Tesla glory/failure

[Archimid]

Please watch the video at 2:16:44 for the best argument, with evidence, that lidar is not needed for full self driving.

[BeeKnees]

https://www.marketwatch.com/story/nvidia-says-tesla-inaccurate-in-self-driving-comparison-2019-04-22

I'm not so sure tesla is wrong.
The pegasus is a board, not a chip.  It holds 2 Xavier chips and two GPUs, plus it uses 4 times the power. 

https://wccftech.com/nvidia-unveils-drive-px-pegasus-first-board-capable-level-5-autonomy-320-dl-tops-post-volta-next-generation-dgpus-500w-tdp/

[Sigmetnow]

Here is video of the full presentation:
Elon Musk (@elonmusk) 4/22/19, 8:10 PM
Tesla Full Self-Driving
 ##https://www.youtube.com/Ucp0TTmvqOE?start=4177
https://twitter.com/elonmusk/status/1120479950828703745
- Warning, this presentation is 4 hrs & 20 mins long
- All cars made since Oct 2016 either have the hardware needed for FSD or are trivially upgradeable
Start at 1:09:37 in.

Elon Musk (@elonmusk) 4/22/19, 9:41 PM
Full Self-Driving video
 ##https://www.youtube.com/tlThdr3O5Qo
https://twitter.com/elonmusk/status/1120502923136520192
2-min time lapse. Lots of highway interchanges.  A number of stop lights and stop signs; you can see them displayed on the screen.
Of all the autonomous rides given that day, I have heard of only one that required intervention, to help get the car into the left turn lane at one intersection.

Hamid Shojaee (@hamids) 4/22/19, 5:56 PM
One more... #AutonomyDay
https://twitter.com/hamids/status/1120446405410205701
Easy to see progress in the ~1 minute video clip at the link: Summon is much quicker, like a human driver!  Attendees described using the Tesla Network app to “Click on a car shown in the parking lot display, and it will pull out of its parking spot and drive to you.”

Tesla4Everman (@Everman) 4/22/19, 7:16 PM
January 15, 2015. @elonmusk,"We will have mainstream cars capable of full autonomy in five years or less. Proving that it is safe & getting regulatory approval is likely to take 2-3 years after that." #Tesla #FSD
 
Thought insane by all then. He'll beat it.
https://twitter.com/everman/status/1120466365834514432
Video of that 2015 press conference, cued to that quote, at the link.
 ##https://www.youtube.com/watch?v=uhDpDNLRVd0  @ 19:35

[Sigmetnow]

See the video beginning 1:17:10 where they describe the system’s two fully independent computers, including power supplies and video feeds.  Musk: any part of this can fail, and the car would keep driving. The probability of this system failing is less likely than a person losing consciousness.  By at least an order of magnitude.

##https://www.youtube.com/Ucp0TTmvqOE

[Sigmetnow]

System on a Chip (SoC)

James Wang (@jwangARK) 4/22/19, 7:11 PM
Here's a comparison of Tesla's new self-driving chip vs. Nvidia's Xavier SoC. Tesla is ~2.5x in perf/watt, ~3.3x when adjusted for die area (proxy for cost). Tesla's real world performance will likely be much higher due to tight coupling of sw+hw+compiler.
https://twitter.com/jwangark/status/1120465245414285314
Image below.

[Sigmetnow]

Tesla unveils car design without a steering wheel, coming within 2 years, says Elon Musk

Unlike most other companies working on self-driving systems, Tesla is taking the approach of delivering vehicles to customers that they can use right now as regular cars but with the hardware required to eventually achieve full self-driving with future software updates.

It is completely different from companies like Waymo or Uber who are working on vehicles solely built for autonomous ride-sharing services.

GM also claims that it is bringing an autonomous Bolt EV without a steering wheel to production by the end of the year.

At the Autonomy Day, Tesla unveiled its own design for a vehicle without a steering wheel....

https://electrek.co/2019/04/23/tesla-design-without-steering-wheel-elon-musk/

Image posted above: https://forum.arctic-sea-ice.net/index.php/topic,2406.msg196425.html#msg196425

[NeilT]

Nvidia Corp. said in a statement Monday that Tesla Inc. was wrong in describing its self-driving computer as more powerful than Nvidia's, which Tesla used until it began developing its own chip.

"Tesla was inaccurate in comparing its Full Self Driving computer at 144 TOPS of processing with Nvidia Drive Xavier at 21 TOPS," a spokesman said in an email. "The correct comparison would have been against Nvidia's full self-driving computer, Nvidia Drive AGX Pegasus, which delivers 320 TOPS for AI perception, localization and path planning."

The statement also contends that "while Xavier delivers 30 TOPS of processing, Tesla erroneously stated that it delivers 21 TOPS. Moreover, a system with a single Xavier processor is designed for assisted driving AutoPilot features, not full self-driving."

https://www.marketwatch.com/story/nvidia-says-tesla-inaccurate-in-self-driving-comparison-2019-04-22

And the SEC were listening.

Apologise up front for the length, but this is highly technical in nature and will not fit into a few words; even with the links.


Actually the Wiki page on Xavier gives 20 TOPS for a single core.

If you do the extrapolation, 7 Xavier cores give the same power as the Tesla chip but consumes  200W to do it.  The Tesla chip is 75W, only 15W more than the Tesla Autonomous 2.5 hardware, but with 21 times the compute power.

Nvidia states that Pegasus gives 2.2 times the Tesla compute power with 2 Xavier cores.  This is true.  However it also consumes 6.6 times the power.  In case nobody was listening, the presentation explicitly stated that the Tesla computer environment was a heat constrained environment.  In other words you cannot just increase the compute power by bumping up the power input, because more power consumed == more heat emitted.

There is also a specific piece of information in the presentation which is critical to the operation of the system.  Batch size.  It was stated that the batch size of Google hardware is 256.  In other words the Google processing environment waits until it has 256 operations before executing.  The Tesla is a batch size of One.

What does this mean?  In pure TOPS measuring the Tesla processor is significantly slower than competitors which batch to a higher value.  However in an Autonomous driving scenario, Tesla outperforms all rivals which use a higher batch value because it responds on even one operation required.

