphin Research believes this performance is mainly due to the higher pricing of the Model Y Juniper launched in Q2, leading to higher car sales prices, and this quarter did not have the negative impact of production halts and ramp-ups on fixed amortized costs in Q1, but rather had the release of economies of scale brought by quarter-on-quarter improvement in sales.

Let's break it down from the perspective of single-car economics:

2.2 Car sales price finally begins to rise after the Model Y Juniper update!

From the perspective of car sales price, in Q2, Tesla's revenue per car sold (excluding carbon credits and automotive leasing sales) was $41,800, rising by $1,800 quarter-on-quarter, while the market expected car sales price to show a quarter-on-quarter decline. The core reason for the rise is due to the price increase after the Model Y Juniper update, offsetting the negative impact of financing promotions and other incentives, and the increase in the proportion of Model Y in the sales structure this quarter, with a decrease in the proportion of lower-priced sales in the Chinese region, leading to a relatively strong performance in car sales fundamentals this quarter.

① Price adjustments for existing models: The starting price of the Model Y Juniper is higher than the old Model Y model, positively impacting car sales price

a. In the U.S.: Model Y Juniper price increase

The starting price of the Model Y Juniper in the U.S. is $59,900, higher than the starting price of the old Model Y, while also raising the prices of Model S/X.

Although Tesla offered a 1.99% financing incentive for the Model Y in Q2 and a $2,000 trade-in subsidy for old Model Y owners, the price increase offset the negative impact of the financing activity.

b. In China: In China, the starting price of the Model Y Juniper also increased by about 3% compared to the old Model Y.

In terms of loan policies, a 5-year interest-free loan was launched for the Model Y just after its launch, while the Model 3 series launched a limited-time insurance subsidy of 8,000 yuan, with the rear-wheel drive/long-range version adding a 5-year interest-free policy.

c. In Europe: In Q2, Tesla had both price increases and decreases for different versions of the Model Y in Europe;

② In terms of model structure: Relatively favorable, as the proportion of Model Y sales increased in Q2, while the proportion of U.S. models is rising

a. In the model structure: Due to the launch of the Model Y Juniper, the proportion of Model Y in this quarter increased quarter-on-quarter (the Model Y Juniper was halted in Q1, with no such factor affecting Q2).

b. In terms of regional proportion: This quarter, the proportion of higher-priced U.S. and other regions increased, also raising car sales prices.

③ FSD business: FSD has not yet iterated to the next generation, and no significant catalyst for large excess confirmation has yet appeared. The substantial growth in this part of the revenue still requires a significant iteration of Tesla's FSD technology and its implementation in Europe and China.

2.3 Car sales cost increase mainly due to tariff impact

After discussing the single-car price, let's talk about the single-car cost. Generally speaking, Tesla's cost reduction comes from four dimensions—1) scale dilution and full utilization of production capacity from sales release; 2) technological cost reduction; 3) natural cost reduction of battery raw materials; 4) government subsidies. Specifically:

Dolphin Research broke down the single-car cost into single-car depreciation and single-car variable cost, and the single-car economic account for Q2 is as follows:

1) Single-car depreciation effect: Single-car amortized cost is declining quarter-on-quarter

This quarter's single-car depreciation was $3,700, with an absolute value decrease of $600 quarter-on-quarter, and the single-car amortized cost rate also decreased by 2 percentage points quarter-on-quarter from 10.7% last quarter to 9% this quarter. Dolphin Research believes the main reasons for the decline in single-car depreciation this quarter are:

a) In terms of production cost: There is no longer the significant profit margin resistance brought by idle capacity and other ramp-up related costs in Q1 (the resistance of the new Model Y halt and ramp-up).

b) Improvement in economies of scale quarter-on-quarter: Tesla's car sales in Q2 were 384,000 units, an increase of about 14% quarter-on-quarter from 337,000 units in Q1, emerging from the Q1 car sales trough, and economies of scale are also improving quarter-on-quarter.

