Rivian: Not So Harmless
Edited by Brian Birnbaum.
Below follows a summary of the deep dive and further down, the full deep dive.
Deep Dive Summary
In the 20th century, the auto industry was about vibes (looks/performance) and manufacturing–which is the case to this very day, with the caveat that software has become just as important. Rivian has managed to coalesce all three drivers to become one of the biggest threats in the auto market over the coming decades.
Rivian´s DNA is tailored to satisfy the needs and wants of 21st century customers. Meanwhile, the majority of the industry is only starting to consider this question. While gross margins remain negative, this may be the price to pay to muscle into the auto industry going forward.
As you will see in Section 2.0, margins are rising rapidly, even in the face of a tumultuous macroeconomic environment. This suggests that, at its core, Rivian may be an excellent organization.
Rivian has created a culture and a platform that enables fast iteration, across its hardware and software stacks (Section 3.0). As a result, Rivian is also a strong contender in the autonomy space, given the robustness of the network it is deploying.
People love Rivians: they have struck a nerve in a way reminiscent of other great brands. This gives them franchising power commensurate with their brand. If Rivian succeeds in its vertical integration and secures a top spot in the software-defined car space, long term shareholder returns will be considerable (Section 4.0).
However, presently negative cash flows threaten the strength of its balance sheet. Rivian may fail to vertically integrate before running out funds, and thus, the risk of bankruptcy is tangible (Section 5.0).
Full Deep Dive
1.0 Lessons From Auto History
In the auto space, vibes drive margins over time. Rivian has vibes.
It is well understood that Rivian is a Tesla competitor (to the extent that it can actually be called a competitor) and that it produces some rather attractive electric vehicles. However, looking at Rivian in that manner impedes a deeper level of understanding of the company.
The world creates idols only to destroy them. Why? Because the public is avid for emotion. - Enzo Ferrari
Tesla has indeed transformed the automotive landscape–but the fundamental desires of buyers have not changed. Although cars are increasingly defined by software, at an emotional level, we still buy them for the same reasons.
One cannot aspire to understand the auto industry without studying its history. As such, I deeply examined one of the biggest success stories in the automotive space: Ferrari. Ferrari was initially a racing organization and, today, is worth $50B+ despite the long-known unreliability of their cars.
The main takeaway from studying Ferrari is that, within reasonable limits, people buy cars because of the way they make them feel. But a car company needs more than vibes to succeed in the space, and Rivian seems to understand this as well as any, which we will get to in due time.
Before founding Ferrari, Enzo Ferrari founded Auto Avio Costruzioni in 1939 and made a name for himself in the racing space. It was not until almost a decade later that Ferrari as we know it today began mass-producing cars for the commercial market. In 1969, having faced some financial difficulties brought about the tumultuous nature of manufacturing, Enzo Ferrari chose to:
Focus on racing.
Delegate manufacturing to FIAT, which acquired 50% of Ferrari.
Since then, Ferrari has been a long-term compounder. The combination of Ferrari’s coolness and FIAT’s manufacturing prowess (which Ferrari kept in their 2016 spinoff) has taken the company a very long way. Ferraris are still known to suddenly burst in flames or not start when required, but people do not care.
Ferrari is undoubtedly a unique case, and I do not imply that everyone buys cars only because of the way they make them feel. But I do argue that this is a strong component of how car companies stay relevant: it gives them pricing power and thus–unless they allow manufacturing to slide–the cash flow to continue growing.
Ferrari chose to stick entirely to super sports cars, but brands like Porsche have done otherwise. The Porsche 911 was also borne on the currents of the racetrack (with plenty of ups and downs). Following the 911’s popularity, Porsche branched off into new models such as the Cayenne. People buy a Porsche for a multitude of reasons, but fundamentally because it makes them feel awesome at a somewhat affordable price.
Both Ferrari and Porsche flirted with bankruptcy in the past, but their vibes kept customers interested. Ferrari chose to drive pricing power via exclusivity, sticking with its traditional sports car, while Porsche has gone for an intermediate strategy (decreasing exclusivity, but driving scale). Ultimately, they both do one thing: make customers feel cool.
Over time, both also managed to produce their cars at satisfactory levels of quality and cost effectively, despite initial difficulties.
