1,150 Yuan Profit Per Vehicle: How Long Can Tesla's AI Bet Last?

The most valuable info in this article is that Tesla made only 1,150 RMB per car sold in Q2 (calculated by non-GAAP net profit), whereas last year that number was nearly 4,000 RMB. Behind the halved profits isn't a sales collapse, but Tesla making a huge gamble to swap the future—turning today's profits into tomorrow's AI compute.

The financial report data itself isn't complex. Total revenue was $28.236 billion, up 26% YoY, beating market expectations; global deliveries were 480,100 units, setting a new Q2 record. But the profit side failed across the board: GAAP net profit was $1.494 billion, down 45% YoY; non-GAAP net profit was $1.812 billion, down 42% YoY. Profit per vehicle dropped from ~3,800 RMB last year to 1,150 RMB, a decline of nearly 70%.

Revenue grew, profits fell—the reason is simple: Tesla is burning cash like crazy. The report shows capex for the quarter was $2.573 billion, up 98% YoY, mainly used for AI infrastructure and capacity expansion. Specifically, this money went toward three directions:

• Dojo supercomputer cluster at the Texas Gigafactory
• Large-scale procurement of NVIDIA H100 GPUs (for training FSD and Optimus robots)
• 4680 battery line modifications and Cybertruck capacity ramp-up

These investments can't be monetized directly in the short term, but Musk repeatedly emphasized on the earnings call that AI is the engine for Tesla's 10x future growth. He's not wrong, but the problem is this bet requires continuous transfusion, while ICE price cuts and EV price wars are draining Tesla's ability to generate cash.

Automotive gross margin (excluding carbon credits) fell from 18.7% in the same period last year to 14.6%, the lowest value in five years. Price promotions are the direct cause, but the deeper logic is that Tesla is trading price for time—using discounts to expand sales volume to maintain factory utilization, while using scale to dilute FSD R&D costs. But economies of scale have a limit; when profit per vehicle drops below 1,000 RMB, Tesla enters the danger zone.

[!note]

Tesla's profit per vehicle fell 70% YoY in Q2 to 1,150 RMB. During the same period, BYD's profit per vehicle was ~900 RMB, but BYD has stronger cost control (vertical integration). Tesla's premium advantage is disappearing.

If you zoom out, you'll see Tesla's financial structure is undergoing a qualitative change. Over the past five years, Tesla's profit growth relied mainly on both delivery volumes and profit-per-vehicle rising. But from 2023 to now, profit-per-vehicle has declined continuously, and profit growth relies entirely on forcing up delivery numbers. Once sales growth slows, the inflection point will appear.

And signs of slowing sales growth are already here. Global deliveries in Q2 were 480,100 units; although a Q2 record, it was only up 0.6% QoQ and below the market expectation of 490,000. The effect of Model Y and Model 3 price cuts in China is diminishing, Europe is hit by tariffs and subsidy rollbacks, and North America sees slow Cybertruck ramp-up. Musk admitted on the call that 2024 delivery growth will be significantly lower than the 38% seen in 2023.

What's truly worth watching isn't the auto business itself, but that Tesla is using money earned from autos to fund an AI dream that may never become profitable. FSD V12.5 is currently pushed to only some users; while the end-to-end neural network is technically stunning, commercial rollout is far off. The Optimus robot is still stuck in lab demo stages. These projects devour billions annually without generating any revenue.

# Tesla 2024 Q2 Capex Breakdown (Estimate)
capex = 25.73  # Billion USD
ai_infra = 8.5  # Dojo + GPU procurement
battery = 6.0   # 4680 lines
cybertruck = 5.0
other = 6.23
print(f"AI Bet Share: {ai_infra / capex * 100:.1f}%")
# Output: 33.0%

One-third of capex is thrown at AI, yet the AI business currently contributes zero revenue. This isn't betting on tomorrow; it's betting on the next decade. But capital markets are running out of patience for long-term stories. After the report release, Tesla's after-hours stock dropped 8% because investors saw: selling cars is becoming less profitable, and new businesses show no return anytime soon.

Tesla's current situation resembles Tesla ten years ago—when Model S mass production ramp-up burned through company cash, leaving them on the brink of bankruptcy. History repeats, except the bet has shifted from EVs to AI. The difference is that back then the EV market was a blue ocean, while today the AI track is crowded with giants like Google, Microsoft, and Meta; Tesla doesn't hold an advantage in compute scale or algorithm accumulation.

Musk is betting that once FSD and Robotaxi commercialize, Tesla can transform from a hardware company to an AI service company like Apple, jumping profit margins from 15% to 40%. But the premise for this assumption is L4 autonomous driving being achieved in 2025, while the entire industry keeps pushing back the timeline for L4.

Tesla's financial report is like a mirror reflecting the ultimate dilemma of tech companies: when the core business matures, you must choose between the profit curve and the future curve. Tesla chose the latter, but the cost is that every car sold contributes only 1,150 RMB of fuel for that future.

Original link: https://www.tmtpost.com/8076744.html