[BidClub_]
Sohn Conference Foundation · · 9 分钟

Chris Drose在Sohn 2025推介Aurora Innovation

Chris Drose

YouTube
TL;DR
  • Chris Drose(Bleecker Street)做空Aurora Innovation(AUR)——这是一家市值140亿美元、基本尚未产生收入的自动驾驶卡车公司,他开场就称其在州际公路上搞的是“公路抢劫”(highway robbery)。 Aurora自成立以来已亏损40亿美元,拥有1800名员工,其中1600名工程师在“手工标注路线”;5月1日完成的首次全自动货运,只是在一个阳光明媚的周日下午,跑了一条平坦、低货量的休斯敦—达拉斯线路。
  • Aurora从当前10辆产生收入的卡车扩张至2027年部署1万辆的增长故事,正受到其OEM合作伙伴的质疑。 上线卡车上的Peterbilt标志被涂黑;Bleecker向PACCAR询问时,对方表示:“在我们看来,自动驾驶卡车走向商业化仍处于相当早期的阶段。任何未经证明足够安全的产品,我们都不会同意商业化。我们还没到那一步。”Drose称,他与Volvo一名高管沟通后了解到,2027年的产品还没到位;他认为时间窗口“更可能是2030年左右,而不是2027年”。
  • 资金账是核心:Aurora称还需要6.5亿—8.5亿美元才能实现现金流盈亏平衡;Bleecker认为“这个数字更接近20亿—30亿美元”。 原SPAC路演材料声称,公司融资额足以撑到产品上线;此后Aurora又以更低价格融资8.2亿美元,自IPO以来股本被稀释50%,并在2月宣布设立5亿美元ATM增发计划。
  • 1月CES上NVIDIA与Continental宣布“规模化部署无人驾驶卡车”,推动股价上涨51%;但这“基本就是”Aurora在2018年宣布过的同一条消息。 一名行业老兵对此“真的感到震惊”:“我看到时简直不敢相信。他们已经宣布过了。”Bleecker认为其使用的“只是现成的NVIDIA芯片”,另一名老兵则表示,“打开引擎盖看,Aurora的自动驾驶技术栈没有任何特别之处。”
  • 在“曲棍球棒式增长和商业化前夜”,内部人士却在离职并卖出股票——这不是Drose想看到的信号。 联合创始人Sterling Anderson已离开公司;他的辞职将在3周后正式生效,届时可以出售4480万股。Aurora的工程负责人于2024年8月离职,总法律顾问于2025年1月离职。CEO Chris Urmson启用了新的Rule 10b5-1交易计划,Reid Hoffman在11月以5美元的价格卖出1480万股,Anderson则以2—3美元的价格卖出价值800万美元的股票。Drose称,Aurora必须把手头剩下的钱全部用于强行推动市场,并补充说:“如果我是Sterling Anderson,我不确定自己会留下来。”
  • 竞争进一步加剧了烧钱问题:Kodiak将在今夏上市,Drose还提到其他商业模式更好、烧钱更少的竞争者。 Waymo已经在道路上取得成功,而Cruise如今已破产。一名行业专家表示,自动驾驶卡车公司需要在产品上线后再存活10年,届时才可能开始抵销成本;Drose认为,以Aurora的商业模式,公司撑不到那时。
摘要 · 为研究而整理的核心内容

1. “自动驾驶卡车领导者”与现实之间:市场交易的是认知差

  • Drose做空的框架是:寻找市场认知与现实之间的偏差。市场认知是,Aurora是自动驾驶卡车领域的领导者。Bleecker接触的人也认为,公司很快就能实现规模化,并且已经拥有清晰的商业化路径。现实则是:“这些人是技术派。他们显然没有认真考虑过商业模式”,而Aurora是“自动驾驶里程的高成本生产商”。
  • Drose援引前CFO对管理层的评价:只能打C或D。