From studying Ferrari and Porsche we can thus infer that many people buy cars for some degree of self-actualization in addition to their basic survival instinct.
Rivian is like Patagonia for cars. In a time of growing consciousness about the health of the planet, this will only increase over time–just like Baby Boomers who have been captivated by the spirit of racing since the 1950s.
Racing will likely continue being relevant going forward, inspiring automotive consumers and OEMs (original equipment manufacturers) well into the future. However, this newfound ecological consciousness has carved a spot for a new source of inspiration.
Rivian´s R1 platform seems to have a franchising power reminiscent of the Porsche 911, but tailored for new generations. Something about it encapsulates the desire to be the eco-conscious bon vivant, heretofore polarizing personality traits that Rivian has merged to successfully mirror the modern man or woman.
The technology and clean sheet approach we've taken with the R1 product line has really enabled the uniquely differentiated product, the features, the attributes, the way the vehicle feels so special. And this is the result of thousands and thousands of tradeoffs we're making between different pieces of content, the way we think about design, the way we think about technology integrating with that design.
And then of course into R2 those same, that same mindset and that same ethos is being applied of course in a smaller form factor and a lower price point.
- RJ Scaringe, CEO and Founder during the Q2 2023 ER.
Nonetheless, while vibes are necessary, they’re not sufficient. Rivian has to bring costs under control, and it is difficult for me to understand why their gross margins are so significantly negative.
I can only guess that, with the automotive industry currently undergoing a tectonic shift, software is poised to radically modify the competitive landscape. Thus, discarding Rivian immediately for its poor financials at present is likely a mistake. While there is indeed no guarantee that the company will thrive, there is much to be learned in studying what the company is investing in exactly and whether the platform it is creating is structurally positioned for success in an auto industry that we cannot even fully wrap our heads around.
In the next section, I dissect this matter in depth.
2.0 Software and Rivian´s Margins
Rivian´s margins are deeply negative, but it may be for a good reason.
New generations not only care about sustainability, but also software, and cars are becoming defined by their software both for the benefit of consumers and OEMs. Software in next generation cars not only delivers a better experience for consumers, but also enables OEMs to operate more efficiently.
Software as the new foundation of the automotive industry enables mass data generation, which is then processed to generate insights. Insights can increase safety (autonomous driving) and reduce maintenance costs for both OEMs and customers.
Software defined cars become more valuable as the network in which they operate grows. As the network in question gets bigger, more data gets processed, which translates into better AI models and, thus, more valuable insights to drivers and OEMs.
Many things have changed. We no longer marvel at cars that can travel effortlessly uphill like we used to 50 years ago. This aspect has become commoditized. In the 20th century, economies of scale enabled FIAT and other industrial titans to make cars profitably and kept other players out for long periods. In the coming decades, network effects are going to do the same but the competitive dynamics will lead to less (and bigger) winners and far more losers.
We now value other attributes, for which consumers are willing to pay a premium. This locus of consumption is currently moving towards software, but only a small number of OEMs can succeed on this front. Rivian´s negative gross margins at present may be the price required to be relevant in this not so distant future.
Consider autonomy, for instance. The more cars that are connected to a network, sharing data about how best to drive in different scenarios, the better the driverless software gets. A network twice the size of another smaller network will not have a 2X better AI, but exponentially better AI. Whoever owns the largest network will deliver the best software, likely by a long shot.
[…] the more training data you have, the better the results.
Basically it’s sort of at 1 million training examples, it barely works; at 2 million, it slightly works; at 3 million, it’s like wow, okay, we’re seeing something, but then you get like 10 million training examples, it’s like -- it becomes incredible. - Elon Musk, Tesla CEO during Q2 2023.
Therefore going forward, while vibes and cost effective manufacturing are paramount, the name of the game for auto OEMs is:
Putting as many cars on the road as possible,
Generating and obtaining more and higher quality data than anyone else and ultimately
Training the best AI models/s possible.
This naturally involves having an infrastructure that permits the collection of data in the first place. Tesla is clearly the current leader in this sense (and others), and although Rivian is far behind, it seems to be far ahead of legacy OEMs that do not even have an infrastructure in place yet.
