Pony.ai's Funding Story: Daydreams in a Winter Storm

By Jiaxiang Shi

Edited by Zhiyan Chen

After eight years, Pony.ai — once China's most valuable autonomous driving company — finally went public last night. It opened up 13%, closed down more than 7%, and now carries a market cap of $4.2 billion.

Pony.ai's founding story began in late 2016, in what was arguably the best year for autonomous driving. Cash-flush internet giants charged in: Uber acquired self-driving truck startup Otto for $700 million; Google's self-driving project Waymo announced its independence. Traditional automakers seeking transformation weren't about to be left behind, unveiling their own autonomous driving plans. Even chipmakers like Intel jumped in, partnering with BMW and Mobileye.

Peng Jun and Lou Tiancheng, then at Baidu's autonomous driving division, decided over coffee to leave Big Tech and start Pony.ai. In those years when mobile internet dividends were peaking, autonomous driving became the area where investors placed their biggest bets. Naturally, this classic strong-technologist duo attracted a roster of top-tier institutions, and Pony.ai's valuation rocketed to the top of the autonomous driving heap.

In some ways, Pony.ai's funding history is a condensed version of NIO's story: from top-tier dollar VCs circling, to embracing traditional automakers' corporate funds, to finally landing sought-after Middle Eastern sovereign wealth funds. It was fortunate enough to ride through different funds' honeymoon periods with autonomous driving.

In June 2021, with financial investors passing the baton, Pony.ai kicked off its IPO process. It should have gone public when capital and confidence were both abundant, ready for market scrutiny. But history allows no hypotheticals. As the situation for Chinese companies listing in the US took a sharp turn for the worse, Pony.ai's IPO was forced to terminate.

Today, autonomous driving is no longer seen as frontier investing, massive losses still fill the prospectus, and Pony.ai's market cap has halved from its last private round. For sovereign or corporate funds that doubled down in later rounds, or early financial investors who three years ago might have seen a $10 billion IPO — the low-hanging fruit now clearly "isn't enough to quench their thirst."

But this may be the naked reality tech investors must face: if a company doesn't ride the wind of a still-hazy technological inflection point straight up, then the young prodigies who are still somewhat green in the business world will not be embraced by the broader market.

Dark Waves spoke with nearly ten of Pony.ai's investors and interviewed founder Peng Jun, trying to parse when exactly is the best moment for venture capital to invest, when emerging tech companies face valuation compression and real commercialization challenges?

The Final Stretch

In June 2021, Pony.ai was just one kick away from the goal.

Four months prior, the Chinese autonomous driving company had completed a $250 million Series C+ round at a valuation exceeding $5 billion. Investors included Ontario Teachers' Pension Plan's Teachers' Innovation Platform (TIP), Brunei Investment Agency, and CPE Sourcepeak.

People familiar with the round said Tesla was surging in the secondary market, Waymo's funding was going smoothly, and large foreign pension funds and sovereign wealth funds — many of them LPs to Chinese GPs — were opening offices in mainland China or Hong Kong. Pony.ai, the highest-valued company across the ocean, naturally attracted heavy bets.

By then, Pony had already accelerated its expansion. 36Kr reported that in December 2020, Pony.ai established its trucking division and registered PonyTron, with the trucking team peaking at around 100 people; in June the following year, it set up a vehicle team of more than ten people in Shanghai's Jiading district, pushing into car manufacturing.

Every move pointed toward going public. "They were seriously considering the IPO," one Pony.ai investor said. At one point, the IPO price exceeded $10 billion, even $12 billion.

But a black swan stopped everything. In August 2021, the SEC called for a "pause" on Chinese company IPOs, and more than a year of preparation was terminated by force majeure. That autumn was also the starting point of dollar funds' decline.

"A high-speed machine suddenly hitting obstacles inevitably causes turbulence," one investor said. First, the car-making plan was indefinitely shelved; multiple core employees of PonyTron left to start their own companies, one of which was even backed by Pony.ai's existing shareholders.

In that eventful autumn, in October, a Pony.ai test vehicle hit a road sign on a median while changing lanes in California, ultimately leading to a recall of a specific version of its L4 autonomous driving system — a world first.

According to Dark Waves, switching to Hong Kong or A-shares was put on the table, with "ongoing discussions about whether there was opportunity and window," but Pony.ai was not yet profitable.

