发布会·中文精读

汽车的未来:马斯克主题访谈

Future of the Car: Elon Musk Keynote Interview

中文精读由 AI 基于原始视频与文稿改写,非官方翻译,可能存在偏差或错误;表述以原视频为准。说明

2022 年 5 月,FT Live 的“汽车的未来”论坛上,主持人问马斯克,特斯拉从零走到今天,最难的是哪一段。他想了想,说,基本上就是一个熊熊燃烧的垃圾箱,里面装满了愚蠢。

说这话的时候他语气很平,没有自嘲的意思。这场访谈里,他和联合创始人 JB Straubel(斯特劳贝尔)第一次一起出现在台上,把那场 2003 年的午饭局讲清楚了,也把第一版 Roadster 从头错到尾的过程讲清楚了。

先摆一组数字:美国历史上出现过几百家汽车公司,今天没破产过的只剩两家,福特和特斯拉。特斯拉自己也好几次差点掉出这份名单。

访谈后半段,话题铺得很开。2030 年两千万辆的目标、星舰的时间表、自动驾驶卡在哪、飞行汽车为什么是伪需求、他为什么要买推特、买下来准备改什么。

下面按主题梳理,数据尽量保留原样。

一场本来聊氢能飞机的午饭

主动打电话的是 Straubel。他当时在琢磨电动飞机,也可能是氢能飞机,方向都属于航空。两人约了午饭,前半段聊得很不顺。

Straubel 后来回忆:“我记不清你当时用了哪些形容词,反正都不太好听。你对氢能飞机不感兴趣,而且理由很充分。”(斯特劳贝尔)

话题转到锂电池,气氛变了。Straubel 讲的是把大量小号锂电池串成大电池包,做出续航几百英里的电动车。这句话放在 2003 年,等于说要破世界纪录,当时没人这么干过。

马斯克当场就懂了。Straubel 说,在他们聊过这个话题的所有人里,马斯克对这件事潜力的理解超过任何人,而且立刻就兴奋起来。

马斯克这边补了另一条线。他对电动车有兴趣很久了,在斯坦福读研时本来打算研究高能量密度超级电容,方向是给电动车做储能器件。后来他退学了,兴趣一直没断。

饭桌上,Straubel 提议他去试驾一辆车:南加州小公司 AC Propulsion 做的 t-zero 原型车,一台概念验证性质的电动跑车。马斯克说好啊。那是 2003 年。

试驾完,他去找 AC Propulsion,非常努力地想说服他们把 t-zero 商业化。对方说不干。

马斯克问:那你们介意我自己来做吗?我造一台商业化的电动跑车。对方说,不介意,没问题。

本来他打算和 Straubel 两个人干。AC Propulsion 的 Tom Gage 提醒他,还有别的团队也想商业化 t-zero,要不要见一见。介绍过来的其中一组,是 Martin Eberhard 和 Marc Tarpenning。

马斯克特意强调了一点:那个时候并不存在一家有实际形态的公司。三个男人,加上 Straubel 和他,然后大家决定合起来,做一个 t-zero 的商业化版本。特斯拉的前史就这么点东西。

两个前提全错,最后只留下 6% 的零件

最初的产品方案听起来很合理:拿路特斯 Elise(Lotus Elise)的底盘,配上 AC Propulsion 的电驱系统,两块拼起来就是一台车。

马斯克后来的评价是,这两个前提从根上就错。

Elise 太小,电池包装不进去。为了塞进电池、加固结构,车重增加了 40%,重量分布和燃油车完全不同,所有碰撞测试作废,车身结构作废。改到最后,跟原版共用的零件大概只剩 6%。

“回头看,从一张白纸开始设计会聪明得多。我们最后保留了 Elise 的一堆限制,却没换来任何零件复用的好处。那是个蠢到令人发指的决定。”(马斯克)

另一半也塌了。AC Propulsion 的技术到量产阶段几乎一项都用不上。Straubel 确认:交付第一个客户的时候,基本没剩下什么。

当时的电机控制器还在用模拟电路。天热一种表现,天冷另一种,油门响应会变,电机输出会变,跑不了两台就得重做。

供应链更离谱。最初的思路是全部外包,结果几乎没有一个外包商能做出他们要的东西。车身供应商直接做不出来,只能换。电池外包给了泰国一家做烧烤的公司。

“这话说出来有点可笑:我们要造史上最先进的电池,然后交给一家烧烤公司去造。他们当然造不出来。”(马斯克)