Further, releasing the most hardware into the market is futile if the platform is not abstracted sufficiently so that iterating on the software side is frictionless. For example, if deploying a new software feature the OEM in question requires calling the cars in and changing a few wires (while a competitor does not have to do that), the attendant OEM will lose. The competitor will iterate faster, please customers more, and create a bigger network.
Studying Rivian, it is evident to me that the company understands this and has invested heavily to:
Deploy a functional data infrastructure from the outset.
Abstract the hardware away.
They have geared the organization towards vertical integration (I will cover this in depth in the next section), with the intention of enabling top decile iteration pace across the board. Rivian has an air of clarity in this sense that is currently scarce in the auto industry.
Tesla´s gross margins have been positive ever since it began producing the Roadster, while Rivian´s are on their way up from the abyss.
It strikes me that in similar developmental years (2010-2012), Tesla had positive gross margins. Even post 2012, with the production of the Model S, Tesla continued to exhibit positive gross margins that, adjusted for price decreases, have only gotten better since–and especially so with the manufacturing leap that brought the Model 3 to life.
There are two ways to look at this scenario:
That Rivian´s negative gross margins are lower than Tesla´s at similar phases of development because the former is an incompetent organization (at least compared to Tesla).
That, because Rivian is late to the party, it has been forced to create an infrastructure capable of competing in tougher conditions (less time and thus, more capital intensive).
Let’s revisit Rivian’s operating margin now that we have covered network effects and the potential winner-/few winners-take-all scenario. If the evolving autonomy landscape ends up accommodating more than one player, and if Rivian´s negative margins are justified, then Rivian is currently a strong contender for position number two.
With over 300 million miles driven by Rivian vehicles, brand awareness continues to grow. - RJ Scaringe, Rivian founder and CEO, Q2 2023 shareholder letter.
These are two big ifs, but I believe they will be discernible in the coming years. In the affirmative, Rivian may occupy a privileged competitive position in the auto industry, with legacy players waking up to reality at the 11th hour.
In the next section, I will dissect Rivian´s vertical integration and where we can find this effort in the financials. A key part of the thesis, Rivian’s success with vertically integrating represents the most concise method of tracking the company´s progress going forward.
3.0 Rivian´s Vertical Integration
Rivian has a strong cost reduction track record now and there are some very promising developments in the pipeline.
The graph below will help you appreciate that, since Rivian went public (Q4 2021), the cost per car delivered has gone through two inflection points:
Q1 to Q2 2022: this inflection point seems to have been due to increased scale, allowing the company to spread out its costs over more vehicles.
Q1 to Q2 2023: according to management, this inflection point was due to to:
the introduction of Rivian´s proprietary Enduro engine. Previously, it was buying engines from Bosch.
the introduction of its LFP (lithium iron phosphate) battery pack.
another inflection due to scale, which has also enabled Rivian to leverage better prices with suppliers.
The more insightful of the two inflection points is the latter, since it involves the vertical integration of key components of the value chain. It also demonstrates Rivian´s ability and commitment to abstract away the hardware layer in the pursuit of frictionless iteration on the software side. In management´s words during the Q1 2023 ER conference call, the Enduro unit is “significantly less expensive”.
Just to be totally clear, when I say “abstracting hardware” away I am talking about creating a hardware stack that does not need to be modified with each software iteration. Thus, a totally abstracted hardware layer would support any software update, without having to call the cars in to modify the architecture.
The Bosch engines that Rivian was employing before introducing its Enduro unit were complex and used many semiconductors (chips). The chip shortage over the last few years created a notable bottleneck for Rivian. The new enduro engine is simpler and uses fewer semiconductors, thus helping to minimize dependencies.
In turn, the LFP battery is cheaper to make than “legacy” batteries made of cobalt and nickel. Further, lithium, iron, and phosphate seem to be abundant within the US, again reducing dependencies. LFP batteries are safer (lower risk of fires), charge faster, and have a longer life cycle, but they also have shorter ranges and less power.
Rivian plans to use this battery technology for all its cars, eventually.
Going into 2024, Rivian will continue reducing costs in two ways:
A next generation architecture, which is expected to reduce ECUs (electronic control units) by 60%, which in turn will reduce wiring length by 25%. ECUs are the tech that controls the functions of a car. A software-defined car is one in which the ECUs have been aggregated and the car is essentially controlled by one or two chips.