There is an aspiration to increase performance by 3* in the next iteration.  This might seem to be a difficult goal until you realise that the hardware is based on 14nm.  Current mobile processors are on 7nm today.  14nm was, almost certainly, chosen for reasons of cost as 7nm is cutting edge.  But in 2-3 years time 7nm will be mainstream.

Tesla will achieve its speed improvement with the same thermal level, or even less.

Nvidia is not even in the same country, let alone the same ballpark.

If you understand what was said in that presentation, you realise that to use Nvidia Tesla must:

Use the Nvidia compilers
Run code the way Nvidia want them to
Be constrained to the way Nvidia want them to execute code and use their processors.

The advantages of ASIC's is that they are Application Specific (Application Specific Integrated Circuit).  Meaning that the processor has been specifically designed to run Automotive AI functions.  Unlike Nvidia who use Graphics processing CUDA cores and force them into the mould of AI.

Tesla didn't just create a new processor.  They created the compilers for it, the language for it and they did the reverse engineering to allow code written for Nvidia chips to be recompiled to run on the new Tesla chip.  This is almost certainly inefficient and significantly less efficient than the Fully Autonomous Self Drive Tesla code which will have been specifically written for the Tesla chip.

Whilst it is not specifically stated, each Operation on the Tesla chip, with Tesla software designed for it, will be more efficient than the same code written for the Nvidia chip.  Hence why Tesla states Images processed as well as TOPS, along with the relative number of images processed by the prior system.  Notably I can't find any data for Images processed for the 320 TOPS Pegasus Nvidia chip.

The Tesla processor is almost certainly processing images at a higher rate than the same software on the Nvidia hardaware.  When Tesla optimise the software and triple the performance, that will increase this differential and make the Tesla chip even faster than the Tesla processor for the equivalent TOPS.  All for the same 75w.  With the current state of Nvidia hardware that would take 675W. 

Think that through and for every hour you drive the Nvidia Drive PX Pegasus, Self Drive is stealing 0.5KWH of your battery.  Or put another way, after 10 hours of driving with the current Nvidia Pegasus (100kwh/340 miles), you have lost 17 miles of driving compared to the Tesla equivalent of 2.5 miles.

Take that same Tesla future 3* extension, to the current Nvidia technology and you are losing around 23 miles of range for the same 10 hours driven.  Think I'm talking BS?  Look back at that Nvidia Drive Wiki Page and you will see that Nvidia are making almost no reduction in power consumption per hardware iteration, for a given TOPS value.  Tesla, on the other hand, is at 1/3 the power on the first iteration.  On the second it will be around 1/6th unless Nvidia makes a revolution in the way they produce their chips; which is not likely because Nvidia has multiple customers and they would all have to do a complete software refresh to gain the benefits.

Musk is not lying when he says that the Tesla processor is the best self drive processor in the world today.  What he is not saying is that only Tesla can use it and that is the main reason it is so much better.  In this market vertical integration beats horizontal every day of the week.

[Sigmetnow]

< If you're already confident about this new chip being plenty powerful enough for FSD, why are you working on a new version? Is FSD not the end goal?
   
<< gotta chase those 9's
   
Elon Musk (@elonmusk) 4/22/19, 8:26 PM
Exactly
https://twitter.com/elonmusk/status/1120483948906332161

Andrej Karpathy concludes: If I were to summarize my entire talk in one slide it would be this. And that includes incidents of flying cars (the rightmost image).

[Sigmetnow]

Tesla (@Tesla)4/23/19, 10:12 AM
Tesla Vision uses path prediction to accurately predict how a road will extend, even when it can’t see around the corner
https://twitter.com/tesla/status/1120691857111916544
Image below. 30-second video at the link.

[Sigmetnow]

“You all used your own neural network in your brains to get here. You didn’t shoot lasers from your eyes to drive.”
- Andrej Karpathy

If Tesla’s non-LiDAR FSD system succeeds, it will disrupt many companies, industries and investors.  Which explains the big commercial push-back on Tesla’s acheivements, in order to protect their own financial interests.

SunTrust's Willian Stein
"We are concurrently skeptical and hopeful about TSLA's claims. If TSLA executes to plan, implications across semis/components are meaningful: positive for analog/mixed signal vendors, mixed for digital vendors, negative for LiDAR vendors.... Investors should recognize that, if the company achieves its autonomous driving goals, combining this with its already-achieved EV technology conspire to establish a ride hailing service that could be quite financially compelling to both the car owner and the rider."

Tesla’s Autonomy Day highlights Elon Musk’s vision for a full self-driving future without LiDAR
https://www.teslarati.com/tesla-elon-musk-full-self-driving-autonomy-day-lidar/


Edit:  Also keep in mind the exponential speed of Artificial Intelligence improvement.

Vala Afshar (@ValaAfshar) 4/20/19, 9:05 PM
AlphaGo initially trained on thousands of human games to learn how to play Go.

AlphaGo Zero skips this step and learned to play simply by playing games against itself; it quickly surpassed human level of play, defeating AlphaGo by 100 games to 0.
https://deepmind.com/blog/alphago-zero-learning-scratch/

https://twitter.com/valaafshar/status/1119769133929979906

[Sigmetnow]

Tesla (@Tesla) 4/23/19, 6:24 PM
This 3D reconstruction shows the immense amount of depth information a Tesla can collect from just a few seconds of video from the vehicle's 8 cameras

https://twitter.com/tesla/status/1120815737654767616
25-second vid at the link:  3-seconds of city street video; then a pan around the 3D street scene.

Alter Viggo (@AlterViggo) 4/23/19, 12:08 PM
Tesla can generate 3D point clouds from their cameras. No LiDAR necessary. This flies in the face of those like Waymo, Uber, and Cruise who say you need (expensive and power hungry) LiDAR to make 3D maps. ...

https://twitter.com/alterviggo/status/1120721213922402304

[Archimid]

The Longest-Range Electric Vehicle Now Goes Even Farther

https://www.tesla.com/blog/longest-range-electric-vehicle-now-goes-even-farther

Beginning today, Model S and Model X now come with an all-new drivetrain design that increases each vehicle’s range substantially, achieving a landmark 370 miles and 325 miles on the EPA cycle for Model S and Model X Long Range, respectively. Using the same 100 kWh battery pack, these design and architecture updates will allow drivers to travel farther than ever before, charging less frequently and getting more range out of every dollar spent on charging.