Peng Jun said in an interview that industry uncertainty was a "pretty big challenge" the team faced — large-scale commercialization was still not in sight — and if commercialization dragged on longer, would they have sufficient resources to sustain development?

One Pony investor said that 2017-18 was still the hazy technology period; by 2021 — much like large language models today — there were no airtight circles in the market. With a little due diligence, anyone could understand what commercialization or technical problems remained unsolved. Combined with wavering confidence in Chinese stocks' future, most people then believed that if Pony didn't go public, it would be hard to raise money in the private market at its $5 billion-plus valuation.

No one knew how long the winter would last. One investor recalled that Pony.ai joked with them then about regretting not having raised more money.

Beyond layoffs and business contraction, they needed to find a directly commercializable path — when L4 was constrained by low-probability but complex scenarios, L2 became a temporary lifeline — even at one point seeking to become an assisted driving supplier. 36Kr reported that after its IPO failure, Pony.ai diverted some resources to developing an L2++ assisted driving project: in Guangzhou's Nansha and Beijing's Yizhuang, Pony.ai-assisted driving vehicles without roof-mounted LiDAR could often be seen on the streets.

Under these challenges, Pony.ai survived.

The Scramble

Rewind to 2016, and Pony.ai's dream beginning.

After deciding to start up, Peng Jun's first call went to Hongshan partner Kui Zhou.

That January, Zhou had attended Hongshan's annual meeting in Silicon Valley, where he met Peng Jun during a visit to Baidu's US research lab. In October, Zhou received Peng's call — he had just returned to Beijing, was considering starting a company with Lou Tiancheng, and wanted to do autonomous driving. Zhou quickly pulled in Neil Shen and Xin Fu to meet Peng.

Lou Tiancheng still remembers that at a dinner before the formal investment, he sat at the same table as Shen, who asked him a raft of questions unrelated to autonomous driving, seemingly random ones, but he answered them all. Later Shen said he wanted to use this kind of meandering conversation to see a person's essential state.

In the Series A, Pony.ai opened up 20% of its cap table. Hongshan led the round at a $70 million valuation with 15% ownership, and went on to invest in four consecutive rounds; it is now Pony.ai's largest financial investor, holding 10.2%. Because it touched both TMT and automotive, Pony.ai was also a rare Hongshan portfolio company with three partners jointly involved.

Zhou told Dark Waves that 2016 was still early days for autonomous driving; Hongshan's read on the sector was that the opportunity was massive but the journey would be difficult, requiring long-term commitment and investment with no shortcuts. Second, this was a startup high ground that required massive aggregation of technical talent.

Peng Jun was the first member of Baidu's US research lab, serving as architect of Baidu's autonomous driving and one of the company's single-digit T11-level engineers. Lou Tiancheng was the undisputed "world's top programmer," had worked on Google's self-driving team, and was Baidu's youngest T10-level engineer, known as "Master Lou." In early investors' eyes, such a combination would significantly attract other technical talent, laying the foundation for autonomous driving.

One investor told us a detail: as long as Master Lou showed up for a meal, classmates from Tsinghua's Yao Class that year were willing to join.

Peng Jun recalled to Dark Waves that they didn't do a massive roadshow for the first round, didn't deliberately chase valuation, only called four investors, of which Hongshan and IDG chose to support. In the company's second round, Hongshan had facilitated NIO Capital's investment in Pony.ai.

Another key early fund was 5Y Capital. In 2017, 5Y Capital partner Zhang Fei first met founder Peng Jun at Pony.ai's Beijing office. Peng described this chat as "meeting a kindred spirit," and Zhang's story of investing in Kuaishou was key to why the conversation with 5Y continued.

Zhang had invested 2 million when Kuaishou was still "GIF Kuaishou" and Cheng Yixiao was the only person there. The next year, when Kuaishou hit a bottleneck, Zhang convinced Su Hua to join by proposing that 5Y and Cheng's team each dilute by half, set aside 50% as an option pool, and give most of those shares to Su Hua's team.

This was a rare moment when an investor changed a company's destiny.

Recalling this meeting to Dark Waves, Zhang mentioned that "James and Tiancheng both have strong engineer traits, and are very pragmatic." When investing in founders, he often worried that overly idealistic people wouldn't be grounded in reality, "but their styles were very balanced." Zhang believed that in a field like autonomous driving that "needs to be redefined," "the only thing you can rely on is the team."