The introduction of the R2 model, which will include from scratch the new Enduro unit, the LFP batteries, and the next generation network architecture.
Indeed, reducing the number of ECUs is one of the key steps to abstract the hardware away. When fully aggregated, the car´s controls get totally virtualized and so software updates have no bottlenecks at the hardware level. In the Q2 2023 ER conference call, founder RJ Scaringe talks about how Rivian has been thinking about this since inception (and excuse the length of the quotes, but I thought I should add some evidence that justifies my ramble about abstracting hardware):
[….] broadly if we wind the clock back many years, we took the decision to develop all the ECUs across the vehicle and so typically, these ECUs would be coming from Tier 1 suppliers, and […] that mosaic of third party, Tier 1 sourced ECUs has to work together.
And it's a very cumbersome network architecture to work with and makes things like over the updates very, very difficult because you have to coordinate across multiple different companies on a software platform and software stack that, you as the manufacturer would know.
We built deep domain expertise in terms of our electronics development capability of course our software development capability. And that, of course, launched into the R1 product where we own the software stack, we own essentially all the ECUs in the vehicle.
So over time, we're consolidating the number of computers we have in the car to be significantly reduced. And next year that first step of that is a 60% reduction in the number of computers in the car relative today.
To that, Senior VP of Software Development Wassym Bensaid added:
I think really the core strength that we have is, we own every single computer on the vehicle. So we're able to not only update the infotainment or the connectivity, but we can update the vehicle controls, we can update the vehicle dynamics, the energy management, ADAS, the way the vehicle drives, the way the vehicle navigates, communicates with the entire world. We're able to create end-to-end unique experiences that really redefine the ownership with our customers and really create that regular connection with our community
Rivian has thus far been using Mobileye to pick up data on the road, and it is now planning to introduce its own proprietary sensor system. The system will be redundant until Rivian jettisons Mobileye. Scaringe wants Rivian to have uninterrupted and early access to sensor data.
The R2 will be revealed in 2024, but production will not start until 2026. Rivian is yet to produce the R1 profitably (although it expects it to be contribution margin-positive by 2024). The challenge that lies ahead is gargantuan.
Tesla as the free cash flow machine that we know and love today only arose after ramping up its manufacturing capabilities to bring the Model 3 to life. The R2 could be something like that for Rivian, but so far this remains speculative. Rivian has been producing cars for a very short time. It’s hard to truly grasp what lies under the hood just yet.
To cap off this section, I would like to mention that I have been eyeing Lucid, although not in depth quite yet. Per the graph at the top of this section, they seem to be following the same vertical integration strategy as Rivian (which is no doubt tracking that of Tesla´s). However, Lucid’s cars aren’t doing quite as well in the marketplace.
4.0 Potential Sources of Outsized Returns
Tesla and Rivian have the potential to each become a TSMC for the auto industry, although the latter less so for now.
For a long time in the semiconductor industry, companies did both the designing and the manufacturing. Notable examples include AMD and Intel. However, for an array of reasons–e.g. the benefits of specializing and the exponentially-increasing investments required to compete in cutting-edge manufacturing–AMD and others decided to stick to designing while letting others (TSMC) do the manufacturing.
Cars combined with software are exponentially more complex than the conventional mechanically-driven car. Having studied Tesla and Rivian, I notice that there is a large difference:
Between them
Between the two and legacy OEMs.
Their pace of iteration seems to be much faster than that of traditional OEMs, and they are carefully crafting their platforms to enable even faster iteration down the line.
It is likely that, over the coming decade, the auto industry will go the same way as the semiconductor industry: a lot of players that currently do both designing and manufacturing will have to revert to only the former.
Consider, at the limit, a face off between Tesla and Range Rover. The latter produces some beautiful and highly desirable cars, but it is no match whatsoever in terms of manufacturing efficiency and even less so when it comes to facilitating the data value chain required for its cars to successfully participate in a network, as described in Section 2.0.
How is Ferrari, for instance, going to bring about a competitive autonomous software if it only puts around 14,000 cars a year on the road?