[Sigmetnow]

The Longest-Range Electric Vehicle Now Goes Even Farther

https://www.tesla.com/blog/longest-range-electric-vehicle-now-goes-even-farther

Beginning today, Model S and Model X now come with an all-new drivetrain design that increases each vehicle’s range substantially, achieving a landmark 370 miles and 325 miles on the EPA cycle for Model S and Model X Long Range, respectively. Using the same 100 kWh battery pack, these design and architecture updates will allow drivers to travel farther than ever before, charging less frequently and getting more range out of every dollar spent on charging.

My Hyundai would still beat them both to New York from LA by several hours. Assuming the Teslas didn't spontaneously explode into fire at some point.

I'll still have ~ $60,000 in my bank account which the Tesla owners won't. So I am unsure why such a new development will lead to Tesla glory (or survival).

Because Hyundais do spontaneously explode into fire.
https://forum.arctic-sea-ice.net/index.php/topic,2406.msg196362.html#msg196362

And Tesla Autopilot makes long trips more pleasant and much less fatiguing.  And some people would rather not pollute every mile of the countryside they drive through, nor expose themselves and their kids to poisonous fumes every time they stop at petrol stations.

Edit:  Oh, and Tesla Models S, X and 3 are statistically the safest, least likely to result in injury or death in a crash, of any car measured by NHTSA.

[Sigmetnow]

EXCLUSIVE: 2019 Tesla Model S Review: From SF to LA on One Charge? - MotorTrend

...Surprisingly, the nav system plots our route down the 101, over the steep Highway 152 pass, and dumps us onto Interstate 5, with its famous obstacle, the Grapevine, which is basically a mountain where it sometimes snows at its summit in the winter. "Is this really the best way?" I ask, thinking that following Highway 101 all the way along the coast, with its relatively gentler grades, would be less of a range challenge. "Just follow the map," we're told.
...
As we pull into the Supercharger stall, our elapsed time from the Bay Area stood at 6 hours, 11 minutes, 359 miles. With 83 kWh used, we had 11 percent of the battery remaining—which equates to 41 more miles at the rate I was going. Right at 400 miles if you add it up. ...

https://www.motortrend.com/cars/tesla/model-s/2019/exclusive-2019-tesla-model-s-review/

[Sigmetnow]

Note that historically, Gizmodo has never missed an opportunity to layer on the snark and derision against all things Tesla.

Elon Musk Was Right: Cheap Cameras Could Replace Lidar on Self-Driving Cars, Researchers Find

In a paper that will be presented at the 2019 Conference on Computer Vision and Pattern Recognition in June, Pseudo-LiDAR from Visual Depth Estimation: Bridging the Gap in 3D Object Detection for Autonomous Driving, Cornell researchers detail a potential breakthrough for autonomous vehicles. Cameras have typically been considered an inferior technology to Lidar given that they’re often installed at low angles, near a vehicle’s bumper, resulting in images that tend to distort objects in the distance which confuses neural networks trying to process and interpret the data.

But by placing a pair of cheap cameras on either side of a vehicle behind its windshield, stereoscopic images are produced which can be converted to 3D data. Because the images are being generated from a higher vantage point, closer to where Lidar systems are typically installed, the 3D data that was generated from the cameras was found to be nearly as precise as what laser scanners are able to generate, without distortion, and at a fraction of the cost. ...

https://gizmodo.com/elon-musk-was-right-cheap-cameras-could-replace-lidar-1834266742/

[Sigmetnow]

The Tesla Show (@TheTeslaShow) 4/24/19, 10:15 AM
Morgan Stanley on their test ride at Tesla Autonomy Investor Day.
https://twitter.com/theteslashow/status/1121054967257677824

[BeeKnees]

Tesla first quarter figures are very poor.  Way below expectations.

$700m loss

[sedziobs]

Cash position took a big hit after the operating loss and $920 million debt payment, down to $2.2B from $3.7B. So far the market isn't reacting strongly. It's actually gone up a bit since the release.

No growth in solar or storage. Ramp up of solar roof, powerwall, Semi, Y, and Roadster are all slated for the next couple years. At this point, I don't see how they can ramp up even one of those without raising capital.

[crandles]

From 63000 vehicle deliveries in Q1 to 90,000-100,000 vehicle deliveries in Q2 and this will only "significantly reduce our loss in Q2".

"return to profitability in Q3" I note this says 'in Q3' not Q3 onwards.

reaffirm our prior guidance of 360,000 to 400,000 vehicle deliveries in 2019

If our Gigafactory Shanghai is able to reach volume production early in Q4 this year, we may be able to produce as many as 500,000 vehicles globally in 2019. This is an aggressive schedule, but it is what we are targeting. However, based on what we know today, being able to produce over 500,000 vehicles globally in the 12-month period ending June 30, 2020 does appear very likely.

[GoSouthYoungins]

Any smartish person can see the pattern. Any sentient person can tell that there is more red beneath the line than green over it.

The business model is pathetic. The company will never make money. No Musk venture will ever make money.

"With a little luck, and a lotta hard work, we think Tesla can make a profit in Q1." lol.

[GoSouthYoungins]

"We are a exponential growth company!"

-The company with pathetically low CAPEX, shrinking 7 quarters in a row, and lower than non-"growth companies". Oh, and no cash. And tons of debt.

[Sigmetnow]

Multiple factors specific to Q1 led to lower production and sales of Model S/X in Q1 — but 90- to 100k S/X deliveries in 2019 are still expected:

Deliveries of Model S and Model X declined to 12,100 vehicles in Q1 compared to our two-year run rate of roughly 25,000 units per quarter. This decline was mainly caused by weaker Q1 demand due to seasonality, pull-forward of sales into Q4 2018 in the U.S. due to the first scheduled reduction of the federal EV tax credit in Q1 and discontinuation of our 75 kWh battery pack. We also had a mismatch between orders and deliverable cars. For example, due to adjustments in pricing mid-quarter, the take rate for the performance versions of Model S and Model X increased faster than we were able to supply.