最后电池生产收回加州,自己做。

还有路特斯。每次路特斯自己遇到财务困难,就把给特斯拉的价格往上抬。成本根本控不住。

马斯克给这段日子的定性很粗暴,一场彻头彻尾、蠢到家的灾难,一路烧过来,公司得从这个坑里爬出来。

这段历史里有个细节值得反复咀嚼。特斯拉当时并不缺资源,也不缺聪明人,问题出在方案本身的美感上。两块现成的东西拼起来,听起来省时省钱,实际上把两套系统的假设全部继承了过来。Elise 的轻量化假设建立在一台一吨出头的小跑车上,换上电池组之后车重逼近一吨半,悬挂、制动、碰撞结构都得重算。AC Propulsion 的控制器假设的是原型车和赛道,换成量产和全球各地气候,散热与标定都得重来。两个假设同时失效,返工量就逼近从零开始。

今天起步,比 2003 年更难

主持人问:你们在无人区里开路,现在有几百家新造车公司在做同样的事,哪边更难?

马斯克的回答很干脆:现在更难。

他看到的最普遍错误,是新公司集体忘掉了特斯拉的头八到十年。“他们只看到成功,然后想,哦这个不错,照着抄一份。”(马斯克)

十七年前,世界上没有特斯拉。这个空白本身就是通行证,允许你犯足够多的错还有活路。今天要回答的第一个问题是:你的生态位在哪?你比现有厂商、比特斯拉强在哪?

他还说了一件容易被忽略的事。当年大厂几乎没有电动车项目,仅有的几个小项目还在关停。如果传统车企当时已经在做,Straubel 大概不会去吃这份苦。所以要有人做,只能由创业公司来做。动机跟“这是个赚钱的好点子”没多大关系。

造车创业的历史是一部恐怖片。坟场里全是你没听过的名字,出名的只有特斯拉和 Lucid 这种极少数。他给新公司的建议是:先做小,在小规模上把错犯完,备足有耐心的资本,然后一点点变聪明。一上来就冲高销量,等于不训练直接去奥运会。

“汽车行业跟自然垄断正好相反。社交媒体、搜索引擎是自然垄断,汽车是超竞争行业,全球都是。那些老厂有既有的客户、经销商、服务网络、工厂、积累了上百年的专业能力。他们是守在堡垒里的老兵军团。”(马斯克)

推特:永久封禁应该极其罕见

主持人用“社交媒体”这个词把话题引到推特。马斯克先立规矩:推特是上市公司,能说的有限,很多问题的答案就是不予置评。他还是把产品思路讲完了。

他的判断是,推特目前是最好的公共广场,换个说法,是最不坏的那个。要变好,先清掉机器人、诈骗、垃圾账号,尤其是“一个人或一个实体操作十万个账号”这种伪造影响力。

第二个动作是开源算法,直接放到 GitHub 上,让所有人看见它怎么工作,也让所有人提改进意见。任何人工干预都要标注出来,让用户知道这条推文为什么没流量、那条为什么爆了。他说现在这个过程太随机。

第三件事是平衡。他认为推特目前有强烈的左倾,因为公司开在旧金山。那里的人未必有意识地左,只是所处环境非常左,在他们自己看来已经算温和。这个偏向损害了美国其他地方对平台的信任。

他给了一个衡量成功的标准:最右的 10% 和最左的 10% 同样愤怒。“这种情况你大概不会得到多少赞美,你能做到的只是把愤怒摊平。”(马斯克)

交易本身呢?他强调远没到完成的程度。股东还没投票,代理文件还没提交,还有一些未解决的问题。他用了“客观上这还称不上一笔敲定的交易”这个说法,最好的情况大概两三个月完成。

关于永久封禁,他说自己和 Jack Dorsey 看法一致:应该极其罕见,只留给机器人、垃圾账号这类没有正当性的账号。

然后他直接说了特朗普。“我认为封掉特朗普是不对的,那是个错误。它疏远了这个国家的一大部分人,而且最终并没有让特朗普失去声音。他去了 Truth Social,美国右翼很大一部分也跟着过去了。结果可能比留在一个所有人能辩论的广场上更糟。”(马斯克)

主持人追问:他去国会山之前煽动人群,有人还带着绞索,这样你也认为封错了?

