XPeng would be worth a sixth of Unitree at J.P. Morgan's targets

J.P. Morgan's RMB995 share-price target for Unitree implies a value of about US$60 billion for the robot maker. In a separate report, its US$10 New York target for XPeng implies about US$9.6 billion for the whole company. Yet XPeng had 45 times Unitree's sales in 2025, mostly from cars. Its IRON robot runs on the same AI chips as its cars and shares most of their suppliers. Mass production is due at the end of this year.

Why one target rose and the other fell

In a note reported on 26 October, J.P. Morgan more than tripled its price target for Chinese robot maker Unitree, from RMB300 to RMB995 a share. The bank still values Unitree at 30 times expected annual sales. Its September target used the sales it expected in 2027. The new target uses 2028, and the bank now expects Unitree to sell RMB13.4 billion that year, up from its earlier forecast of RMB5.3 billion. It expects robot sales to grow faster than it previously thought.

In a separate report on 26 October, J.P. Morgan cut its New York price target for XPeng from US$11.50 to US$10 a share. In November 2025 its XPeng target was as high as US$50 a share. The bank expects weak demand for cars in China to last into 2027.

What investors agreed to pay for a stake in Dogotix

In August, outside investors agreed to put US$600 million into Dogotix, the company that holds XPeng's robotics business. The agreement implies a value of US$6.3 billion for Dogotix as a whole. At the deal price, XPeng's stake would be worth about US$4.7 billion after the planned transfer of shares to staff leaves XPeng with about 74%. The outside investors agreed to buy shares with extra protections, including buyback rights under certain conditions.

What the targets and the funding deal imply

XPulse calculations. The last three rows add up to all of XPeng. The US$2.0 billion is what is left once the robot stake and the cash are taken away. Nobody has set a price for those businesses.

Analysts welcomed the deal in August. Morgan Stanley said what mattered most was the first outside price for the robot business, which lets investors value it apart from the cars. Citi called it strong backing from well-known technology investors. DBS wrote that XPeng’s share price gave the robots little value, so success there could lift it. Six of the eight did lower their XPeng targets within days, but because of weak car demand, not the robots.

Where IRON stands

IRON has come a long way in a year. In a video XPeng posted on 8 September 2025, for the IAA motor show in Munich, an earlier version made a cup of coffee.

An earlier version of the IRON robot, white with a black visor, pours hot water from a black kettle into a coffee filter on a glass jug, at a white table with two glasses, a grinder and a tin, against a black background. — IRON makes pour-over coffee. Picture from XPeng’s video for the IAA show in Munich, 8 September 2025.

Two months later came the new IRON, at XPeng’s Tech Day. It moved so smoothly and naturally that many viewers thought someone was inside. The next day He Xiaopeng had the covering on one leg cut open on stage, and the robot kept walking.

On a dark stage, He Xiaopeng in a grey jacket watches the new IRON robot walk past, the white covering cut away from its lower left leg to show the metal joints inside. — IRON walks on with one leg opened up, as He Xiaopeng watches, 6 November 2025. Picture from XPeng.

Now IRON is moving into production. XPeng opened its production line in September and aims to start mass production by the end of 2026. On 28 October Gu Jie, a key figure in XPeng's robotics team, wrote that IRON's solid-state battery had come off the line in small batches. On 4 November He Xiaopeng, XPeng's chief executive, went further, writing on X of “the solid-state battery we’re now mass-producing for the robot industry”.

Its first job is set. From early 2027 XPeng plans to use IRON to greet shoppers and answer questions in its stores in China. Store work is already a real market. Counterpoint, a research firm, puts guiding and retail at about a fifth of the humanoid robots sold in the first half of 2026. The rest go mainly to shows, teaching and data collection. Unitree already sells humanoid robots by the thousand, mostly for labs and classrooms.

What the robots could bring XPeng

IRON gives XPeng a second business that grows out of the first. It runs on XPeng's own AI chips and shares much of the AI work behind the cars. More than 85% of its suppliers also supply XPeng's cars, He Xiaopeng says. So progress in one business can help the other, and costs can fall as both grow.

XPeng is also its own first customer. It has about 740 stores in China and more than 500 abroad. IRON can learn its job there, in front of real shoppers, before it is sold to anyone else. Deliveries abroad are planned from the first half of 2027, and sales to other companies from the second half. He Xiaopeng has said XPeng expects to keep a bigger share of each robot sale than of each car sale.

It is still early, and XPeng says so itself. He Xiaopeng calls a humanoid at least 20 times harder to make than a car. So XPeng starts with work it controls, in its own stores. Those stores will also be IRON's first test in daily use.

If it works, the prize is large. He Xiaopeng's long-term hope is more than a million robots a year by 2030. Outside investors have already put a US$6.3 billion value on the business before its first sale.

What comes next

XPeng reports its third-quarter results on 16 November. That is the next chance to hear how the year-end plan for mass production is going, and whether the Dogotix funding has closed. AI Day, XPeng's yearly technology show, is also due later this month.

After that, the real test begins: what an outside customer pays for IRON, and whether they come back for more.