——
Discussion during the Q1 financial results call stressed how difficult the Q1 “wave” production (making batches for overseas during the first half of the quarter, then for North America in the second half) was for production, transport, and delivery efforts alike, making minor transport slips result in deliveries delayed to Q2.  In fact, one-half of all Q1 deliveries happened in the last ten days of the quarter!  They are therefore transitioning to a more regular, non-wave strategy for Q2 and forward, which should lessen stress on personnel, improve scheduling, and reduce overtime costs.  And with the many Q1 deliveries shifted to Q2, Q2 deliveries should be great — targeting 90- to 100k.

Quick delivery times for Model 3 orders are made possible by production batches.  Despite using this method, Tesla inventory remains well below industry average.

Unlike Model S and Model X, we do not build Model 3 vehicles to order. Rather, given its significantly higher volume, we build different variants of Model 3 in batches (including regional versions), and every vehicle that leaves the factory initially becomes inventory.
While in inventory, those vehicles are then matched to a specific order made by each customer.
…
… our global Model S, Model X and Model 3 inventory (including vehicles in transit and vehicles owned by our sales and service organizations) at the end of Q1 equaled 30 days of sales, less than half of US industry average and in line with our historical numbers.

—-
Only 3.5% of the trade-ins for Model 3 are Model S, indicating minimal cross-demand.  The majority of trade-ins for the Model 3 are non-premium vehicles, showing the market for the car extends well beyond the premium segment.

Model 3 was yet again the best-selling premium car in the US in Q1, outselling the runner-up by almost 60%. This is not surprising given that, for the first time in history, the price of an electric vehicle is lower than its gas-powered equivalents. While global premium vehicle sales reached 8 to 9 million units (depending on definition) last year, the Model 3 is attracting buyers from other segments. Since introduction of Model 3 Standard Range and Standard Range Plus, 69% of trade-ins were non-premium vehicles, indicating that Model 3 is demonstrating appeal beyond the premium segment. Our global expansion for the Model 3 has just begun, competing in a segment that is vastly larger than just the US. Model 3’s average selling price (ASP) in the US remains strong, as a majority of these orders are for long range or all-wheel drive versions. We are also seeing increasing take rates of our Autopilot options, as this suite of features improves.

Average Sale Price for Model 3 has been increasing ever since the Standard Range was made available February 28.  Standard Range Plus with Autopilot seems to be the sweet spot for new orders.
“We continue to target a 25% non-GAAP gross margin on Model S, Model X and Model 3, depending on variant mix and option take rates as our product offerings change.”

[Sigmetnow]

From the financial call: Although Musk has stated previously that Fremont had no room for Model Y production, the Tesla team has come up with a way to “append” existing building and warehouse space that would make it possible.  Expect an announcement of the chosen location in the next few weeks.  Tooling and equipment has been ordered, so no delay is expected.  Also, lessons learned with Model 3 production mean that future production lines will incorporate significant CapEx savings.

We intend Model 3 to be the first step in a platform which we can cost effectively and quickly replicate across geographies and vehicle types. We have spent years developing this platform, and Gigafactory Shanghai and our planned Model Y production line will be the first to reap the benefits of this investment. Learning from our experience, we can now build a second-generation Model 3 line in China that we expect will be at least 50% cheaper per unit of capacity than our Model 3-related lines in Fremont and at Gigafactory 1. Our Model Y manufacturing capacity will have the same simplicity as the line planned for Gigafactory Shanghai.

—-
A potential robotaxi fleet of one million cars, enabled via an OTA software update as regulations allow.  Only Tesla has this potential.  Other ride-share companies have a few hundred (more expensive) self-driving vehicles at most.

A custom-made robotaxi capable of running about a million miles using a single battery pack, with all the sensors and computing power for full autonomy, should cost less than $38,000 to produce. We believe low vehicle cost, low maintenance cost and an expected powertrain efficiency of 4.5 miles per kWh should make this the lowest cost of ownership, and to be the most profitable autonomous taxi on the market.

——

OUTLOOK
 Although we are driving towards higher internal goals, we reaffirm our prior guidance of 360,000 to 400,000 vehicle deliveries in 2019, representing an increase of approximately 45% to 65% compared to 2018. Please note that vehicle production will be significantly higher than deliveries, as it takes several weeks to transport cars from California to distant customers, especially in other countries, where they must also be processed by customs. Deliveries, production and customer orders, which are all materially different, are often conflated when analyzing Tesla.

If our Gigafactory Shanghai is able to reach volume production early in Q4 this year, we may be able to produce as many as 500,000 vehicles globally in 2019. This is an aggressive schedule, but it is what we are targeting. However, based on what we know today, being able to produce over 500,000 vehicles globally in the 12-month period ending June 30, 2020 does appear very likely.

We continue to target a 25% non-GAAP gross margin on Model S, Model X and Model 3, depending on variant mix and option take rates as our product offerings change.

In response to the operational challenges we experienced with international expansion in Q1, we are in the process of balancing our regional vehicle builds throughout the quarter. This provides an opportunity for additional cost efficiencies in our factory, supply chain, logistics operations and delivery centers.

With the recently announced product improvements on Model S and Model X, as well as continued expansion of Model 3 globally, we expect our order rate to continue to increase throughout the year as our production levels increase. We believe we will deliver between 90,000 and 100,000 vehicles in Q2. Although it is possible to deliver a higher number of vehicles, we believe it is important to begin unwinding the "wave" approach to vehicle deliveries, where overseas cars have been made in the first half of the quarter and North American cars have been made in the second half. This puts extreme stress on Tesla, negatively affects our working capital needs and adds to our cost structure.

Tesla First Quarter 2019 Update
https://ir.tesla.com/static-files/b2218d34-fbee-4f1f-ac95-050eb29dd42f

[Sigmetnow]

“Our 2019 capex, the vast majority of which will be to grow our capacity and develop new vehicles, is expected to be about $2.0 to $2.5 billion. We believe this amount should be sufficient to continue to develop our main projects, such as Gigafactory Shanghai, Model Y and Tesla Semi, as well as for the further expansion of our Supercharger and service networks.

