Chinese electric vehicle manufacturer Xpeng announced Thursday a decision to increase the size of its U.S initial public offering (IPO) by more than a third after realizing Wall Street can swallow anything so long as it promises a greener tomorrow.
Co-founded in 2014 by two former executives from China’s GAC Group, the EV startup has already managed to produce around 20,000 vehicles for the Asian market. It also became engaged in an intellectual property dispute with Tesla (which claimed Xpeng stole its Autopilot source code) in 2019 and ran afoul with California’s Department of Motor Vehicles after failing to submit disengagement reports on its self-driving test vehicles in 2018.
Such hurdles don’t seem to have slowed the company’s rise to prominence, however. Xpeng is adept at fundraising, amassing well over a billion dollars through strategic partnerships in just the last two years. Meanwhile, the adjusted IPO filed on the New York Stock Exchange this August now targets a cool $1.5 billion USD.
It makes sense in the broader context. EV manufacturers frequently garner valuations that overshadow established automakers that are many times larger — and substantially more profitable. Xpeng itself only has two facilities located in China that produce its G3 sport-utility vehicle and brand-new P7 sedan. While we’ve never driven one (as it’s not sold in North America), the G3 has been praised for its build quality and impressive tech inclusions. Autocar even said it offered “near-Tesla levels of autonomous driving capability” in 2019.
There’s substantially less information floating around about the P7, but it’s clearly targeting the Tesla Model 3 and looks a lot like many other electric sedans we’ve seen debut in the last year or two. Porsche also helped develop the chassis, giving it some extra credibility as a performance model.
According to Reuters, Xpeng’s filing states that it will price its shares at $15 each. It originally suggested they’d go for between $11 to $13 when the deal was launched, however.
From Reuters:
The Guangzhou-based automaker had planned to sell 85 million American Depository Shares (ADS) but increased that to about 99.7 million shares following higher-than-expected demand from investors, according to the filing. Investors now value Xpeng at over $11 billion and its shares will start trading in New York on Thursday.
There is also a so-called greenshoe option in which another 14.96 million shares can be issued within the next 30 days that would allow Xpeng to raise a further $224.4 million.
BofA Securities, Credit Suisse, and J.P. Morgan. are joint bookrunners for the offering.
Xpeng CEO He Xiaopeng said the additional funding will be funneled into research and development and expanding sales. Vehicles (both future and present-day) are targeted between 150,000 yuan ($21,804) to 300,000 yuan to appeal to the market. Currently, that only includes China, though the business aspires to tackle both Europe and the United States.
[Image: helloabc/Shutterstock]
The Chinese market is a lot more accepting of, and conducive for, EV makers, than in America.
If the idea here is to take money from American investors to produce EVs for the Asian market, it may be a lucrative venture for the capitalists.
If the idea is to raise money for more vaporware in the US, I think that has run its course.
Not sure why anyone would *want* Tesla’s Autopilot code.
Zing of the day.
Tesla actually has some really good training data. I’d love to get my hands on it even though I’m using different hardware and a different generation of AI.
“Not sure why anyone would *want* Tesla’s Autopilot code.”
What do you know about Tesla’s Autopilot code? Another empty statement?
When working with machine learning, the training data is the hard part, and the code is relatively easy.
Let me give you a one-liner to explain why: with sufficient training data and computing power, a full-self-driving AI could easily start the day with more driving-miles (“experience”) than any human will gather in a lifetime.
With enough computing power, every release of the software could be trained and tested over more miles than any human driver will ever experience.
That is the endgame for this. The code is a sideshow compared to the training data — for this particular type of software.
@Luke42: It’s not just for training, it has huge value for testing. The AI technology I’m working on is less dependent on traditional AI training, but learning how it reacts to real-world situations is important. That training data is huge though.
@Inside: Chill; it’s a half-joke. We all know the outcomes of people who mistake AP’s Level 2 capabilities for FSD, which it is not.
As much as I support Tesla, I have no interest in AutoPilot or vehicle autonomy in general.
In the age of hyper inflation, anything is worth that much money.
The only thing that is inflated at the moment are stock prices.