Chapter 387: Typical Style, The Art Of The Deal
“Is it the L5 Autonomous Driving we understand, or the L5 Autonomous Driving shaped in the public mind by the media?” Musk asked.
The world’s first L5, a product that has never appeared before, just as Musk experienced earlier, is smoother than all current Marketplace and Laboratory Autonomous Driving performances, but exactly what kind of L5 it is, Musk still wants to ask clearly.
The suddenly emerging L5 is a bit like a sixth-generation fighter, where everyone was previously at fifth-generation, and you suddenly bring out a full sixth-generation, I have no way to say you’re not, but also don’t want to admit you are.
“Understood? Shaped?” Lin Ran repeated before asking: “Are you saying it’s like the Autonomous Driving Technology promoted by the media, where once it reaches L5, it absolutely won’t make mistakes?”
Musk nodded seriously: “Exactly, in media promotion, it seems technology can solve all problems, and once the highest level L5 Autonomous Driving is achieved, it will never make mistakes, its driving will always be correct, and it will absolutely never have safety accidents.
We know that humans still make mistakes, and Autonomous Driving will similarly.
The meaning of L5 is just that its probability of making mistakes is even smaller than humans, and it can adapt to driving needs in any scenario.
But it will still make mistakes.”
The car was still quietly driving on Shanghai streets, and along the way, this strangely shaped car also attracted countless onlookers and photos.
Some knowing passengers even stuck their bodies outside the window to take photos, wanting to see the people inside through the glass.
On Weibo, car reviewers from all sides were already boiling over.
This time it was really boiling over.
“I always knew Huawei had an internal vehicle project led by their 2012 Laboratory, I thought it was Zun Jie, thinking it would probably stack a lot of New Technologies on it.
Now it seems it’s Silly Girl!
Damn, this is a bit too surprising, honestly I’ve always firmly believed that the world’s first L5 will definitely come from Chinese Companies, we have the best market, the richest road conditions, the most diversified scenarios, and the most competitively intense automakers.
But I still didn’t expect it to be this fast, so fast that 2024 hasn’t even ended yet, and there’s already L5.
When will it launch? Which model will get this Autonomous Driving set? Starting at a million or will it be rolled out to all models?
And most importantly, how will other automakers play?
My first reaction was worry for XPeng, I’ve already broken out in a sweat for XPeng, they’ve always been the most aggressive, the ones pushing the pure Vision proposal the hardest, what now?
After thinking carefully, everyone will discover, no difference, this is an indiscriminate attack on all automakers.
Once Huawei rolls it out to all Harmony Intelligence series models, whether you use pure Vision or Sensor combination, you’ll directly face the impact from this L5 Autonomous Driving, this is dimensionality reduction strike, all non-Harmony Intelligence series automaker bosses probably won’t sleep well tonight.
On the contrary, Lei Zong might have less pressure, they sell more Emotional Value, and aren’t known for Autonomous Driving anyway.
But I believe Huawei won’t go too far, won’t eat the entire market, and will truly act like Bosch, selling this proposal to all Chinese Companies automakers.
This requires Wisdom, requires coordination from Yanjing side, when treating China as a whole, I believe this is a sweet trouble.
Finally, a complaint, Silly Girl is really too ugly-sounding, I know it’s a homage to Magic Phone, with Ran Shen’s age, he probably watched Magic Phone as a kid, but when it officially launches, I beg you, please change the name okay.
Giving everyone some Emotional Value, I won’t mock Huawei’s Classic of Mountains and Seas naming anymore, compared to Silly Girl, Classic of Mountains and Seas has much higher Prestige.”
“I finally know what that exposed test car on the road was before, at the time everyone was still joking, saying which company was desperate to design a model similar to Ideal MEGA, or a slightly smaller Ideal MEGA.
Now we know, it was Ran Shen, Deep Red lives up to being the well-deserved No.1 in domestic Artificial Intelligence, now verifying it again, the first L5 Autonomous Driving.
Musk felt bitter inside, thinking this trip to China was to inspect Tesla for the last time before taking office as NASA Director, maybe he’d be stationed at NASA from then on, unable to come to China for a long time, unable to see China’s Gigafactory.
Who would have thought, no surprise, only shock, whether it was Ran Shen’s deadly dialogue throughout the conversation process, or the final L5 Autonomous Driving experience, Musk probably won’t be able to sleep.”
“My observation point might differ from others, I focus more on the Technology itself.
Is it L5? I don’t know, but Ran Shen says it is, so I believe it definitely is, Deep Red doesn’t make cars, doesn’t need to make cars, and Ran Shen himself doesn’t need to gain attention this way.
Then talking about points observed during the entire live broadcast, first everyone noticed, in some narrow lanes, when the red light turns green and starting, there were multiple Electric Vehicles and delivery motorcycles quickly weaving around.
Traditional Autonomous Driving would generally be overly conservative, getting stuck in place afraid to move.
But this car didn’t hesitate, it directly made a smooth starting decision, even with a behavior mode similar to an old driver flooring the accelerator to grab the time window.