马斯克的回答是分层处理。错误和有害的推文应该删除或限流,账号可以临时禁言,永久封禁不行。永久封禁从根本上破坏了推特作为城市广场的可信度。他把这个决定称为“道德上错误,而且蠢到极点”。

至于会不会影响特斯拉,他的答案是不担心。订单排到一年以后,可能要限制接单,眼下 0% 的精力花在需求上,全在产能、工程和供应链上。

两千万辆是愿望,别当成承诺

2030 年做到年产两千万辆,相当于今天丰田和大众加起来。主持人请他描述那时的工厂、布局、车型。

马斯克先把定性说清楚:这是 aspiration,愿望,谈不上 promise,承诺。他也重申这个场合不做产品发布。

这个数字的来源是全球大约 20 亿辆汽车和卡车。要真正推动能源转型,每年至少得替换掉其中 1%,也就是两千万辆,这才算对可持续能源有实际影响。

从四个工厂走到可能二十个工厂,难度和已经走过的那段差不多。他给了一个参照:特斯拉的年增速比历史上任何大型制造商都快,包括福特 Model T。

最大的不确定性,他指向三年后可能出现的原材料约束。正极有两条主流路线,磷酸铁锂和镍。铁极其丰富,地球上 32% 的成分是铁,第二大元素是氧,占约 30%。磷稍微麻烦一点,也常见。锂到处都是,谈不上稀有元素。

真正的瓶颈落在设备上,把矿石加工成电池级材料的设备。这部分他们在和供应商一起解决。

主持人说磷酸铁锂的回收性比镍基电池差很多,会不会影响整体环保表现。马斯克直接打断:不对,磷酸铁锂完全可以回收,镍基也一样。主持人回了一句,我接受纠正。

能源的账很小:每平方公里 200 兆瓦

为什么非要长到这么大?

马斯克说,可持续能源有三根柱子:电动交通、固定储能、太阳能和风能,再加上地热和水电。风和光是间歇的,太阳不照、风不吹的时候,得靠电池顶上。

他给了个可以自己算的框架。太阳辐射到地面的功率大约是每平方米一千瓦,也就是每平方公里一千兆瓦。组件效率按 20% 算,每平方公里能出 200 兆瓦。

“拿这个数去除总用电量,就知道需要多少平方公里。全欧洲可以用西班牙一块不算大的地方供起来,全美国可以用犹他或者德州的一个角。当然实际上会铺开,这只是说明需要的面积很小。”(马斯克)

储能电池需要的面积更小。他给出的估算是,给整个美国储能够用的电池,占地大约一平方英里。

“跟差一点完全是两回事,差得很远。大多数人只是没做过这个基础算术。不难的,200 兆瓦每平方公里,除一下。”(马斯克)

至于特斯拉是否已经完成了使命,他分两层回答。让整个汽车行业转向电动化这件事,特斯拉成功了。当年他们刚起步时,大厂在这个方向上什么都没做,长期否定电动车的可行性。是特斯拉开始抢走市场份额,才逼着他们改主意。

让世界真正转向可持续能源,还早得很。

星舰:运力是全世界所有火箭的一千倍

火星的时间表,他给的是:无人星舰三到五年内到火星,如果顺利,载人任务在 2030 年之前。窗口每 26 个月一次,行星对齐才有一趟顺风车。

星舰的关键词是完全、快速、可复用。猎鹰 9 号已经算现役最能复用的火箭,一级和整流罩能反复使用,大概省下 70% 的发射成本,二级每次都得扔掉。

星舰两级都回收。按理论值,助推器一天可以发射 10 到 12 次,飞船受轨道限制,一天 3 次。运力约 100 吨入轨,土星五号级别;如果做成一次性版本,大概 250 吨;完全复用的前提下继续优化,150 吨是可能的。

“如果有一支星舰舰队,每天发射一次或多次,星舰送上轨道的载荷会超过地球上所有其他火箭加起来的一千倍。”(马斯克)

为什么非要去火星,他的理由带着物理学的冷感。太阳最终会膨胀,把地球上所有生命抹掉。如果真关心生命的未来,成为多行星物种、最终成为多恒星物种,唯一的方向就在这。这一步总得有人先迈。

