🏢 🚙 🤖 The Physical World Tech Newsletter

By Sam Cash // Physical World Technologies Newsletter

🏢 🚙 🤖 Issue #43; tesla bull / bear analysis, atomisation of work and too many unicorns





Subscribe to our newsletter

By subscribing, you agree with Revue’s Terms of Service and Privacy Policy and understand that 🏢 🚙 🤖 The Physical World Tech Newsletter will receive your email address.

🏢 🚙 🤖 Issue #43; tesla bull / bear analysis, atomisation of work and too many unicorns
Welcome to my newsletter, where I discuss thoughts and news on the intersection of the built world and technology #retail #mobility #realestate #tech
Please contribute to the community by forwarding this to someone who might enjoy it also. 🙏🏼
You can follow me on Twitter here

Tesla: bull or bear?
The following startup adage feels true of Tesla today as it ever has:
“The battle between every startup and incumbent comes down to whether the startup gets distribution before the incumbent gets innovation”
Tesla has been running fast to capture the value and growth of electric vehicles. There are innumerable complex challenges of building an electric vehicle - not only do you have nail production of a normal vehicle, something OEMs have been doing for decades, but you also need to layer in hardware, software, battery technology, electric drive trains and sensor fusion.
The crux of the bull v bear argument here is; bulls argue that Tesla is uniquely positioned to build a defensible tech stack, which current OEMs are not native to, bears argue that Tesla is unable to compete with existing manufacturers on the table stakes of car production and delivery. Let’s explore both sides of the argument;
Full stack:
  • One of the primary arguments for Tesla is how vertically integrated their vehicles are from point of purchase to custom software, hardware, charging network, to delivery and servicing
  • We’ve seen a large number of fully vertically integrated companies, such as Apple, build strong moats around proprietary hardware + software;
  • Importantly, vertical integration has allowed Tesla to move away from selling their vehicles based on “range” but to concentrate consumers on the whole value proposition and drive lock-in. This is disruption; changing the basis of competition
Data aggregation
  • To date, much of the race for Level 4/5 autonomy has been about aggregating high-quality, variable data to parse through machine learning models
  • Musk made an early strategic decision to fit cars with more cost effective computer vision, instead of the AV industry standard LIDAR, in order to be able to aggregate data IRL
  • Its believed that Tesla has over 1bn miles of real-world, predominantly human-driven data versus Waymo’s 10m+ of autonomous data - importantly in many valuable real world edge cases
Software + Autopilot:
  • whilst software progress tends to be less exponential than hardware (remember Moore’s Law), Tesla has been providing a superior experience and lock-in through software
  • Car complexity will shift from mechanical complexity, to software and hardware complexity. Retrofitting an existing vehicle with software is arguably more difficult than building a modular vehicle around a software stack. Cars today have about 50m lines of code, increasing to 600m lines of code in the future
  • Tesla has re-codified the car stack - allowing vehicles to be updated and repaired over-the-air, near instantly. Future autos will run on a central control board + operating system, though its unlikely there will be a predominany OS for the majority of vehicles due to low fragmentation in the auto industry
  • Tesla’s Autopilot runs effectively at Level 2 autonomy. This level of autonomy far outstrips the ADAS systems currently available in most high-end luxury autos
  • Autonomous continues to be Musk’s stated future value driver for the company - in April he announced that Tesla would deploy a fully-autonomous ridesharing network as early as Q2 2020
  • To put a further point on this, Tesla has 100,000s of vehicles with sensors collecting real world data in both autopilot and driven by humans
Autonomous Hardware
  • Tesla recently announced a proprietary computer, powered by two system on chips (SoC) armed with a CPU and GPU, which they’ve aplty named FSD (full self driving)
  • This new hardware is coupled with an accelerated neural net, and is capable of processing at a frame rate a quantum higher than NVIDIA’s Drive Xavier chips
  • If these claims shake out, this is a leapfrog on existing hardware currently used by competitors and could give the company a partly defensible position until incumbents catch up
  • One of the gating factors for EVs has been range, due to costly and heavy lithium-ion batteries
  • Power increases for these batteries have been incremental, though cost curve declines have accelerated, especially for Tesla
  • Tesla batteries, produced with Panasonic, produce more power at a lower price point - this has allowed the company more palatable unit economics and a considerable development lead on competition
  • This is a hard one to quantify as its partially intangible but its worth noting that Tesla have benefited from a software-first approach to car production whilst not incurring the technical debt of legacy vehicles, unlike OEMs