Here it demonstrates not just yielding, but prediction and game, the model clearly made high-confidence predictions on surrounding vehicles’ Movement intentions, bluntly put, the model itself is very confident, this is in line with Huawei series Autonomous Driving’s past aggressive confidence.
Then when crossing complex intersections, I observed that when entering a crossroad, there was a large truck on the left, and a pedestrian illegally crossing on the right.
The car didn’t wait dead, but did a small amplitude probing forward movement, which traditional Perception systems can’t achieve, this shows it’s not only relying on Vision, but also combining Radar signals, having probability modeling for occluded targets.
From my professional angle, this is the difference of L5: it won’t fall into ‘stop if can’t see’ like L2/L3, it dares to decision-make under incomplete information.
Finally, overtaking and lane-changing, when going from Baoshan onto the inner ring elevated, Ran Shen they encountered a slowly driving taxi, the System decisively changed lanes to overtake.
This behavior previous Autonomous Driving could also do, but in old Autonomous Driving, overtaking was a high-risk decision, generally needing strong rules + huge safety Redundancy.
When I saw it I specifically timed it, according to my timing, Silly Girl through Trajectory prediction + real-time game modeling, completed the overtake in just a 1.5-second time window, this is another embodiment of human Driver style.
These I think are why we feel it’s smooth, because its improvements are all-around, truly driving like an old driver, rather than the jerkiness past Autonomous Driving would show.
Finally, the only thing I’m curious about now is just one thing, Silly Girl’s performance in some extreme environments, which wasn’t demonstrated this time.”
On Weibo, Silly Girl L5 had already shot to the top of the hot search.
Self-media Bloggers in the car circle all voiced their views.
In Huawei Official’s comment section, there were more ordinary netizens, everyone asking when it will be released, many users holding coins waiting to buy even said they won’t buy until Silly Girl launches.
Now that there’s L5, no matter how flowery you smart driving promoters hype, I have to wait.
For car sales people, the sky fell in an instant, how to sell this car now? Especially brands that sell mainly on smart driving features.
Shifting sight back to the car, Lin Ran smiled after hearing: “Elon, your heart is already panicking.
Humans make mistakes, so how could our L5 autonomous driving possibly be the L5 as promoted by the media?
It will of course make mistakes and err.
We need to further clarify some concepts, namely what kind of autonomous driving can be called L5 autonomous driving.
For me personally, if its error rate is lower than the human accident rate in all scenarios, then I think it can be called L5.
If we demand stricter standards, then in every category of scenario—highways, urban roads, extreme weather, etc.—its error rate must be lower than the human accident rate, and its death rate must be lower than the death rate caused by human drivers; Silly Girl is at this stage.
If we demand even stricter standards, I think that’s the pinnacle L5, which requires it to drive better than any human driver in any scenario; this is very hard to quantify, and we are currently still working toward this goal.
Musk said softly: “L5.1, L5.2, L5.3?”
Lin Ran: “That’s what I personally think.”
Musk’s trip to China could be said to have come with high hopes and left disappointed, not completely disappointed, but half disappointed.
The appearance of Silly Girl made him realize that things had changed; the emergence of autonomous driving would affect Tesla and even the global car industry.
At the same time, Musk also thought, does this mean that Deep Red has taken another key step on the road to general AGI?
This made Musk even more regretful; why doesn’t this Chinese entrepreneur open-source like other Chinese companies? Alibaba or ByteDance would open-source their technology, but Deep Red never mentions open-sourcing; how could today’s young people forget the great internet spirit of open-sourcing?
On the special plane back to Florida, Musk thought this.
He wasn’t going to Washington; big T hasn’t moved into the White House yet; he was going to Florida, to big T’s estate in Florida.
Musk spread out that note folded like a map on the long table.
“China’s conditions are more straightforward than I expected.”
There were four lines on the paper:
America: Pressure the 4v semiconductor industry, Move advanced production capacity to America, lead American companies to hand over high-end production capacity to Japanese companies and Korean companies, and gradually hollow out 4v’s economic lifeline.
China: Use tariffs and industry tilt policies to precisely strike 4v companies, make the capital market lose support, transfer moon semiconductor-related technology, provide Lunar North Pole landing technology, and cooperate on research into moon semiconductor process.
Coordination synergy: On the surface, both sides oppose each other, but in reality, each gets what they need, avoiding direct exposure of coordination traces.
Exchange conditions: America must acquiesce to Japan exporting semiconductor-related technology and equipment to China.
Musk heavily
He raised his head: “They made it very clear, this is not about tariffs, not about the market, what China wants is to break free from the shackles in the field of semiconductor technology.
They want to be able to directly buy the key links from Japan: materials, equipment, process, and so on—everything related to semiconductors.
China wants a key to unlock it.”
“Interesting chips, interesting transaction.” Big T said.
The prospective Secretary of State sitting next to Big T said: “I don’t think we should reach such a deal with China.
If Japanese companies’ semiconductor exports to China are opened up, they could close the gap with America within ten years.”
Musk corrected, “Sorry, based on the results of my observations during this trip to China, it doesn’t need ten years—five years, five years is enough.”
“Ten years’ expectation is too optimistic.”