2) Single-car variable cost: Affected by tariffs, rising

This quarter's single-car variable cost was $32,200, rising by $1,000 quarter-on-quarter, still mainly affected by tariffs (Q2 tariff costs increased by $300 million, with $200 million reflected in this quarter's car sales variable cost, equivalent to a negative drag of $520 on car sales cost)

3) Automotive gross margin begins to recover due to the rise in car sales price and the decline in single-car depreciation cost:

Finally, due to the decline in single-car amortized cost this quarter, but the rise in single-car variable cost, the final car sales gross margin this quarter was 15% (excluding carbon credits), finally emerging from the Q1 car sales trough, improving by 2.5 percentage points quarter-on-quarter!

III. How will Tesla's car sales fundamentals evolve in 2025?

3.1 Q2 delivery volume exceeded market expectations, but has already been priced into the stock price

Tesla's actual delivery in Q2 was 384,000 units, exceeding the sell-side expectation of 380,000 units and the buy-side expectation of 360,000 units seen by Dolphin Research, mainly due to the U.S. market and other regions' sales exceeding expectations. By region:

By region:

① In the Chinese market: Tesla's retail sales and market share in the Chinese region both declined quarter-on-quarter in Q2. Due to the intense competition in the new energy vehicle market, even though Tesla just launched interest-free promotions for the new Model Y, compared to the speed of new car launches by Chinese peers, the competitiveness is clearly insufficient, and the market share is also continuously declining.

② In the U.S. market: Tesla's sales and market share in the U.S. region both rose quarter-on-quarter in Q2, as the Beautiful Act has already been confirmed, and the federal tax credit of $7,500 for new energy vehicles will end on September 30, 2025, officially pausing from October 1, 2025, so the U.S. market saw pre-consumption before the IRA expiration.

③ In the European market: Tesla's performance in the European market was also relatively average, previously affected by Musk's political activities causing brand value damage, which was particularly significant in Europe, as corporate fleet buyers account for a high proportion of sales in Europe.

④ In other markets: Other markets significantly exceeded market expectations this quarter, mainly due to the joint efforts of the Asian markets (Malaysia/Korea/Thailand).

3.2 Short-term car sales fundamentals still face headwinds, Model 2.5 mass production and launch delayed

For Tesla car sales, the new cheap model "Model 2.5" is still the most critical variable, and its production and delivery progress is crucial to Tesla's car sales fundamentals this year. Although repeatedly rumored to be canceled, this earnings call confirmed that this model will continue to be launched, but the mass production progress is delayed by about a quarter, launching in Q4.

Currently, market expectations for Tesla's sales are at 1.65 million units (before the Q1 earnings report, market expectations were at 1.8 million units), and buy-side expectations may have already fallen to 1.6 million units or even below (a decrease of 200,000 units compared to 1.79 million units in 2024), mainly due to the average performance of Tesla's existing models in the first half of the year, and the main sales increment this year is actually mainly in this Model 2.5 model, with the Model Y six-seat version expected to contribute relatively limited increment.

In terms of regions, One Big Beautiful Bill Act's $7,500 IRA subsidy will end on September 30, although U.S. demand in Q3 may be released in advance to some extent, the decline in sales in the U.S. region after the subsidy phase-out in Q4 is inevitable, so it is extremely dependent on the mass production and launch progress of the cheap Model 2.5.

This delay in the mass production and launch plan of Model 2.5 may still lead to a further downward revision of the overall 2025 car sales expectations, and the specific mass production ramp-up progress of this model still needs to be closely monitored.

(Note: The market previously expected Model 2.5 to be launched first in the U.S., in Europe in August, and in China at the end of 2025, making it difficult to achieve large-scale delivery before Q4. This time, Model 2.5 delivery is delayed again).

IV. But more important than car sales fundamentals is the vastness of Robotaxi/FSD and Optimus

For Tesla, the current stock price trend ($330) is already relatively decoupled from the car sales fundamentals, and investors' real bet is on Tesla's vast business: Robotaxi/FSD and Optimus, so the substantial progress of these two businesses, the next generation's progress, and Musk's planning implementation time are particularly important, and these two narratives may have significant breakthroughs at the end of this year and early next year, with short-term narrative logic unverifiable:

① In the Robotaxi/FSD business:

a. In Robotaxi:

Currently still proceeding as planned, officially launched in Texas, USA in June, with the first batch of 20 unmodified Model Ys put into small-scale trial operation.