  • This has allowed them to produce a customer experience a quantum better than incumbents - from point of purchase to delivery and in-car experience
  • Whilst the trend for electric vehicles is strong, the trend for car ownership far less so. With the unbundling of the car, especially within cities, and a slew of new low-friction transportation options driven by smartphone usage, we’re at or near “peak car” ownership
  • Auto loan defaults are at record highs - meaning that even with a lower price point vehicle in the Model 3 and government subsidies, Tesla is likely to continue to find it hard to find new customers
  • Part of the growth narrative has been staked on China, where the company are building a production (Giga)factory. Tesla on a unit basis is being outsold by local players such as BYD, currently the largest global manufacturer of electric vehicles - they sell 30k units a month v Tesla’s ≈10k monthly units
Musk’s management style and claims
  • Whilst there are too many fictitious claims to note, the zenyph of these claims might in the now infamous tweet of having raised private financing at $420 (Bloomberg have put together a claim tracker)
  • Musk has consistently made hasty and borderline ficticious announcements about product releases, in and around, earnings calls to partly divert the narrative from financials to product
Supply Chain and Manufacturing:
  • Tesla has been consistently stretching payables to vendors - as at end of 2018 the company had $5bn in AP liabilities
  • This friction between Tesla and its vendors continues to excarbate the company’s inability to manufacture vehicles in line with guidance, now dubbed “manufacturing hell”
Autopilot / Autonomous
  • Tesla’s Level 2 autonomous system has been implicated in a third fatal crash recently - calling into question the safety and viability of their software and their timeline to autonomy
Management revolving door
  • There has been an exodus of talent leaving the company over the last year - over 90 executives have left the company since the start of 2018
Unit economics / financials
  • Currently Tesla makes an EBIT loss of about $3,500 per on its $35,000 Model 3
  • The company is in a near continuous cycle of equity and debt sales to keep the company afloat. They recently raised $2.7bn after having burnt through $1.5bn in Q4 2018
  • With OEMs such as Audi, Volvo, Nissan, - at best they will see margin erosion, at worst they will see loss of market share of EVs
Whilst there are no doubt other significant elements to add to both sides of the argument - these tend to be the crux of the focus for the majority of speculators.
(Other) News
Venture / Startups
There are currently 452 “unicorn” tech companies, this has been driven by companies staying private longer and more late-stage venture capital, hopefully this dislocation between price and value will see a reversion to sensible valuations (thinking about writing a longer piece on this) // this pricing drift is also not being “bought” by Wall Street/public markets (WSJ paywall), typically where VCs liquidate their positions, as recently witnessed by meh listings of Uber, Lyft, Pinterest etc. // The fundamentals of product-market fit, a great primer by Holloway Guides // The New Yorker looks at the IndieWeb movement of owning your own servers/data, a partial remedy to misappropriation of data by platforms // A fantastic deep dive by a16z on investing in the podcast ecosystem // betaworks’s John Borthwick breaks down his thoughts on Zuckerberg’s podcast - he highlights how Zuck’s mental models might have lead to blind spots in data privacy
Future of Work
Awesome write up by Alex Danco on “cooking as a service” (this is must read in my view); in my view there is a wider trend of atomisation of work as technology has removed friction from internal and external coordination of firms, giving rise to new services and work modes // Lightspeed’s Alex Taussig writes about the “business in a box” - how horizontal platforms are providing individuals, solo-preneurs with turnkey economic opportunities // “The AI gig economy” how perceived “AI” is actually a mechanical turks powered by gig workers, to clarify this is not the world we want to live in
You can now use NYC’s Citibike through the Lyft app. Last year the company acquired Motivate, the operators of a number of urban docked bike sharing schemes. It’s important not to understand how important this is for Lyft as ride sharing platforms look to go multi modal and building a superior full stack offering in a city as large as NYC // Speaking of which, VOI just announced a bike sharing partnership with the city of Stockhom // Nissan is now using computer vision for its autonomous program instead of Lidar, partial validation for Mr Musk who has claimed Lidar is a defunct technology // Uber released their Q1 ‘19 earnings, the first as a public company, spoiler alert: it wasn’t good, still though the stock is up (!)
Thanks for reading!
Did you enjoy this issue?
Sam Cash // Physical World Technologies Newsletter

The intersection of the physical world and technology; with a focus on future mobility,real estate, retail and cities.

In order to unsubscribe, click here.
If you were forwarded this newsletter and you like it, you can subscribe here.
Powered by Revue