New Street Research analyst Pierre Ferragu upgraded shares of Tesla from Hold to Buy and his price target is now $578, up from $400. He foresees Tesla generating $100B+ sales and $16 E/P by 2026. Based on the assumptions and forward P/E ratio of 75, he estimates Tesla could trade for $1,200 per share by 2025.
Tesla now trades for 116.28 times 2021 earnings estimate. For comparison, GM has a forward P/E ratio of 6.93 and Toyota’s forward P/E is 16.16. In 2019, Toyota sold more than 6.5 million vehicles, compared to about 0.37 million for Tesla. Yet, Tesla’s market value, at $362B, is almost double of Toyota’s market capitalisation of $185B.
Is this justifiable? If so, how? Maybe no one has a clear answer. While we can make any wild assumptions on the economic value, the penetration rate of EV and Tesla’s market share to justify any given valuation, one thing seems to be warranted. It is wrong to compare Tesla and other automakers. And, indeed, Tesla is very different from other vehicles on the road now in one essential aspect. Tesla is a computer while other cars are not, just like smartphones are computers while feature phones were not.
In Tesla, all the peripherals (e.g. sensors, motors, batteries, steering and so on) are directly connected to a powerful central SOC (system-on-a-chip) that comprises of neural processing units, GPU, CPU, ISP, DRAM controllers and so on, and are directly controlled by the SOC. On the other hand, other vehicles are still mainly highly skillfully assembled metals and wires, with fragmented electrical components (that, though, sometimes are computerised ECUs) scattered and embedded (and sometimes connected each other) all over the chassis. Most manufacturers are trying to consolidate and integrate these fragmented ECUs into an integrated ECU (like one in Tesla), but it will be years away. Thatis why many manufacturers are struggling with things like OTA strategies, “connected cars” and so on. On the other hand, Tesla, as a native computer, can handle anything that computers can handle without complex tricks.
Some time ago, Mercedes-Benz announced the alliance with NVIDIA for “Software-Defined Computing Architecture for Automated Driving”. This seems to be a right move, and the new architecture will be rolled out starting from … 2024. Other major manufacturers have similar timelines for their own integrated ECUs. One needs to note that this has been already implemented and rolled out across the fleet of Tesla. Tesla actually was using NVIDIA for the previous generation architecture and is currently on the third generation, which they claim 7 times more powerful than NVDIA’s chip. All the more importantly, Tesla can achieve those innovations by themselves while other manufacturers sourly lack these essential hardware/software skills.
More than 10 years ago, some people turned computers into telephones and feature phone industry perished and digital camera industry plummeted. We are not sure what will happen when computers are turned into automobiles. It might not be such a big thing, as most of the traditional car makers now seem to assume (maybe correctly). In this case, Tesla share might be in the huge bubble territory. Or, it might change the whole industry. In this case, the opportunity will go far beyond just automated driving and Tesla seems to be one of the best-positioned companies to leap forward in this uncharted market.
At the least, Tesla cars are now only computers running on the road, learning how to drive by themselves, with the trainers (called owners) who are willing to pay to train Tesla’s software. And several years worth of lead in these areas might not be easily reversible.
The downside for Tesla is big and clear, and the upside is elusive but huge. I will hold on to the existing long position for now.