Although there are still many problems in the actual road test performance, Dolphin Research still believes that this is essentially an inherent flaw in the end-to-end route, but in terms of running through the business model (Tesla's current end-to-end route has lower overall costs and strong generalization of the technical route, which can be quickly expanded to other cities and regions), it is still much more likely than Waymo's route.

Tesla's previous plan was to expand to multiple cities in the U.S. by the end of the year (generalized end-to-end algorithm supports rapid expansion to other cities), aiming to reach 1,000 Robotaxis within a few months, while expanding geographically to San Francisco, Los Angeles, and San Antonio in the U.S., and according to Tesla's previous plan,

it is expected to start contributing significantly to the financial statements in the second half of next year.

This earnings call's plan for this part of the business remains unchanged, with Tesla advancing regulatory approvals in multiple locations such as San Francisco and Nevada, aiming to cover about half of the U.S. population by the end of the year, with service areas and models showing exponential growth.

b. In the FSD business: The time inflection point is expected to achieve the next major breakthrough in Q4 this year and early next year

Currently, Tesla's FSD (intelligent driving for passenger cars) is still based on the V13 software version and HW4.0 hardware, and the market is still eagerly anticipating Tesla's next-generation intelligent driving version's major update, which is another major leap in the intelligent driving end (referencing Tesla's V12/V13 version directly switching to end-to-end routes, bringing a significant performance leap):

The market is still anticipating the FSD 14.0 version on the software side and the HW5.0 version on the hardware side, and the HW 5.0 (latest called AI 5) version's hardware computing power is expected to reach over 2000+ TOPS, and compared to the existing model, the model parameters supported by the FSD 14.0 version will reach several times the current level, and its performance is expected to leap to another level, thereby driving the penetration rate of FSD and the continued expansion of Robotaxi.

In terms of progress, according to research, the current plan is to start internal testing of the V14 alpha version in Q4 2025, and officially launch the V14 official version in Q1 2026, and Musk also stated at the 2025 shareholder meeting that "HW 5.0 + V14 is the last piece of the puzzle to L5, and Tesla owners can sleep in the car in 2026."

② In the Optimus business: Gen 3 is expected to start mass production for the first time at the end of 2025 and early 2026

According to Tesla's plan in the Q1 conference call, by the end of 2025, thousands of robots will be put into trial use in Tesla factories, with confidence to achieve an annual production of 1 million units within 4-5 years, expected to reach this goal in 2029 or 2030.

This earnings call mentioned that the Gen 3.0 version design is complete, with a prototype to be launched within 3 months, mass production next year, and a target of achieving an annual production of 1 million units in 4 years, with progress basically consistent with previous plans.

According to research, Optimus is expected to launch the Gen-3 complete function at AI DAY in October 2025, with the Gen-3 version achieving significant performance improvements and continued cost reduction (cost reduction expected to reach $16,500), and in terms of production planning, small-scale trial production is expected in Q1 2026, and large-scale mass production is expected to start in Q3 2026, so this mass production and release inflection point may also arrive in Q4.

III. On the expenditure side: Still heavily investing in AI's vast business

Tesla's R&D and sales expenses this quarter continued to increase, with R&D expenses at $1.59 billion, exceeding market expectations of $1.34 billion, still due to increased investment in AI intelligence and new vehicle series development, while sales and administrative expenses this quarter were $1.37 billion, exceeding expectations of $1.27 billion, mainly due to marketing investment for the launch of the Model Y Juniper.

Finally, due to the exceeding expectations on the gross margin side, offsetting the rise in operating expenses, the overall operating profit this quarter was $900 million, rising by $200 million quarter-on-quarter, with an operating profit margin of 4.1%, also rising by 2 percentage points quarter-on-quarter.

In terms of free cash flow, this quarter's operating cash flow saw an increase due to improved fundamentals, but capital expenditures are still heavily investing in AI business, with capital expenditures at $2.4 billion this quarter, continuing to rise by $900 million quarter-on-quarter!

Tesla added 16,000 H200 GPUs in Texas, bringing the Cortex training cluster to an equivalent computing power level of 67,000 H100s, still heavily investing and paying for the vast business!

Finally, although this quarter's operating cash flow rose by $400 million quarter-on-quarter, due to the rise in capital expenditures by $900 million quarter-on-quarter, the final free cash flow still fell by $500 million quarter-on-quarter to $150 million.