When investors think about the semiconductor space, chipmakers like Texas Instruments and Nvidia typically come to mind. There's a good reason for that. Both are leaders in different verticals of the industry. But a growing number of companies are now designing their own chips, and those custom-built technologies can have a big impact on a business. For instance, Apple's M1 chips have supercharged its lineup of electronic devices, and Amazon's Graviton chips have made its data centers more efficient.
Building on that idea, Tesla (TSLA 0.75%) looks like a smart, albeit non-traditional, way to invest in the semiconductor industry. Here's why.
Tesla doesn't sell chips or provide foundry services to customers, but it does design its own chips for internal use. That includes the chip that powers its full self-driving (FSD) software and the D1 chip that will make its Dojo supercomputer one of the fastest artificial intelligence (AI) training machines in the world. But let's start with the core business: electric vehicles (EVs).
Last year, Tesla delivered over 936 thousand EVs, up 87%, capturing an industry-best 14.4% market share. Chinese automaker BYD ranked second with a 9.10% market share. Moreover, Tesla's manufacturing efficiency continued to improve as production increased at its factories in the U.S. and China. The company posted an industry-leading operating margin of 14.6% in the third quarter, and that figure ticked up to 14.7% in the fourth quarter.
Collectively, Tesla generated $53.8 billion in revenue in 2021, up 73%, and the company posted its 10th consecutive quarter of generally accepted accounting principles (GAAP) profitability, earning $4.90 per diluted share, up from $0.64 per diluted share in 2020. But in the years ahead, Tesla's proprietary semiconductors could take its electric car business into more profitable verticals.
Artificial intelligence and robotics
Let's start with self-driving cars. CEO Elon Musk believes Tesla could have a fully autonomous vehicle by the end of this year. That vehicle would be powered by its proprietary FSD chip -- a technology that, when released in 2019, was praised by some analysts as being six years ahead of the competition. The FSD chip is important for two reasons. First, Musk thinks FSD software will ultimately be the most important source of profitability for Tesla's EV business. Second, he also believes that Optimus -- an autonomous humanoid robot teased during the AI Day event last year -- could be more significant than the EV business over time. The FSD chip (or its next iteration) would be the AI engine that powers the Optimus robot.
Tesla's D1 chip, also introduced at the AI Day event, is the other half of the picture. The FSD chip will run the FSD software that makes Tesla's cars and robots autonomous, but the neural networks (AI algorithms) at the heart of that software must first be trained to understand the world and make decisions. With that in mind, the D1 chip was designed to make Tesla's Dojo supercomputer one of the fastest AI training machines in the world. That infrastructure will have a positive impact on Tesla's ability to build fully autonomous cars and robots, but Tesla also plans to use Dojo to offer neural network training as a service to other companies. In other words, the D1-powered Dojo supercomputer could help Tesla break into the cloud computing industry.
Here's the big picture: Tesla is a leader in the EV industry, but its semiconductor expertise could make it a leader in self-driving cars, autonomous robots, and AI cloud services. In turn, those high-margin verticals would supercharge its profitability. Case in point: ARK Invest believes autonomous ride-hailing platforms will generate $2 trillion in annual profits by 2030. That's why this under-the-radar semiconductor growth stock is a smart buy.