The Series B led by 5Y was the round that truly put Pony.ai on the investment map. By then Pony.ai had a prototype vehicle and a team of about twenty-some people. "It was a very good fundraising," Peng Jun's schedule was packed every day, seeing at least five or six investors daily.

Sustar Capital partner Shen Xiao remembers meeting Peng Jun and Lou Tiancheng at a hotel in the US, with no preamble, pulling out a whiteboard and diving straight into L4 technical details. "Actually we were still quite far from L4 then, LiDAR and drive-by-wire redundancy were both very immature."

"But you believed L4 could be achieved, and that L4 could dimensionally reduce to beat L2," Shen Xiao told Dark Waves. After the second meeting, Sustar decided to invest.

Because it was so sought-after, Pony adjusted its valuation in this round "more than three times." The B+ round started almost immediately after, at a $900 million valuation. "From July to September, constantly iterating sensor solutions. The market was good, and everyone accepted this kind of jump."

Lin Haizhuo, founding partner of ZY Capital, said that before 2020 Pony.ai basically maintained a rhythm of raising every half year. When he connected with the team, the B round had already closed, and the B+ round received over 40 term sheets, ultimately selecting about ten funds. Lin got in because he was an alma mater connection with Peng Jun and Lou Tiancheng.

Also investing in this round, VMS Group managing director Gong Biao told Dark Waves, "the market indeed had very high expectations for autonomous driving." Some investors believed that "the market had a fear-of-missing-out mentality, so they'd rather be wrong than miss out."

A Brain Seeking a Body

Toyota's entry in 2020 was particularly important in Pony's funding history. For intelligent driving companies, lacking car-making capabilities themselves, bringing in an automaker was essential. If autonomous driving is the brain, then the car is the body.

"For a person, no matter how brilliant the brain, it needs a healthy body," Lou Tiancheng said in an interview.

That February, Pony.ai received a $400 million investment from Toyota at a $3 billion valuation — triple the previous round. Their connection traced back to August 2019, when Pony conducted public road pilots in Beijing and Shanghai using Toyota's Lexus RX vehicles.

To complete this round, Peng Jun flew to Japan six or seven times that year, with negotiations stretching over half a year. From specific project cooperation, to strategic partnership, and finally to investment, "there was no important milestone or decisive moment with Toyota, it was a process of slowly heating water, getting hotter and hotter," Peng Jun told Dark Waves.

Toyota told Dark Waves that its main reason for investing in Pony.ai was that for the next 5-10 years and beyond, China would be the world's largest market for highly autonomous mobility services. To advance Robotaxi commercialization as early as possible, they chose "to cooperate with local technology enterprises."

Why did Pony.ai win out?

One investor told Dark Waves that at the time, Roadstar — another star autonomous driving company — was closer to Toyota. But Roadstar later fell apart due to divergent business visions and equity distribution, with its original CTO leaving to found DeepRoute. "Pony seized this opportunity." Peng Jun's version, however, was that Roadstar never even met Toyota's senior leadership, and was far from an investment.

There's a sidebar here. Before Xiaomi officially announced its car-making, it had also considered investing in Pony.ai. Peng Jun and Lei Jun discussed it multiple times. But Pony's valuation had already exceeded $5 billion, plus Xiaomi preferred L2 assisted driving, so it ultimately chose to buy DeepMotion for $500 million — a cheaper autonomous driving solutions company. A DeepMotion investor said, "Xiaomi's core DNA is extreme cost control on top of making good products, having good value for money."

Unlike Horizon Robotics, which focuses on L2 assisted driving and serves multiple automakers, Pony bet its chips on Robotaxi (driverless taxis), which follows the L4 route (vehicles can be fully autonomous in most situations), with control resting in the autonomous driving companies' hands.

So Pony and Toyota's cooperation is better described as joint R&D — for example, its fifth, sixth, and seventh generation autonomous driving systems were all first launched on Toyota vehicle platforms.

But Toyota also established its autonomous driving department in 2018, externally invested in Uber's autonomous driving, and domestically invested in Momenta — Pony.ai was not Toyota's only choice.

The Real Test

"Top companies have multiple lives," one Pony investor said with feeling. This perfectly summarizes Pony's fate after its IPO was blocked.

Fortunately, life-saving money still came.