Operating cash flow less capex should be positive in every quarter including Q2. As the impact of higher deliveries and cost reduction take full effect, we expect to return to profitability in Q3 and significantly reduce our loss in Q2.”

https://ir.tesla.com/static-files/b2218d34-fbee-4f1f-ac95-050eb29dd42f

[Sigmetnow]

Other notes from the financial call:

• Maxwell Technology acquisition is on schedule, expect close in mid-May.

• “Tesla Insurance” is in the works, expect launch in about a month.

• Average Sale Price for Model 3 (most recent data) is about $50,000.

• Analyst: “Why don’t you raise more capital for growth?”
Musk: “It’s healthy to be on a spartan diet for a while.” Capital is not a constraint for our growth.

• Giga 3 progress is going “incredibly well.” “I get daily photos.”

Edit:
• Can still buy the $35k Model 3, via phone or in store. “Will remain there.”

[ASILurker]

Yay, TSLA drops to $250, then bounces way up to $257, see it ain't too bad.

 

[ASILurker]

Personally I'd like to hear what the Chairperson of the Tesla Board has to say about these matters.

Under oath, in a witness box, being cross examined by a bevy of Lawyers in a court of Law preferably.

[oren]

Terrible quarterly results on all counts. Surprisingly Tesla still insisting on its annual delivery guidance, which I think is what prevented a further slide in the stock, in the face of a possible capital raise. Should they deliver 90k-100k cars in Q2 as they are guiding, it will be a positive surprise in my view.
Doing away with the "wave" production pattern is a big operational plus, though will impact quarter-end cash positions once it is implemented.
Looking forward, it all still depends on demand. Should enough orders materialize, I believe Tesla can be profitable and grow. Should orders shrink or even remain at Q1 levels, expect big trouble.

[ASILurker]

It's only just begun ....

Wedbush analyst Daniel Ives downgraded shares of Tesla Inc. to neutral from outperform following the company's first-quarter earnings report, in which the company posted a sizable loss. Shares are down 1.2% in premarket trading Thursday.

"The demand story at Tesla is quickly changing and the company has unfortunately not adjusted to an evolving [electric-vehicle] landscape (especially in the U.S.) with the well thought out marketing and distribution logistics needed to manage this difficult and complex hand holding process for customers, employees, and investors," Ives wrote.

"To this point, in our 20 years of covering tech stocks on the Street we view this quarter as one of top debacles we have ever seen - while Musk & Co. in an episode out of the Twilight Zone act as if demand and profitability will magically return to the Tesla story."

He called management's outlook "aggressive" while arguing that the company " is not taking aggressive enough cost cutting actions and shutting down future endeavors to preserve capital and give a sustained path to profitability for the Street."

Ives lowered his price target 25% on shares to $275 from $365.

https://www.marketwatch.com/story/tesla-stock-downgraded-at-wedbush-after-earnings-debacle-2019-04-25

The Twilight Zone is an apt comparison.

[NeilT]

Interesting figures.

Now maybe let's see why the markets are not so worried about Tesla right now.

Tesla made a loss of ~$700m
Tesla had 10600 vehicles in transit at the end of Q1 and could not make delivery due to the protracted transit time.
The average price of a Tesla Model 3, delivered, reached $60,000 in 2018.

So Tesla had around $630m in direct revenue floating around the Ocean at the end of the quarter.  Probably more because we know that the Model 3 configurations going to the international market were premium models.  But let's stick with that figure for a want of better evidence.

So Tesla had manufactured and started delivery of vehicles which would, if they had actually been delivered, brought Tesla to near break even.

Then, of course, we have the fact that Tesla repaid almost $1bn of debt in March 2019.

I had a quick scan through the Q1 2018 analysis and could not find anything about paying back debt in Q1 2018, only how much cash they were burning.

So, a quick analysis you wont' see anywhere else, comparing Q1 2018 with Q1 2019.

In Q1 2018 Tesla lost ~$700m, had no reported deliveries in the pipeline and paid back no debt.
In Q1 20198, Tesla lost ~$700m, had ~630m in revenue in the pipeline and paid back £930m in debt.

On my quick calculations that means that, one time factors (such as debt repayments aside), Tesla generated $1.65b more actual cash flow than in 2018.  The fact that $630m of the actual cash will only have arrived by now, rather than in Q1, is irrelevant.  If it were a case that Tesla had exceeded the available cash on hand to continue manufacturing and had to go to the market because of this lack of revenue, this would be different.

But Tesla did not.  Tesla paid back debt in Q4 2018 and still strengthened cash on hand.  Tesla used that cash to pay back debt in Q1 2019 and had enough cash on hand to continue operating.

Even more so, Tesla, after paying back that debt, virtually broke even on cost v deliveries and continues to have enough cash on hand to operate for Q2.

There may be $182m debt to pay back in April, but there are no really major debt repayments, now, for the remainder of 2019.

If I can do this simple extrapolation, so can the analysts.  Tesla is making money out of making cars.  Tesla has $9bn in debt and has to pay it back, over time.  This is going to depress profits for some time to come.

What these results do not mean is that Tesla is going to fail.  The share price reflects that.

Let us remember Cleantechnica's Impossible Trick #9.


I note that the Q4 2018 financial statement listed $139 million in profit (unadjusted), but the Cash position was very different.

Tesla said its cash position substantially improved by $1.45 billion, despite spending $230 million to repay convertible bonds during the quarter.

There is a trail of articles all over the net about how Tesla can't pay its debts, never makes a profit, is going to fail, has impossible challenges to continue operating.

The figures do not reflect the position.  I stuck my stake in the ground and continue to defend it.  Tesla will post its first profitable year in 2019.

If that were all, then it would be enough.  But it is not.  The Models S and X will see a significant refresh, Model Y will launch, the Semi build out continues towards launch in 2020 and Gigafactory 3 continues to build out.

All paid out of revenue as Tesla is not looking for more funding.

Another quarter before it's worth discussing again.  The trend continues to fund the business growth and operations continue.

Q2 indicators will tell the tale.  If we see continued decline in sales and additional loss of revenue, then things will look bleak.  But these figures are far from bleak.  One time costs which rob profitability, revenue removed from the report due to timing issues in delivery.  These are not fundamental indicators unless they happen every quarter.