2. 华尔街喜欢的合作伙伴,正在质疑商业化时间表

  • 5月1日上线卡车上的Peterbilt标志被涂黑,颇为醒目。Bleecker向PACCAR询问时,对方表示“仍处于相当早期的阶段”,并补充称,任何尚未被证明“超级安全”的产品都不会商业化:“我们还没到那一步。”Aurora称,Volvo的自动驾驶卡车将在2027年准备就绪;但Volvo正在自动驾驶业务及其主工厂裁员。Drose称,他与Volvo一名高管沟通后了解到,该产品还没到位;他认为时间窗口“更可能是2030年左右,而不是2027年”。

3. 催化剂反复使用,技术趋于商品化

  • CES上的NVIDIA公告推动股价上涨51%,但重复了2018年的说法:Aurora当时称将在“未来几年内”让汽车和卡车上路。如今Aurora聚焦卡车,理由是卡车业务拥有更清晰的商业和监管环境。
  • Drose认为,这项技术使用的“只是现成的NVIDIA芯片”。一名行业老兵表示,Aurora虽然做得非常好,但“打开引擎盖看,其自动驾驶技术栈没有任何特别之处”。另一名行业老兵对Aurora再次宣布这一合作“真的感到震惊”。1600名工程师手工标注路线,凸显了这项业务的高昂成本;Drose还以如今已破产的Cruise为例。

4. 填补资金缺口,盯紧内部人士

  • Aurora称需要6.5亿—8.5亿美元才能实现现金流盈亏平衡;面对公司“长期过度承诺、交付不足”的记录,Bleecker认为实际数字为20亿—30亿美元。自IPO以来,Aurora又以更低价格融资8.2亿美元,股本被稀释50%,并宣布设立5亿美元ATM增发计划。
  • 收尾证据是:Anderson离开公司,并以2—3美元的价格卖出价值800万美元的股票;他的辞职将在3周后正式生效,届时可以出售4480万股。Hoffman在11月以5美元的价格卖出1480万股,Urmson则启用了新的Rule 10b5-1交易计划。Drose称,Aurora必须把剩余资金全部用于强行推动市场,并表示如果自己处在Anderson的位置,不会留下来。
Chris Drose

We’re going to talk about highway robbery happening on our country’s interstates. Bleecker Street is short Aurora Innovation, a $14 billion market-cap, largely pre-revenue autonomous-trucking company. It completed its first fully autonomous freight run on May 1 between Houston and Dallas. All right, I knew we’d get AI or autonomous in here somewhere.

It lost $4 billion since inception, and right now it has 1,800 employees, 1,600 of whom are engineers basically manually annotating routes. They were able to complete 1 route on a sunny Sunday afternoon, on a straight line between Houston and Dallas. There’s not much freight that moves between Houston and Dallas. It’s flat, not hilly. It would be a very boring drive—perfect for an autonomous truck.

1. Aurora’s Scaling Story Unravels

Now it’s a scaling story. Congratulations: They got the tech where it needed to be, and now it’s a scaling story. They have 10 revenue-generating trucks right now, and they claim to be scaling to 10,000 trucks deployed by 2027. Our research unveils key issues with that time frame. Their OEM partners disagree with Aurora on that timeline. There’s new competition: Kodiak is going public this summer. There’s also a lot of competition with better business models and less burn.

We think Aurora needs to raise about $750 million more to get to commercialization. Aurora says it needs to raise $750 million to get there, but it has a long history of overpromising and underdelivering. We think it needs to raise $2 billion to $3 billion. Meanwhile, executives are selling and executives are leaving, so we think it’s a pretty good setup.

2. Perception Masks The Reality

In general, we try to understand what the perception of a company is in the market and whether reality is sufficiently different. That is what makes something an attractive short to me. The perception, when we started looking at this, was that Aurora is the leader in autonomous trucking. We believe Aurora spent $4 billion, was able to take the driver out first, and now the real challenge begins. It is the high-cost producer of autonomous miles, which we think matters a lot as the business model of autonomy develops over the next decade.

3. Aurora Lacks A Business Model

We’ve talked to people who said Aurora has a business model, that it will scale soon, and that there’s a clear path to commercialization. As we said, it doesn’t really have a business model. These are tech people; they clearly have not thought about the business model. Wall Street loves partnerships—that was something that was said a lot. They have partnerships with Volvo and PACCAR to actually get the trucks on the road. Unfortunately, both of them disagree with Aurora on the scaling timeline. Also notable: A former CFO gave Aurora management a C or a D grade. Not amazing.

