XPeng's Robot Just Got a $6.3 Billion Valuation. Its Stock Hit a 52-Week Low.

XPeng's Robot Just Got a $6.3 Billion Valuation. Its Stock Hit a 52-Week Low.

On Monday morning XPeng announced that its robotics business had raised more than $900 million in a first funding round, at a post-money valuation above $6.3 billion, which the company called the largest single-round private financing in the history of China's embodied-AI industry. That same morning, shares of the parent, XPEV, traded down about 4.6%, within cents of a 52-week low, down roughly 42% for the year.

A record AI round and a falling stock on the same day is a contradiction worth sitting with. The obvious reading of the headline — a company just created a $6-billion line of business, buy the stock — is exactly what the market refused to do. The reason matters more than the headline, because it tells you which of XPeng's businesses investors believe they actually own.

The two events were really one event. XPeng reported second-quarter earnings before the open, and those numbers were the news. Revenue was RMB19.74 billion, about $2.9 billion, up 8% from a year earlier on a gross margin of 20.7%. Deliveries, at 103,295 vehicles, were nearly flat year over year. The adjusted loss ran wider than analysts had expected, and the company guided third-quarter revenue and deliveries below what Wall Street was looking for. The robot round got the press release. The earnings got the price.

Then there is the valuation itself, which is where the day turns genuinely strange. The parent is worth about $11.1 billion in the public market. The robot unit just received a private mark of $6.3 billion. Because XPeng says it keeps control of the unit and keeps it consolidated, a public shareholder already owns most of that $6.3 billion. Now do the subtraction: $11.1 billion of market value has to absorb the robot stake and the company's roughly $6 billion of cash and short-term investments, and that leaves almost nothing left over to pay for the car business — the thing that delivered 103,000 vehicles in a single quarter and generates on the order of $11 billion of trailing revenue.

Two markets, one company, two wildly different prices for the same asset. Something has to give, and the way to find out what is to look at what each number actually measures.

A private round is a negotiated number, not a market-clearing price. Nobody can sell shares of the robotics unit tomorrow at $6.3 billion; there is no public market in it. The round was led by IDG Capital and Gaorong Ventures, with Tencent and Alibaba in as strategic holders. These are not indifferent checks. By the standards of the sector, $6.3 billion isn't even an outlier — Figure AI, an American rival that likewise discloses almost no revenue, is privately valued at $39 billion. And one account of the round's structure suggested that a meaningful share of the $900 million came from XPeng's own subsidiary and its leadership rather than outside investors. Private valuations in a frothy sector are partly a fundraising signal. You cannot rule out that the $6.3 billion is one.

The public market, by contrast, clears. It cleared at about $11.60, and it treats XPeng not as the owner of a $6.3 billion robot but as a carmaker that still loses money and just guided below expectations.

That points at the question the valuation hype skips: has anyone actually demonstrated they want this robot? The machine, IRON, was unveiled in late 2025, but there is no price, no order book, and no customer. The first units are slated for XPeng's own stores and campuses, with a stated target of more than 1,000 units a month by the end of 2026, and the first deliveries to anyone outside XPeng in 2027. Using the product on yourself before selling it to strangers is the right way to start. But it is a test of whether the product has a customer. It is not revenue.

Seen that way, the round is honest in a way the $6.3 billion headline is not. The money is real, mostly external, and it funds an expensive bet at nearly no cost to public shareholders, who keep their stake and their consolidation. That is close to free optionality, and it's a good deal for a shareholder if you think the founder — who took direct control of the robotics unit midyear — can make physical AI real. But an option is a price paid for a chance, not a value. The value appears on the day robots are sold to people who don't work for XPeng, at a price, at a margin, and that day is at best a year away.

So here is the thing to watch, and it's not the $6.3 billion. It's whether the car business keeps being what the market prices it as: flat deliveries, stubborn losses, guidance below expectations. For now the market has answered, and the answer is written in the 52-week low — a robot valuation you cannot trade is worth less than a car business that's still losing money, apparently by a lot. The stock will start paying attention to the robot on the day someone outside XPeng pays for IRON. Until then, a record AI round is a signal about where serious money wants China's embodied-AI exposure, and a reminder that the stock is priced on cars.