(Sorry for the long post, but I didn't know where to cut it down :D )
So far, I've just been reading along and enjoying all the great stock picks. Two months ago, I came across an article in the Handelsblatt about the auto industry and discovered Horizon Robotics
$9660 (+7,86%) , and since no one here has discussed this stock yet, I took the plunge and am eager to hear your thoughts. It’s definitely a gamble, but I’d appreciate your help in assessing it.
1. What does Horizon do: Horizon Robotics (founded in 2015, IPO in Hong Kong at the end of 2024) develops high-performance, energy-efficient AI chips (the so-called “Journey” series) and the corresponding software stacks for advanced driver-assistance systems (ADAS) and autonomous driving. In short: They’re building the brain for modern smart cars.
- The Technology—BPU (Brain Processor Unit): Standard computer chips (CPUs) are all-rounders. However, AI tasks in autonomous driving—such as simultaneously processing 12 camera streams, radar, and lidar data in real time—require extremely specialized computing power. Horizon Robotics has developed its own architecture for this purpose: the BPU (Brain Processing Unit).
- Hardware-Software Symbiosis: Unlike traditional chip designers, Horizon develops the BPU hardware and the autonomous driving algorithms in parallel. The chip is physically engineered to process the specific calculations for neural networks and modern foundation models (such as Transformer models for end-to-end driving) at lightning speed.
- Efficiency beats raw power: Many manufacturers primarily advertise astronomical TOPS (Tera Operations Per Second) figures. Horizon optimizes its BPUs (such as the current Nashand future Riemannarchitecture) for the highest frame rate (images processed per second) with minimal power consumption. This prevents the chips in the car from overheating or draining too much of the electric car’s range.
2.🚀 The “Journey 6” Generation: The Game-Changer
With the latest Journey 6series, Horizon Robotics has carved out a massive technological lead, securing the company market share even against giants like Nvidia.
- True SoC (System-on-a-Chip) integration: The Journey 6 series combines the CPU, GPU (graphics), MCU (security), and BPU (AI accelerator) on a single, extremely small piece of silicon. This reduces the space required on the circuit board by 50% and effectively turns the chip into a “plug-and-play” solution for the automotive industry.
- Extreme scalability: The Journey 6 is available in various configurations. From the small J6B for simple brake assist systems in affordable entry-level cars to the J6Pflagship model with a massive 560 TOPS of computing power for highly complex, urban autonomous driving. An automaker develops its software just once and then scales the hardware according to the vehicle’s price class.
- Development time cut in half: Thanks to this entirely new architecture and the accompanying open-source software tools, R&D efforts for automakers are reduced significantly. According to the company, the time to market for a new automotive system is reduced from the previous standard of 18 months to just 8 months .
3. Macro Trend: “ADAS” – Growth: The market for advanced driver-assistance systems is growing rapidly. In China, the ADAS market is projected to grow at an annual compound annual growth rate (CAGR) of over 12% by 2033; some global estimates for the software sector even range from 15% to 17% because end customers expect smart, self-learning cars. Level 2— and Level 3systems are currently evolving from a luxury feature to the absolute standard in the mass market (China is already significantly further ahead than Europe in terms of broad public acceptance). Here, Horizon Robotics benefits from “China Speed”: They develop and iterate significantly faster and more cost-effectively than Western competitors.
4. Partnerships:
The VW Roadmap: Volkswagen and its software subsidiary Cariad have, as part of their joint venture Carizon (together with Horizon Robotics), officially announced plans to develop Level 2++ systems specifically for the mass market. In official statements, the company confirmed that these systems will form the basis for the mass-market introduction of Level 3 functions.
BYD’s Pricing Strategy: BYD, too, is now equipping its volume models in the under-30,000-euro price range with complex ADAS packages (some powered by Horizon chips) as standard features in order to compete in the price war. This technology is simply a prerequisite for selling cars in China.
5. The Failure of the “Automated Drive Alliance” (ADA) Between VW and Bosch
Launched with great fanfare in early 2022, the ADA was intended to become the European counterpart to Tesla and Nvidia. Thousands of developers from Bosch and VW’s software subsidiary Cariad were to collaborate on developing Level 2 (driver assistance) and Level 3 (highly automated driving) systems for mass production.
- Why did the alliance fail?
- Cultural and structural obstacles: Two sluggish behemoths (corporate structures) attempted to develop agile software. Power struggles and differing software philosophies severely slowed down the project.
- Missed the “China Speed” train: While Bosch and VW argued over definitions in Europe, their Chinese competitors (BYD, Li Auto, Xiaomi) pulled ahead technologically, leaving them in the dust. VW dramatically lost ground in China—its most important market—when it came to “smart EVs.”