The biggest change here is that Musk is not playing up the upsides, he's playing them down.  Which sets expectations which are easier to exceed.  UPOD, something Musk has to learn in order to stop Bears eating his company alive.

[NeilT]

Ives lowered his price target 25% on shares to $275 from $365.

Well given that Tesla shares are currently at around 258 that's a BUY isn't it??

Also ignoring the fact that there is some $600m odd of revenue about to be added to the Q2 figures when they release, just as there were in Q4 2018.

This is really interesting, share price advice is lowered by 25% but is still higher than the current share price. This is going to drive the price of the shares lower...

Only on Wall Street.

[ASILurker]

Ives lowered his price target 25% on shares to $275 from $365.

Well given that Tesla shares are currently at around 258 that's a BUY isn't it??

Not for smart people, who know the basics of stock markets, no. (sigh)

May 20th TSLA goes below $200 / share!

Also ignoring the fact that there is some $600m odd of revenue about to be added to the Q2 figures when they release, just as there were in Q4 2018.

This is really interesting, share price advice is lowered by 25% but is still higher than the current share price. This is going to drive the price of the shares lower...

Only on Wall Street.

Only in your mind.

Maybe just admit you don't have a clue what you're talking about when it comes to Tesla, analyst's comments/advisories, and the share markets either? 

NeilT's biggest problem is he doesn't know how ignorant and stupid he really is!

[magnamentis]

all things highly tuned apparently run bust just before they explode in our faces but
that aside, sales, demand and all that stuff that is permanently posted here to support
"glory" does not matter from the point of no return that has been reached already
once the debts reached a certain level.

i dunno how many million cars GM or Chrysler sold the year before they filed but it were many and it did not help at all because they were not able to pay their bills and amortize their debts.

they both fell like a try on first serious gust and GM would be gone for good with gov-bail-out of some kind back then.

further i agree with "Lurk" as to stock prices and that discussion and on both counts in his post on that topic.

tesla is definitely overindebted and will very soon lose significant market share once people understand by comparison that range and acceleration are not the only qualities that count.
at the end tesla will even technologically join the fate of U.S cars from earlier than the 70ies
agains japanese and/or german cars after that era.

While the last paragraph is an opinion and speculation from my part (educated guess if you like), the part about finances is a fact that will hit like a bomb is peaceful till the split second before it goes of. only after it went everyone is either muted, deaf or say uppsss.... that was ugly or all of those.

[crandles]

Interesting figures.

Now maybe let's see why the markets are not so worried about Tesla right now.

Tesla made a loss of ~$700m
Tesla had 10600 vehicles in transit at the end of Q1 and could not make delivery due to the protracted transit time.
The average price of a Tesla Model 3, delivered, reached $60,000 in 2018.

So Tesla had around $630m in direct revenue floating around the Ocean at the end of the quarter.  Probably more because we know that the Model 3 configurations going to the international market were premium models.  But let's stick with that figure for a want of better evidence.

So Tesla had manufactured and started delivery of vehicles which would, if they had actually been delivered, brought Tesla to near break even.

$630m revenue, yes maybe. But gross margin is near 20% and that stock is in balance sheet. So that is only around $125m of the $700m loss. Nowhere near enough to bring "near break even".

[crandles]

Interesting figures.

Now maybe let's see why the markets are not so worried about Tesla right now.

Tesla made a loss of ~$700m
Tesla had 10600 vehicles in transit at the end of Q1 and could not make delivery due to the protracted transit time.
The average price of a Tesla Model 3, delivered, reached $60,000 in 2018.

So Tesla had around $630m in direct revenue floating around the Ocean at the end of the quarter.  Probably more because we know that the Model 3 configurations going to the international market were premium models.  But let's stick with that figure for a want of better evidence.

So Tesla had manufactured and started delivery of vehicles which would, if they had actually been delivered, brought Tesla to near break even.

$630m revenue, yes maybe. But gross margin is near 20% and that stock is in balance sheet. So that is only around $125m of the $700m loss. Nowhere near enough to bring "near break even".

Stock in balance sheet was up from $3.1bn to $3.8bn. So it may well be possible to release a little less (going to have some not none in international transit) than $.7bn in cost and $.875bn in revenue. So it is looking like less than $.175bn doing it this way.

Certainly can't release that revenue without also wiping out that stock asset.

[sedziobs]

$630m revenue, yes maybe. But gross margin is near 20% and that stock is in balance sheet. So that is only around $125m of the $700m loss. Nowhere near enough to bring "near break even".

The stock is on the balance sheet, yes. Inventories increased by $700m during the quarter. However, I don't think gross margin of transit/inventory vehicles is reflected in net cash flow. In the $1.5B cash loss, cost of those vehicles has been realized but revenue has not (shown as $676m loss in operating assets and liabilities). I could be wrong on this, I'm just an amateur observer. 

So while revenue from vehicles in transit wouldn't be enough to bring operating income to break even, it would be enough to bring cash flow to break even before debt payment.  Cash flow seems to be the more relevant metric over the next few quarters.

[magnamentis]

with hundreds of millions of debts a company has to produce huge profits over a decade or more to get into a healty state which is why it won't happen, neither are such kind of profits anywhere in sight nor will the "boom" or kind of "monopoly-like" situation last long enough, once the market share is 1/10 while EV sales are 5 times higher it will get ugly with TSLA finances.

of course, as i state from day one, there are ways but each of those ways means that TSLA as a standalone profitable automaker won't last. What it does not mean is that the brand will disappear necessarily or that the technology will be lost, i just believe that TSLA will be purchased or there will be a merger on a big scale and if a company is not able to survive on it's own it's in the meaning of this thread a fail, even though, also as i always point out, TSLA as an "event" as a developer and as an important part of transition to electification etc. will remain a success, even once the enterprise as such will have been swallowed by market mechanisms so to say.

[crandles]

$630m revenue, yes maybe. But gross margin is near 20% and that stock is in balance sheet. So that is only around $125m of the $700m loss. Nowhere near enough to bring "near break even".

The stock is on the balance sheet, yes. Inventories increased by $700m during the quarter. However, I don't think gross margin of transit/inventory vehicles is reflected in net cash flow. In the $1.5B cash loss, cost of those vehicles has been realized but revenue has not (shown as $676m loss in operating assets and liabilities). I could be wrong on this, I'm just an amateur observer. 