In March 2022, Pony.ai announced the completion of a $190 million Series D led by China-Arab Investment Fund, with Ontario Teachers' Pension Plan, 5Y Capital, ZY Capital, Carlyle Investment Group, Cuoming Investment, Raumier Limited, Evodia Investment, ASSETKEY LIMITED, and others.

In October the following year, Pony.ai announced another $100 million investment from Saudi Arabia's NEOM and its investment fund NIF (NEOM Investment Fund).

Pony.ai's Series D valuation was $8.5 billion. People close to this round said that for Middle Eastern countries, as dollar funds cooled and political winds shifted, they instead saw themselves picking up good companies in a valley. "In the investment, industry probably accounts for 80%, with the remaining 20% calculating financial returns."

At the time, Middle Eastern money was flowing more toward traditional energy projects; TMT-related projects were less common, and Pony.ai was among the earliest. "Dollar funds were gradually drying up, RMB funds had shorter exit timelines, making it hard to invest in projects with long commercialization paths like autonomous driving. So the Middle East was the most promising," Peng Jun told us, saying Middle Eastern money was the result of their "targeted search."

However, according to Dark Waves, some shareholders invested through a combination of secondary shares plus new shares. One investor said that when they bought in, "new shares were priced around $7 billion, secondary shares around $1 billion."

The halo effect sustained Pony.ai's position as the top autonomous driving project, but its real test lay in the secondary market. In contrast to its broad embrace in the private market, Pony.ai's market cap on Nasdaq was cut in half from its last private round.

Multiple Pony investors told Dark Waves that the current price is a range where capital markets can generally reach consensus.

Peng Jun also responded to Dark Waves that private market valuation mainly depends on resources and external confidence; after entering the secondary market, valuation no longer matters, "whether it's a mule or a horse, take it out for a spin and you'll know immediately." Internally, they also believe they must seize this rare listing window.

Three years ago when preparing to go public, Pony.ai's valuation exceeded $10 billion — higher than NIO's current market cap; three years later, Pony has obtained driverless mobility service permits in four first-tier cities — Beijing, Shanghai, Guangzhou, and Shenzhen — with over 250 Robotaxis, yet the market's valuation is less than half what it was then.

Its prospectus says it all: net losses of $148 million in 2022, $125 million in 2023, and $51.78 million in the first half of this year. The commercialization of autonomous driving is almost identical to the question of when the VR era would arrive: every year is year zero, every year waiting for the "iPhone moment." The secondary market, which calculates valuation with an eye to the bottom line, clearly isn't willing to pay for a not-yet-profitable Robotaxi story.

Six years ago, when Li Jiaqing discussed Joy Capital's logic for investing in Pony.ai, he said, "What needs to be determined is, is this the most important direction of transformation? If yes, we invest money. For early-stage investment, especially technology investment, it's hard to think with mature industrial logic. Asking right away, what's the business model? I don't know."

Redpoint China managing director Lü Huangxian also told Dark Waves that they knew at the moment of investment it was a long shot.

Buying into technology dreams, trading time for high-multiple growth — this was VC's fundamental logic for a long stretch. Today, as a realism narrative focused on fundamental discounting has become mainstream, investors who poured into unicorns in the mid-to-late rounds face a reality of meager profits or even losses.

If internet fundraising was a series of flowing capital feasts, then Pony.ai illustrates a tech company's fate after investors return from dreams to reality.

Multiple forces created its situation: the sudden shift in US-China relations, the pivot in investment institutions' preferences, the difficulty of fundraising and the DPI-only mentality, and more. But it won't be an anomaly, and may even become the norm.

The more important question is: is extreme realism correct? What kind of capital soil does true innovation need? Don't forget, autonomous driving came from daydreams.

One Pony investor told Dark Waves that the right asset matters more than the right time. This may be self-consolation; perhaps they still remember Pinduoduo's story: under $30 billion at IPO, breaking issue price at least four times in its first year, to over $130 billion today. Perhaps Pony's listing is only the starting point.

Two months ago, renowned angel investor and Neu Venture Capital founder Jerry Neumann announced he would no longer invest in new companies. In bidding farewell to his career as a "solo venture capitalist," he said that startups now no longer seem to be driving progress, but more like doing business — perhaps this is the current state of investing.

Either way, may the fishermen still be willing to go out to sea, and may there still be daydreams in the winter.

Image source: The Night Train (David Cox, 1849), Birmingham Museum and Art Gallery, Birmingham