- For Horizon Robotics, this has three fundamental implications:
- Full maturity of the “Carizon” joint venture: VW has withdrawn resources from the failed Bosch project and is now expected to focus on its partnership with Horizon Robotics (in Asia) in the joint venture Carizon -> The situation in Europe remains unclear, as they are currently searching for a new partner there. Horizon is thus no longer just “one supplier among many,” but VW’s tech brain for the Asian market.
- Bosch has to go shopping: Bosch, now left without its major customer VW for this specific alliance, cannot afford to fall behind. Bosch is making a massive pivot and is now integrating the Journey 6platform from Horizon Robotics into its own systems in order to continue offering competitive products to global automakers at all. So Horizon wins on both sides.
- The “lock-in” effect for Western OEMs: The failure of the ADA shows other Western manufacturers (such as BMW, Stellantis, or Mercedes) just how risky and expensive in-house development can be. As a result, Horizon Robotics is establishing itself as the “plug-and-play” alternative for the global market.
6. Moat
Pillar 1: Extreme Switching Costs (The Lock-in Effect)
This is likely Horizon’s strongest moat. An automaker doesn’t just switch to a different chip supplier for its next model.
- Hardware-Software Integration: When an automaker (such as BYD or VW) builds its driver-assistance system based on the Journeyarchitecture and the SuperDrivesoftware from Horizon, the entire code, sensor alignment, and validation are tailored to it.
- Barrier to switching: Switching to a competitor such as Nvidia or Mobileye would mean years of new development and hundreds of millions in R&D costs. Once an automaker is in the Horizon ecosystem, it generally stays there for the next generation of vehicles as well.
- Joint Ventures: Through the Carizonjoint venture with VW, Horizon has even bought its way directly into an OEM’s corporate structure. A lock-in effect could hardly be any deeper.
Pillar 2: Regulation and Geopolitics (The Local Shield)
Normally, geopolitics is a risk, but in the Chinese market, it serves as a massive moat for Horizon.
- Data Sovereignty: Modern cars collect terabytes of video data. The Chinese government now rarely allows this highly sensitive geographic data to be processed by foreign companies or even taken out of the country.
- Localization Requirement: Western manufacturers (such as VW, BMW, and Audi) must collaborate with local players like Horizon in order to even get their cars on the road in China equipped with modern ADAS. Horizon benefits as a sort of “toll booth” for Western automakers in the Asian market.
Pillar 3: Algorithm-Hardware Symbiosis & Data Flywheel
In the field of machine learning, the rule is: whoever has the most real-world data wins.
- The data flywheel: Since Horizon holds a market share of nearly 48% among domestic brands, millions of cars are on the road equipped with its chips. These collect real-world driving data, the algorithms improve, the system becomes safer, which in turn attracts more automakers. A classic network effect.
- Co-optimization: Because Horizon develops not only the chip (BPU) but also the software (algorithms) itself, it can perfectly coordinate both components. This leads to extreme energy efficiency that traditional “general-purpose AI chips” often fail to achieve.
⚠️ Risks:
The moat is also under attack:
- The “in-house” threat: The biggest risk is that massive volume manufacturers like BYD will eventually say, “We’re big enough; we’ll develop our own chips now.” Tesla has shown the way. Although BYD recently postponed the release of its own chip, this remains a sword of Damocles.
- Nvidia’s dominance in the high-end market: While Horizon dominates the mass market, Nvidia (e.g., with the Thorchip) is still the gold standard in the ultra-luxury and robotaxi segments. Horizon must prove that it can consistently keep up technologically even at the very top of the market.
- China: I guess I don’t need to explain this further, do I?
7. Financial Review
This highlights the mixed reality of Horizon Robotics: it’s delivering on growth, but burning through cash when it comes to profitability.
Actual Value (FY/25) vs. Analyst Estimate
Revenue
3.76 billion RMB vs. ~3.65 billion RMB (+3.0% 🟢)
Gross Margin
64.5% vs. 62.0% (+2.5% 🟢)
EPS (loss per share)
-0.81 RMB vs. -0.37 RMB (-118% 🔴)
Net loss (total)
-10.47 billion RMB 🔴
Brief Analysis: While Horizon Robotics saw revenue (a strong increase of nearly 58% year-over-year) and gross margin (64.5% is outstanding for the automotive sector), the immense cash burn caught the market off guard. Earnings per share (EPS) missed analysts’ expectations by a staggering 118%, as research and development (R&D) costs for the latest chip generations have skyrocketed. It is precisely this stark “miss” on EPS that is the main reason for the weak stock performance this year—but perhaps it also presents an opportunity?
What do you think?
@Tenbagger2024
@Multibagger
@Raketentoni
@EpsEra
@Get_Rich_or_Die_Tryin