So while revenue from vehicles in transit wouldn't be enough to bring operating income to break even, it would be enough to bring cash flow to break even before debt payment.  Cash flow seems to be the more relevant metric over the next few quarters.

Yes re not in cash flow. I am an observer who happens to be a Chartered Accountant, and while not expert on US GAAP, I know that much.

However break even cash flow from operating activites is a terrible position. There was $.3bn in investing activities and $.65bn in debt financing and while the latter may be high in this quarter, you are very unlikely to get net income from these. Even total cash flow zero if extended forward means the fixed assets get used up for nothing.

Doing a different calc, 90k-100k vehicle deliveries for Q2 has been suggested. 63k deliveries in Q1. If they manage 50% more deliveries 94.5k then the automotive gross profit gets increased 50% from $.65bn to $.975bn an improvement of $.325bn and this reduces the loss by less than half.

If we get the automotive margin to increase back to 25% from 20%, that would be an enormous stretch unless you believe they have really struggled with lack of some parts but still had all the labour and other non part costs. Then automotive gross profit goes from $.65bn to $1.31bn.
This enormous stretch still doesn't quite wipe out that massive $.7bn loss in Q2. I am much more inclined to believe the 25% automotive gross margin is unlikely to return in the next few quarters.

Maybe Q3 becomes profitable only because of Fiats payments to pool vehicles?

[sedziobs]

break even cash flow from operating activites is a terrible position.
...
Maybe Q3 becomes profitable only because of Fiats payments to pool vehicles?

Right, break even cash flow is obviously not sustainable indefinitely. I just meant to imply that the cash burn rate shouldn't be that high going forward. They're going to have to get back to operational profitability soon. They will need to get S/X numbers back to 25k and sustain 90-100k Model 3. I'm guessing part of the reason for projecting another loss in Q2 is that upgraded S/X won't be delivered until the second half of the quarter at the earliest, likely suppressing deliveries again.

[sedziobs]

I don't understand what is happening with energy storage. Tesla states:

Energy storage production in the second half of 2018 was limited by cell production as we routed all available Gigafactory 1 cell capacity to supply Model 3. Some Gigafactory 1 cell production has been routed back to the energy storage business, enabling us to increase production in Q1 by roughly 30% compared to the previous quarter.
...
Energy generation and storage revenue in Q1 decreased by 13% over Q4 2018. This decrease was mainly driven by lower solar deployments that fell from 73 MW to 47 MW sequentially, which was partially offset by a 2% increase in storage deployments.

An increase in production by 30% but sales by only 2%. Does this mean they also have a sizeable inventory of energy storage?

[oren]

They have a large backlog of energy storage (some people have been waiting for Powerwalls for two years). Storage is fed by batteries from many suppliers, but it seems this quarter they used more of their own batteries and less from external suppliers.
Still I would have expected storage to grow, while their solar is obviously shrinking. Maybe the global demand for batteries caused a shortage among the external suppliers.

[crandles]

I don't understand what his happening with energy storage.  Tesla states:

Energy storage production in the second half of 2018 was limited by cell production as we routed all available Gigafactory 1 cell capacity to supply Model 3. Some Gigafactory 1 cell production has been routed back to the energy storage business, enabling us to increase production in Q1 by roughly 30% compared to the previous quarter.
...
Energy generation and storage revenue in Q1 decreased by 13% over Q4 2018. This decrease was mainly driven by lower solar deployments that fell from 73 MW to 47 MW sequentially, which was partially offset by a 2% increase in storage deployments.

An increase in production by 30% but sales by only 2%. Does this mean they also have a sizeable inventory of energy storage?

I certainly hope so. However gross margin falling from 11.5% to 2.4% also sounds awful. What causes that large a gross margin drop? Maybe they have had to write off some of that increased production/inventory?

[Sigmetnow]

Elon Musk (@elonmusk) 4/25/19, 1:44 PM
UK Model 3 order page goes live next week, followed shortly thereafter by Japan, Australia, New Zealand & Hong Kong

https://twitter.com/elonmusk/status/1121469940815425536

—-

Tesla (@Tesla) 4/25/19, 10:57 AM
Newly upgraded Model S and X drive units rolling down the production line at Gigafactory 1

https://twitter.com/tesla/status/1121428109599707136
Photos at the link.

—-

Elon Musk (@elonmusk) 4/24/19, 2:09 PM
As mentioned before, Tesla full self-driving option price will increase after May 1. You can order online at Tesla.com.

https://twitter.com/elonmusk/status/1121113876488081408
   

ValueAnalyst (@ValueAnalyst1) 4/24/19, 2:27 PM
It's unlikely that Tesla would up the price of FSD, which today costs $5000 before delivery and $7000 after delivery, unless if the value proposition was also about to increase, soon.

https://twitter.com/valueanalyst1/status/1121118404847570945

[oren]

this thread is a total waste of people's time and energy. That TESLA matters is a MYTH 

this Tesla thread will remain a sad joke.  

and for the record *again* - I do not care whether Tesla succeeds or fails, because it's totally irrelevant to anything really important

And yet you have overtaken sig as the number one poster on this thread...

[ASILurker]

this thread is a total waste of people's time and energy. That TESLA matters is a MYTH

this Tesla thread will remain a sad joke. 

and for the record *again* - I do not care whether Tesla succeeds or fails, because it's totally irrelevant to anything really important

And yet you have overtaken sig as the number one poster on this thread...

My quotes there are all very logical, rational, justifiable and well reasoned. Even though they are my own personal opinions they are as valid as any others.

YOU ACT LIKE A MONGREL DOG OREN!

And you are about as smart as one too.

[oren]

From Wikipedia:

There are currently four planned powertrains for the Model Y: Standard Range, Long Range, Long Range with Dual Motor All-Wheel Drive, and Performance. All models are planned to be available by late 2020, apart from the Standard Range model, which is planned to enter production in early 2021.

BTW, the Tesla Pickup is to be unveiled this year, with production date probably in 2021 (my guess) or 2020. The Tesla Semi is to start production in 2020.

