XPeng's Double Miss: Spending to Grow, Not Burning Out

XPeng reported a wider-than-expected loss last week. Revenue fell short of some analyst forecasts. The stock had already lost roughly 40% of its value in 2026.

The obvious reading is that the company is struggling to grow its way out of losses. The only problem is that the math says something different.

XPeng's Q2 revenue came in at ¥19.74 billion ($2.91 billion), up 51% from the previous quarter. The loss per share was ¥1.29 on a non-GAAP basis, wider than the consensus expectation of ¥0.98. Revenue landed near the lower end of company guidance and below some Wall Street forecasts around ¥20.9 billion.

That's the double miss. But what actually widened the loss — and whether it matters — is a different question entirely.

Growing pains, not broken math

The net loss jumped to ¥1.34 billion from ¥480 million a year ago. Three expense lines explain the gap.

Research and development surged 32% year-over-year to ¥2.91 billion. Sales and general administrative costs rose 15% to ¥2.50 billion. And government subsidies — a non-recurring revenue source in the ¥1 billion range a year ago — declined. Strip those three items out and the operating picture doesn't look like a company losing its footing. It looks like a company spending to launch.

Vehicle gross margin held at 12.1%, flat from Q1 and below the 14.3% a year ago. That's the product-generation transition in action: older, higher-margin models phasing out while new ones ramp. The overall gross margin, though, improved to 20.7% from 17.3% a year ago. The lift came from services and technology revenue — up 94% year-over-year to ¥2.70 billion, with a 75% margin — largely from its engineering partnership with Volkswagen.

When you're spending ¥2.91 billion on R&D in a single quarter to develop AI-driven autonomous systems, new vehicle platforms, and a robotics division, the net loss widens. That's not a margin problem. That's an investment decision. The question is whether the spending is buying future capacity or burning cash with nothing to show for it.

The delivery ramp tells the real story

XPeng delivered 103,295 vehicles in Q2 — essentially flat year-over-year but up 65% from Q1. The sequential jump came from a June that delivered 40,126 vehicles, a 2026 monthly record.

The company guided Q3 deliveries to 115,000–121,000 units, a 11–17% sequential increase. July brought 38,027 deliveries, which means XPeng needs roughly 38,500–41,500 per month over August and September to hit the top of its range. That's right in line with recent performance.

The constraint on the upper end isn't demand — it's the ramp of the Mona L03, a new compact SUV that started deliveries in late July. The model launched July 16 at ¥123,800 and pulled in over 46,000 orders in its first hour, including more than 40,000 pure electric and over 10,000 extended-range versions. Supply bottlenecks pushed wait times to 13–17 weeks for the BEV. Deutsche Bank expects Mona L03 monthly deliveries to reach about 15,000 units this year, with full-year volume potentially hitting 150,000 in 2027.

That's a material new revenue stream coming online in Q3 and Q4. XPeng also teased the G9L, a larger SUV, without pricing or timing.

The cash position

XPeng held ¥40.48 billion ($5.97 billion) in cash and short-term investments at June 30. The Q2 burn was roughly ¥1.6 billion versus ¥3.8 billion in the year-ago quarter. A company that burns roughly ¥1.5 billion per quarter and holds ¥40 billion in cash has roughly 26 quarters of runway at the current pace — more than six years. Even if the ramp into Mona L03 and G9L accelerates spending, the runway stays long.

The company announced on the same day as earnings that its robotics subsidiary, Dogotix, secured $900 million in funding commitments at a $6.3 billion post-money valuation. XPeng retains control — its stake dilutes from 100% to roughly 68% fully diluted — but the company consolidates the unit and keeps upside to what could be a $6+ billion business. The deal also brings in external capital that doesn't drain XPeng's automotive cash pile.

The valuation disconnect

XPeng trades at roughly $12 per share, giving it a market cap around $11 billion. The stock is down roughly 40% in 2026, down over 50% from its one-year high.

The market is pricing this as an EV maker that can't turn a profit, stuck in a brutal China price war with no clear path to breakeven. That narrative ignores what the numbers actually show:

• Deliveries are ramping 65% sequentially with a new model generating supply-constrained demand.
• Q3 guidance calls for 115,000–121,000 deliveries, a 11–17% increase over Q2.
• Revenue guidance for Q3 is ¥21.7–23.4 billion, implying 10–19% sequential growth.
• Cash on hand covers the current burn rate for years.
• Vehicle gross margin is stable at 12.1%, not collapsing.

Compare that to the Chinese EV peers: Li Auto trades at a positive EV/EBITDA of 10.6 and a market cap of $13.3 billion, while NIO trades below $10 with a negative P/E and a smaller market cap. XPeng sits at $11 billion — between them — with the same fundamental reality that none of these companies are cash-flow-positive automakers today. They're all burning capital to scale. The difference is which one has the delivery ramp, the product pipeline, and the runway to survive the burn.

The break condition

The thesis here isn't that XPeng is about to start printing money. It's that the market is treating a product-transition quarter — with higher R&D, heavier SG&A, and a declining subsidy stream — as evidence of structural decline. If Q3 deliveries hit the top of guidance at 121,000, driven by Mona L03 ramping past the supply bottleneck, and gross margin holds or improves as the product mix shifts, the loss-per-share story changes. The same ¥3 billion in R&D and SG&A spread across ¥23 billion in revenue looks different than spread across ¥19.7 billion.

The risk is real. Vehicle gross margin declined year-over-year, and the China EV price war shows no sign of ending. If Mona L03 fails to scale to the 15,000-per-month range Deutsche Bank expects, or if G9L pricing undercuts the margin recovery, the burn rate stays elevated with no offset.

But the double miss on revenue and EPS doesn't signal a broken business. It signals a company spending to grow into revenue that's already starting to show up in deliveries. The ¥40 billion cash pile means XPeng doesn't need the stock market to fund the next 24 months of execution. That's the disconnect: the stock price is pricing in a company running out of runway when the balance sheet says it has years.