T.A. Eady - Comments
Technology Writer
Contributor's Links: Wondrous Robotic Futures

I write about the future of technology, with a focus on the roboticization of cars. For more of my writing, subscribe to my ... more

Latest Comments
Tesla's Executive Turnover Seems About Average (Updated)
2 years ago

I should also mention: in the Cappelli and Hamori study, the standard deviation was 3.7 years.

In this article: TSLA
Tesla's Executive Turnover Seems About Average (Updated)
2 years ago

I think Tesla is one of the most exciting technology companies in the world. The opportunity that gets me excited is autonomous ride-hailing, which could be a $10 trillion market by the early 2030s (https://ark-invest.com/research/self-driving-cars). Unlike any other company, Tesla’s production cars are collecting real world training data every day that Tesla’s neural networks are learning from. The volume of training data is most likely the main bottleneck to the development of fully self-driving cars. This is an area where Tesla can sail ahead of the competition because of its data advantage.

That’s the primary reason I’m invested in Tesla. It’s primarily about the immense opportunity of self-driving cars. A 10% market share by the early 2030s would propel Tesla to a $1.5 trillion valuation (at a 1.5x revenue multiple) and a ~30x return. If you’re willing to invest on a 15-year time horizon and take a risk on an uncertain technology, then I think Tesla is a really exciting company to invest in.

In this article: TSLA
What Jim Chanos Gets Wrong About Tesla
2 years ago

Here is what Elon Musk has said about collect data on the safety of autonomous cars:

“Even once the software is highly refined and far better than the average human driver, there will still be a significant time gap, varying widely by jurisdiction, before true self-driving is approved by regulators. We expect that worldwide regulatory approval will require something on the order of 6 billion miles (10 billion km). Current fleet learning is happening at just over 3 million miles (5 million km) per day.”

www.tesla.com/.../master-plan-part-deux

Tesla plans to collect billions of miles of data in “shadow mode”: https://ytcropper.com/cropped/5S5b3e33e7d93c5

In this article: TSLA
What Jim Chanos Gets Wrong About Tesla
2 years ago

Great question. Thank you. I agree with Chanos that it is dangerous for a company like Tesla to be reliant on the capital markets. This risks a situation where perception matters more than reality. Regardless of the reality, if perception is negative, then capital markets might not supply the needed cash. Perception is largely outside Tesla’s control, and many actors have a vested interested in turning perception negative, or have a grudge or emotional animus against the company. Human psychological bias and systemic media bias are working against Tesla here, too: medium.com/.../the-truth-about-tesla-c4b6d7e43ad7

So, the faster Tesla gets to GAAP net profit and positive free cash flow, the better. The company’s fate will then be decided by reality, not perception. I believe in the technology, so I think if Tesla can be left to develop and deploy its technology over many years, it will be an incredible success.

In this article: TSLA
What Jim Chanos Gets Wrong About Tesla
2 years ago

Hi Brian. Tesla does collect images from the production vehicles’ cameras. This presentation by Tesla’s Director of AI shows some of the images Tesla has collected and how it goes about labelling them: https://vimeo.com/272696002

Some Tesla customers have found their cars uploading gigabytes of data to Tesla’s servers: electrek.co/.../tesla-autopilot-data-floodgates/

The cars are not recording video all the time. They may collect short video clips at random moments, or collect video when certain conditions are met (e.g. a construction zone is detected) and that triggers a recording.

Tesla wants to make a custom ASIC (en....pedia.org/.../Application-specific_integrated_circuit) to run deep neural networks. Currently, I believe it uses Nvidia GPUs. Designing a purpose-built chip makes sense because it allows for potentially much greater efficiency than a GPU. Tesla also believes that vertically integrating as many components of the self-driving system as possible will lead to greater efficiency.

In this article: TSLA
What Jim Chanos Gets Wrong About Tesla
2 years ago

To build public trust in autonomous vehicles, I think there needs to be data transparency and scientific, statistical evaluation of the data. We need to prove that autonomous vehicles are safer before we deploy them at large scale. It is natural and human to have doubts about handing over control to a computer, and I think only robust statistical evidence can put those doubts to rest.

In this article: TSLA
Tesla's Competitive Advantages: Direct Sales, Data, And Software
2 years ago

Note: it should say "Disclosure: I am long TSLA", not "Disclaimer".

In this article: TSLA
The Tesla Model 3 Vs. The Porsche 911 And Other Beautiful Cars
2 years ago

The new Roadster will no doubt be a head-turner!

In this article: TSLA
1 to 8 of 8 comments