On the positive end, a new AI chip design and updates on autonomous driving capabilities encouraged investors. However, a typically ambitious presentation by CEO Elon Musk that kicked off with a bizarre dancing “robot," proclamations on universal basic income, and Tesla’s plan to produce a 125-pound humanoid robot provoked some skepticism.
Amid the uncertainty over how to assess the presentation, shares have remained close to flat in premarket trading Friday as many ponder the possibilities that Tesla could be nearing new breakthroughs, while also considering the possibility that another flashy presentation is just that - all flash.
A History of Overly High Hopes
Of course, unusual and perhaps overambitious presentations by Tesla are nothing new.
Musk has famously overestimated his abilities, promising everything from solar shingles to be produced by his most dubious acquisition, Solar City, to the rollout of both semi and cybertruck fleets, to a rocket-assisted roadster. Yet, perhaps most famous was the promise of a fully autonomous fleet of “robotaxis."
“By the middle of next year, we'll have over 1 million Tesla cars on the road with full Self-Driving hardware, complete at a reliability level that we would consider that no one needs to pay attention,” Musk said in 2019. “Meaning you could go to sleep in your - from our standpoint, if you fast-forward a year, maybe a year, maybe a year and three months, but next year, for sure, we'll have over 1 million robo taxis on the road.”
For sure, there were 1 million less robotaxis on the road in 2020 than promised by Musk in the year prior.
In fact, the comment on sleeping in your car could be quite a problem for Tesla legally as the National Highway Traffic Safety Administration opens an investigation into crashes that Tesla’s full self-driving (FSD) technology potentially contributed to, while German regulators rail against false advertising on this front.
In terms of this liability, the lofty robotics aims communicated by Musk at present could prove problematic in the same way false promises about robotaxis and FSD are at the present moment.
The State of the Robotics Race
However, even if the robot is not simply an overly optimistic pronouncement, the push toward leading in the crowded robotics space opens other enormous issues.
While Musk conjectured that Tesla could be considered the “world’s largest robotics company," his company is in reality not even the largest automaker in the robotics space.
South Korean automaker Hyundai might be the biggest hurdle after announcing the completion of its $1.1 billion acquisition of Boston Dynamics earlier this summer. Very much in contrast to Musk’s comical presentation, the capabilities of the company’s robots are almost frightening.
“This transaction will unite capabilities of Hyundai Motor Group and Boston Dynamics to spearhead innovation in future mobility,” Hyundai Chairman Euisun Chung said of the transaction. “The synergies created by our union offer exciting new pathways for our companies to realize our goal - providing free and safe movement and a higher plane of life experiences for humanity.”
Meanwhile, Japanese stalwarts Honda (HMC) - Get Honda Motor Co., Ltd. Sponsored ADR Report and Toyota (TM) - Get Toyota Motor Corp. Sponsored ADR Report have been toiling in the robotics space, especially in humanoid robots, for decades. Memorably, Honda’s recently retired ASIMO was the first robot to walk bipedally and also the first to climb stairs, becoming nimble enough over its development to play soccer with a U.S. president.
Musk joked that the codename for the new Tesla bot is “Optimus Subprime." Clearly a joke, referencing the leader of the autobots faction in the beloved "Transformers," the codename might also unwittingly allude to Tesla’s late start in challenging fellow automakers that are far farther along in the development of these technologies.
Perhaps more aptly put, suboptimal.
Chips on the Table
To be sure, there was actual hardware that could prove highly valuable to Tesla’s future. This was found most prominently in the new, proprietary chip for training artificial-intelligence networks in data centers unveiled by Ganesh Venkataramanan, senior director of Autopilot hardware.
According to Venkataramanan, the new D1 chip, which will be crucial within Tesla’s Dojo supercomputer system, uses a 7nm process, with 362 teraFLOPS (floating point operations per second) processing performance. For reference, Sony’s SNE PlayStation 5 sports 10 teraFLOPS while Nvidia’s Titan V GPUs tout 110 teraFLOPs of power.
Clearly, the breakthrough for in-house chip design is big news as vertical integration of company technology has long been a key underlying idea for Tesla’s biggest bulls.
Comparing Tesla to Apple (AAPL) - Get Apple Inc. (AAPL) Report in a research note upon Tesla’s 2019 FSD computer announcement, ARK Invest analyst James Wang proclaimed that vertical integration would allow Tesla to “out-execute and out-innovate its competitors." He noted his belief that in-house chip design could help Tesla leapfrog its competition by as many as four years by streamlining design for its cars’ specific needs.
“Instead of waiting for Nvidia’s next gen processors, Tesla hired chip designers and built its own,” Wang added. “Tesla’s processor isn’t better than Nvidia’s - it’s just better for its own workload.”
Still, there is ample room for caution on this bullish factor yet. When pushed during the Q&A section on the potential issues for the Dojo supercomputer, a Tesla engineer explained that not everything has been sorted out, but that he remains confident the team is “on the path” to solving remaining issues.
Ever the optimist, Musk added that the rollout is likely due next year. For investors bitten by over-optimism rather frequently in recent years, a modicum of skepticism on the claims might be worthwhile.