[NeilT]

$630m revenue, yes maybe. But gross margin is near 20% and that stock is in balance sheet. So that is only around $125m of the $700m loss. Nowhere near enough to bring "near break even".

That's backwards isn't it?  The net manufacturing cost was $511m and the expected profit was $125m.  So the sum carried on the balance sheet was the $511m manufacturing cost and not the expected 20% profit.

The $125m is the difference between what was carried on the balance sheet and what was expected to be earned with the delivery over and above manufacturing costs.  Simple delivery at 0% profit would have net the $511m carried on the balance sheet as a net cost of operation.

Put this into context of the overall automotive market (Mercedes Benz making a 16% drop in revenue, year on year, for the month of March) and Tesla is not doing to badly.  Mercedes is not an outlying marker either, it is indicative of the general trend in motor vehicle sales in 2019.

Tesla made the right call to move sales to International markets.  The problem, as always, is how long it takes to deliver the vehicles.  With the direct to market approach of Tesla, there is a direct financial impact to the failure to deliver on a quarter boundary.  Something mitigated by the dealer network infrastructure in traditional automotive sales.

Take a step back.  Tesla has $2.2bn in cash on hand and just paid back almost $1bn in debt on the quarter.  Tesla has ~$600m of Guaranteed cash flow in pipeline for delivery and that will go into Q2 stats without incurring any additional operating cost.

Contrast Q3 2018.  Down to under $2bn in ready cash, uncertain ramp up of vehicle deliveries, Analysts claiming that Tesla could not manufacture the Model 3 at a profit at ANY price, more than $1bn in debt to repay in the next 2 quarters, Everyone expecting that Tesla would have to go back to the market for money just to get the Model 3 out of the door in volume.

Never mind the fact that Tesla was still developing the Semi, the Model Y for 2019 launch, solar products and growing all of the Gigafactories.

Some times it takes a quick view in the rear view mirror to understand the difference in the position of Tesla in 2018 and the position in 2019.

At the same time, in the last few weeks, significant proportions of the roof have gone on at Gigafactory3.  Roofing is hard, walls are a very different thing and significant progress will be seen in days.  As soon as the walls are in, utilities will start to build out ready for manufacturing equipment.

The financial position of Tesla, in 2019 is radically different to 2018 and the top line figures simply do not show it.  In Q1 2018 Tesla had 2 very premium models and <200,000 production capacity.  In Q1 2018 Tesla was Years in backlog to faithful pre orders with a long delivery time in sight and only the Model 3 as a future prospect. Solar was a major drag and Tesla looked like it was going to have to go back to the markets, possibly more than once, to fund Model 3 volume production capability.

Today?

Tesla misses being profitable by a sum which is a small proportion of the revenue for the quarter.  More importantly, had Tesla been in a position to convert the loan it paid back, it would have made a large profit.  Instead it paid back more in debt than it made in losses.

Back in Q3 2018 people were sceptical about Tesla's ability to deliver Gigafactory3.

Today?

Well take a look. (scroll down to the pictures).

I believe it was GSY who stated that Tesla would own the majority stake in a mud field at the end of 2019.  I count some 19 cranes working on the roofing.  I also noticed the train lines being built into the train station.

The Tesla financial report for Q1 2019 is being held up as a reason for Tesla to be in _real_ trouble.

Tesla could have refinanced that ~$1bn debt and would have ended the quarter significantly profitable with revenue shifting into Q2 due to delivery delays.  Instead Tesla chose to pay back debt out of operating cash, whilst continuing to grow the operational capacity by ~100%.

What happens if Tesla actually makes a profit by the year end?  They pay tax to the tax man out of those profits.  On the other hand Tesla could continue to end every year neutral whilst growing the business and paying down debt on the profits made by manufacturing.

Just remind me again.  When did Google last pay a dividend?

[NeilT]

BTW, the Tesla Pickup is to be unveiled this year, with production date probably in 2021 (my guess) or 2020. The Tesla Semi is to start production in 2020.

Plus the Models S and X refresh.

Tesla announced a 10-percent improvement to the top driving range for both its Model S fastback and Model X crossover Tuesday—resulting in what it claims will be EPA-official ranges of up to 370 miles and 325 miles for the two models, respectively.

370 miles blows all the competitors out of the water today.  This is for the 100KW/h battery too.  Not for a larger battery.

Plus V3 supercharger support.

A much needed refresh which should drive some key sales when they start delivering as it brings Model3 advances into the S and X range.

[Sigmetnow]

Tesla Gigafactory 3 on track to break China’s record for fastest factory buildout

... Part of the reason behind Gigafactory 3’s incredible construction speed is the urgency of the project, with Tesla CEO Elon Musk aiming to start Model 3 production by the end of the year. Shanghai official Chen Mingbo highlighted this urgency in March, urging Tesla and its construction partner to finish the initial Phase 1 buildout by May.

To make such an aggressive timetable into reality, Tesla and its construction partner are operating 24/7 on Gigafactory 3’s buildout. The nonstop work on the site has been a real difference maker, allowing some sections of the Phase 1 area to enter the roof paving stage roughly a month after the first pillar of the factory was built. ...

https://www.teslarati.com/tesla-gigafactory-3-china-record-construction/

[crandles]

$630m revenue, yes maybe. But gross margin is near 20% and that stock is in balance sheet. So that is only around $125m of the $700m loss. Nowhere near enough to bring "near break even".

That's backwards isn't it?  The net manufacturing cost was $511m and the expected profit was $125m.  So the sum carried on the balance sheet was the $511m manufacturing cost and not the expected 20% profit.

The $125m is the difference between what was carried on the balance sheet and what was expected to be earned with the delivery over and above manufacturing costs.  Simple delivery at 0% profit would have net the $511m carried on the balance sheet as a net cost of operation.

Not sure what you think I have backwards. Revenue $630 million, 20% gross margin is $125million (or $126million) so the stock on balance sheet is $505 million. Maybe I could have stated that more clearly. I have no idea where your $511m came from.

Anyway you were claiming the $630million brought Tesla near break even.

So Tesla had manufactured and started delivery of vehicles which would, if they had actually been delivered, brought Tesla to near break even.

but it is actually more like reducing the $700m loss by about $125million or 18% reduction in the loss which I don't think constitutes bringing near to break even.