On May 1, they took the driver out and did this freight run between Houston and Dallas. You’ll notice that they’re scaling with a Volvo truck, but PACCAR’s Peterbilt has built the majority of their trucks so far. Both OEMs are going to help them scale. You’ll notice that the Peterbilt logo is blacked out on the truck. There was a lot of press and excitement around this event, and we were wondering why.

4. Partners Reject The Timeline

So we asked PACCAR, and they said, “In our view, it’s still pretty early days on the road to commercializing autonomous trucks. We will not agree to commercialize anything that is not proven to be super safe. We’re not there yet.” Basically, they told Aurora to take it off. They said, “Yeah, leave us out of this one.”

On the industry scaling timeline, an industry expert said they need to survive 10 years past launch, at which point they may start offsetting their costs. It’s a very expensive business, and unfortunately, they don’t have the business model to really make it 10 years, in our opinion.

5. Volvo Pushes Launch To 2030

PACCAR is 1 truck partner, and Volvo is the other. Volvo has an autonomous truck that Aurora says will be ready in 2027. Volvo has a long history of missing deadlines in autonomy. We talked to a senior Volvo executive about the 2027 Volvo product. It’s not there yet, because it’s already 2025. I think the time window is really about 2030, not 2027. We don’t think Volvo signed off on this, and we don’t think they’re there yet. Meanwhile, Volvo is doing layoffs in autonomy and at its main plant. I think it’s going to take a lot more capital and a lot longer than Aurora is saying.

6. Competition Undercuts Aurora’s Tech

As we discussed, there’s a lot of competition. Aurora wants you to think that it has a tech and business lead. Another industry veteran said Aurora has done a very good job, but under the hood, there’s nothing special in its AV stack, or autonomous-vehicle stack. There are a lot of engineers working on this manually, and that’s very expensive. Look what happened to Cruise; it is now bankrupt.

In January, on the eve of commercialization, Aurora went to CES and announced that it was partnering with NVIDIA and Continental to deliver driverless trucks at scale. The stock went up 51% on this news. That was basically the same news they announced in 2018, when they said NVIDIA and Aurora would collaborate to build a next-generation autonomous-vehicle compute platform. At that time, in 2018, they said, “We will have cars and trucks on the road in the next couple of years.”

Now they’re just focusing on trucks. They pivoted to trucks because they said it was a cleaner commercial and regulatory environment. Meanwhile, Waymo is already succeeding on the roads. We talked to another industry veteran, and he said he was genuinely shocked that this is what they announced: “I couldn’t believe it when I saw it. They have already announced it.” Meanwhile, we think it is just off-the-shelf NVIDIA chips.

7. Aurora Burns Cash As Insiders Exit

Aurora has a very long history of understating how much capital is needed in this business. Its original SPAC presentation claimed it was raising enough money to get to launch, which would have gotten it to what happened this May. Since then, it has raised an additional $820 million at prices that are much lower than where they are now. It has diluted its share count by 50% since the IPO and announced a $500 million ATM in February. Last week, it said, “We are going to need $650 million to $850 million more to reach cash-flow breakeven.” However, we think that number is closer to $2 billion to $3 billion.

Meanwhile, you’re seeing insiders leaving and selling, which is not what you want to see on the eve of hockey-stick growth and commercialization. Its co-founder, Sterling Anderson, left the company. Aurora’s engineering lead left the company in August 2024, just a few months after presenting at its analyst day. Its general counsel left the company in January 2025.

CEO Chris Urmson just entered a new Rule 10b5-1 plan. Reid Hoffman sold 14.8 million shares in November at $5 per share, and Sterling Anderson sold $8 million worth of shares between $2 and $3. He can sell 44.8 million shares post-resignation, with that resignation becoming effective in 3 weeks.

A former PACCAR senior employee is another key scaling partner for them. Aurora has to go out there with whatever money it has left and try to force the market. I’m not sure I would stick around if I were Sterling Anderson. We are short Aurora. There will be a full report going up on our website shortly.

Chris Drose在Sohn 2025推介Aurora Innovation — 文字稿与摘要 | BidClub