There are a number of potential outcomes here, but I see the potential for Tesla and, potentially, for Rivian to branch off as the “fabs” of the auto industry. This would be one source of potentially outsized returns over the long run.
The second potential source of outsized returns, which is borne on the first, is licensing autonomy and the hardware stack required for it to work. Tesla has already expressed interest in doing so and Rivian may be able to do so as well–so long as there’s room for two.
5.0 Financials
The income statement is exhibiting signs of convergence, yet the company´s balance sheet and cash flows are a source of concern.
A glance at the above graph reveals that, indeed, Rivian´s operating leverage is increasing. Revenue is growing fast, tracking the company´s output, and gross profit and net income seem to have turned a corner since Q1 2023. It’s also noteworthy that Rivian has managed to print somewhat converging financials amid a period of inflationary pressures.
In Q1 2023, according to Rivian CFO Claire McDonough, the company saw lithium prices rise 115% thus far in FY2023. Further, rising rates have made cars far less affordable for consumers. Put in such context, the graph below takes on a different feel. Rivian is not doing all too bad, all things considered.
Further, Rivian´s deal with Amazon somewhat distorts the top line. In 2019, Amazon made a major investment in Rivian, for a 17% stake. The deal included an exclusivity clause, whereby Amazon was going to buy 100,000 commercial vehicles from Rivian by 2030. There are two issues with this deal:
We do not know how much of Rivian´s revenue depends on this deal.
Amazon has been somewhat under delivering on their side, buying less vans than expected.
This is a major additional source of uncertainty in the thesis.
On the other hand, the company´s liquidity–or lack thereof–is a source of concern. While it has an ample net cash position of more than $6B, Rivian is burning plenty of cash, as cash from operations came in at -$1.36B in Q2 2023. It is spectacular to see how Rivian has not gone out of business already, even though 9 years went by between its founding and the production of its first car.
Management clearly has excellent fundraising skills, but there is no telling when or if vertical integration will succeed. In Q1 2023, management guided for a CapEx of $2B through 2023, so Rivian needs to achieve positive margins at some point in the next few years else financial trouble awaits.
It’s good to see stock based compensation trending down, but as always, this is not a black and white subject. Per the Glassdoor reviews, it seems that employees are not too happy with Rivian´s austerity measures and I wonder whether this will have a negative impact in the pursuit of positive margins.
6.0 Conclusion
A clear addition to the watchlist.
Rivian is an impressive organization, advancing well in the face of a gargantuan challenge. Its cars have that X-factor, and customers love them. They have struck a nerve and now have the opportunity to franchise over the coming decades, just like Porsche has done.
Rivian´s negative margins add a great deal of uncertainty, but rapid financial convergence during tumultuous macroeconomic times says much of the company. A deep look into Rivian´s vertical integration reveals unique DNA as compared to what we are used to seeing in the auto space–one far more fitting for the times ahead.
Over the long term, if Rivian can surmount its manufacturing challenges and its R1 platform is indeed franchisable, shareholder returns can be excellent. Especially so if the autonomy space accommodates more than one winner, opening up the possibility for Rivian to become a fab for car brands.
Currently valued at approximately $19B (6.5 times sales), if Rivian goes the Porsche way, the upside is reasonable. If it goes the TSMC way (becoming a fab), the upside is perhaps exponential (Section 4.0).
Rivian´s balance sheet affords it a few years of liquidity. However, there is scarce visibility on when or if they will achieve positive margins, the risk of bankruptcy is tangible.
I will be tracking Rivian´s evolution in the coming quarters, looking for signs of continued vertical integration. While I am not convinced of starting a position here, I do acknowledge that this company is something special.
Until next time!
⚡ If you enjoyed the post, please feel free to share with friends, drop a like and leave me a comment.
You can also reach me at:
Twitter: @alc2022
LinkedIn: antoniolinaresc
I noticed the comment about Range Rover. I have a new Defender and have received 3 OTA updates which addressed everything from infotainment to chassis, engine and transmission updates - basically the full stack. I was impressed.
"How is Ferrari, for instance, going to bring about a competitive autonomous software if it only puts around 14,000 cars a year on the road?"
Well Ferrari is an experience and Luxury. It is made to feel. It will never be autonomous because their buyers would love to manually drive and feel