Xpeng's first-quarter 2026 gross margin hit 20.6%, ahead of Tesla's 16.9% automotive margin, and the Chinese EV maker is now pushing its own AI chips into a robotaxi fleet and a humanoid robot, both aimed at mass production before Tesla ships Optimus or Cybercab at scale.
He Xiaopeng runs a car company that Western investors have mostly ignored for years. That's getting harder to justify. Xpeng's first-quarter 2026 gross margin hit 20.6%, the company reported, comfortably ahead of the 16.9% automotive gross margin Tesla posted for its second quarter of 2026, or 16.3% once you strip out regulatory credit sales. Five percentage points separate Xpeng's margin from where it stood a year earlier. It got there even as first-quarter revenue fell 17.6% year over year to 13.03 billion yuan, and the company posted a net loss of 1.78 billion yuan.
Then the second quarter happened. Xpeng delivered 103,295 vehicles between April and June, a 65% jump from the 62,682 it delivered in the first quarter. Much of that rebound came from the ramp-up of the GX, the company's new flagship SUV. Xpeng has guided second-quarter revenue of 19.6 billion to 20.8 billion yuan, which would put combined first-half 2026 revenue above 32 billion yuan. The company reports full second-quarter results before the US market opens on August 24, and the question hanging over that report is whether the delivery surge held the margin near 20% or whether ramping a new SUV ate into it.
While Tesla still hasn't shipped a Cybercab or put Optimus into real production, Xpeng is racing toward both a robotaxi fleet and a mass-produced humanoid robot in the second half of 2026. It's doing that on chips it designed itself. Xpeng's robotaxi runs on four of the company's in-house Turing AI chips, delivering a combined 3,000 TOPS of computing power, which Xpeng says is the highest of any mass-produced vehicle in the world. The company plans three self-developed, Level 4 robotaxi models, with pilot operations starting in Guangzhou, where Xpeng is headquartered. The humanoid robot, called IRON, stands under 170 centimeters, runs on an all-solid-state battery, and packs three Turing chips of its own. Xpeng is targeting 1,000 units of IRON in 2026, with mass production beginning in the second half of the year. That's a small number next to Tesla's public ambitions for Optimus. But Xpeng isn't trying to out-promise Tesla. It's trying to out-build it, on hardware it doesn't have to buy from Nvidia or anyone else.
Xpeng puts its founder in charge of the humanoid robot race Xpeng CEO He Xiaopeng is taking direct control of the company’s robotics unit as its IRON humanoid program moves toward mass production. The decision shows Xpeng wants to be seen as a physical AI company, not just another EV maker in China’s crowded market. - xpeng humanoid robot development under founder leadership - how xpeng ev maker pivots to physical ai
Frankly, the chip strategy is the real story here: the margin number alone doesn't cover it. Tesla still leans on outside silicon and on Optimus and Cybercab timelines that keep sliding. Xpeng is shipping vehicles today at a higher gross margin than Tesla's core auto business, while running its robotics roadmap on chips it builds in-house instead of sourcing them. Profitable cars funding a vertically integrated robotics business is the part of this story that Western investors have been slow to price in.
None of this means Xpeng has won. A single strong quarter doesn't erase years of losses, and a run of 1,000 humanoid robots is a pilot program, not a business. But the direction He Xiaopeng has set is unmistakable, and it points straight at the parts of Tesla's story that used to look untouchable.
Also read: Americans Moving to Canada: The Financial Decisions That Cannot Wait Until After the Move • Defense-Tech Roll-Up Lyntris Falls 11% in Its NYSE Debut as LYNX • A UAE Startup Hit $18 Million ARR With 14 People and No VC Money