A middle-aged White Person man wearing an Army uniform wrote three words on the Whiteboard: 4v—America, 4v—Mainland, Japan tech—transit.
“The first two are transaction structures we can accept, and the third is the problem.”
He turned to look at Big T, “If we default to allowing Japan semiconductor technology exports to China, that would be equivalent to putting a knife in their hands in the global industry competition.”
An executive from IBM said: “Moreover, there is no gray area in this part.
This is the core problem of export restrictions.
“If we even slightly acquiesce and get caught, the entire Congress, media, and even allies will chase us relentlessly.”
Big T tapped the desk, his tone somewhat playful: “But the chips they’re offering aren’t small at all.
Turn 4v into an empty shell, and in the end, what we get in hand is high-end semiconductor production capacity, high-end employment positions, and a large amount of cash.”
Old John Morgan laughed, “Five hundred billion, Mr. President, at least we can directly squeeze five hundred billion US Dollars out of 4v.
In the long term, we can continuously harvest 4v through some operations, and the revenue will not be less than one trillion US Dollars.
“No need to give it to Congress, no need to give it to the national treasury.”
His tone was full of temptation.
Young John Morgan defected to the Donkey Party and deeply cooperated with Lin Ran. After Big T showed signs of winning, the Morgan family maneuvered Old John Morgan into Big T’s core team.
And Old John Morgan is well-versed in the ways of eating, drinking, and having fun, and he’s very good at flattering, playing golf is his specialty; after all, since the 1990s, he’s been kicked out of the family core industry and has focused on being a playboy, and these are all essential entertainment activities.
Old John Morgan is in New York, and in a certain sense, he is also old acquaintances with Big T, so he quickly integrated into this small circle.
The direct descendant of the marriage between the Morgan family and the Adams family, this pedigree has immense value in America.
Old John Morgan undoubtedly feels he must act. China breaking the blockade is the future, while we harvesting cash is the present.
The diplomatic advisor opened the folder: “This is also why they push Japan to the forefront.
What they want is not for us to directly sell technology, but for Japanese companies to directly export to China.
They gave a very appropriate angle. Japan’s technology is not the most advanced, and there is still a gap with ASML and America’s semiconductor companies, so they feel that we can accept this condition.
On the other hand, from Yanjing’s perspective, as long as America turns a blind eye, no longer forcing Tokyo to completely block exports, they can bypass America’s restrictions.
This is actually the most sensitive point.
Japan has two factions, one pro-America, one pro-China.
If Tokyo senses that we are loosening up, those industry factions will immediately push for equipment and materials export permits.
The deputy Secretary of State firmly opposed: “That is the core lifeline, we cannot hollow out 4v ourselves while also building bridges for them.”
The industry consultant was more hesitant: “If 4v is completely hollowed out, this can make America the only country with a truly complete semiconductor chain in the short term.
But if China obtains Japan’s technology, their catching-up speed will increase exponentially.”
Big T turned his gaze to Musk, “Elon, what do you think?”
The current Musk is at the height of favor, the number one favorite in Big T’s team.
“I think hollowing out 4v is completely feasible.
But everyone will think we are loosening the reins on China.
What I want to say is, similarly, we also need to obtain moon landing technology from China.
The current competition is on two tracks, one is space, one is semiconductors.
We use semiconductor time to exchange for space time, and China’s attached transaction condition is the entire wealth accumulated by 4v over decades.
This is the essence.
I think to achieve cooperation, the previous president’s trillion moon fund is uncertain how much is left.
My opinion is the same as before, we should not compete with China for the moon, we should go straight to Mars.
Obtain moon landing technology from China, build a small base on the moon, used for promotion to maintain America’s advantage in the field of aerospace, but in reality we won’t invest too much energy in this aspect, we will go all out to Mars, build a Mars base, modify Mars.
Musk’s subtext is, cooperating with China lets America’s public get the feeling of winning, ensuring our election four years from now can still win, ensuring that in the field of aerospace we maintain an advantageous position through promotion.
Then put all resources into Mars, win to the point of numbness.
Theoretical winning and actual winning, walk on two legs.
Big T said in his characteristically mumbled accent: “They don’t want us to publicly sign documents, they want us not to stop Japan.
From their perspective, this is just a tacit approval, and in the international system, tacit approval itself is sometimes more effective than policy.”
The long table fell into a period of silence.
Big T finally spoke: “A and B we will implement—core extraction of 4v, porcupine strategy.
Keep C as insurance too.”
He raised his head, his eyes carrying that campaign-stage momentum: “As for China’s demands, we won’t give them, won’t unlock.
Then, he changed his tone: “But, we can pretend to be negotiating.”
He pointed his finger at Musk: “You continue to be their ear, give them hope, but don’t loosen up.
We need to make them think there is an opportunity. This way, they will continue to cooperate.
Then find a way to make them hand over the moon landing technology, Elon, you can select some not-so-critical semiconductor technologies from Japan, give them a little sweetener first.
But the truly core keys, we absolutely will not hand over to them.”
Typical Big T style.
Want all the benefits, but don’t want to give up a single chip.