原型容易,量产难

中国长期会占特斯拉市场的 25% 到 30%,其余四分之三在别处。

上海工厂的由来,他讲了一个细节。中国政府多次邀请他去建厂,他说不做 51% 本地持股的合资,除非规则对所有人一起改。后来规则改了,他才进来。

近期不会在中国再建新厂。重心是把柏林和奥斯汀两个新工厂推到高产量。

主持人转述了大众 CEO 赫伯特·迪斯(Herbert Diess)的问题:怎么同时在中国、德国、德州三地爬产?这在有百年经验的行业里都是硬仗。

马斯克的回答是,制造业的难度被严重低估了。他和团队 100 个人花六个月,能做出任何你能想到的车型原型。把原型变成大批量生产,是一万人干两年,而且这已经比行业快三倍。

“造车的难题,99% 在生产。原型是容易的那部分。”(马斯克)

他提到的这组比例值得单独记住。100 人六个月做原型,一万人两年做量产,中间差的东西远不止人数。一个零件可以手工打磨,一条产线不行。任何一道工序慢三秒,整条线的产能就按这个三秒来算。

他拿自己的爬产记录当底气:历史上任何造车公司、任何制造大型复杂产品的公司,都没有特斯拉这个增速。哪怕这个速度慢下来一点,也还是全球最快。

至于印尼建厂、英国建厂,他两次都把话推回去:这个场合不做重大公司公告。

自动驾驶卡在局部最优里好几年

主持人提了一个常见的质疑:AI 圈很多人认为人类水平的人工智能还很远,你怎么敢这么快部署完全自动驾驶?

马斯克的回应是先拆掉前提。开车不需要通用人工智能,不需要理解抽象概念,只需要在驾驶这一件事上和人类神经网络打平。

“任何人开几年车之后都有体会,开车的认知负荷并不高。你可以想别的事,听音乐,和人聊天,同时还能安全驾驶。所以要对齐的只是开车这一部分的神经网。”(马斯克)

他也承认这条路走得比预期慢,原因很具体:

自动驾驶是那种有很多假山顶的问题。你以为快到了,结果进度开始渐近,先是线性的,然后看起来像对数曲线,慢慢平掉,因为你陷进了一个自己没意识到的局部最优。(马斯克)

他的判断是,现在不再被困在局部最优里了。当然他补了一句,我也可能错。

纯视觉路线的争议一直在。激光雷达的支持者认为多一种传感器多一层保险。马斯克的逻辑是,人靠两只眼睛和一颗大脑就能开车,说明视觉加上足够强的神经网已经够用,多传感器融合反而增加标定和决策的复杂度。这个判断对不对,beta 视频里的表现会给出答案。

进度怎么验证?目前 beta 项目大约十万人,并没有藏着掖着。他建议别看发布会,去看 beta 用户自己发的视频,看变化发生在哪里。

时间预测是今年内达到比人类驾驶更安全的水平。更远的判断是,最终大概会是人类十倍的安全度,按受伤概率衡量。

被救的人不知道自己被救了

主持人问得直接:那些在事故中死去的人,算不算为了救更多人付出的代价?

马斯克先纠正了一个事实。特斯拉从来没有说过 Autopilot 不需要注意力,从第一版开始就在反复强调,开启时必须确认,每次开启都会提醒,提醒到令人厌烦。所谓“先承诺能自己开、后来没做到”的说法不成立。

然后他讲了另一面。

被 Autopilot 救下来的人,不知道自己被救了。他本来会撞上前面的车,或者撞上行人、自行车,但他不知道。全球每年大约一百万人死于车祸,大约一千万人重伤。就算你救下 90%,剩下那 10% 的家属还是会起诉你。(马斯克)

他接着说了自己的立场:坚信要做善的实质,做善的表象没有意义,并且极度鄙视那些把表象看得比实质更重的人。所以答案很简单,挨骂也做,被起诉也做。

隧道被低估,飞行汽车被高估

被问到个人交通的下一个创新,马斯克说隧道被低估了,隧道从不让人失望。

主持人用纽约的罗伯特·摩西举例:修了一堆高速,结果只是带来更多拥堵,隧道凭什么不一样?

马斯克说,所谓“诱导需求”是他这辈子听过最蠢的概念之一。如果加路会让交通更堵,那把路删掉不就通畅了?真删了,会有人上街抗议。

他给的解释是维度。城市里楼是三维的,人在几十层里住着,却要靠一张二维路网把人送进送出,还要在几乎同一时间完成。必然堵。往上或者往下打开第三维,路网和楼的维度就对上了。隧道不是一层,想要几层就几层,层数由把拥堵压到多低决定。

拉斯维加斯已经有验证。会展中心到 Strip 的环线在跑,接下来要把所有酒店和机场连上。当初在克拉克县投票,测试隧道只是勉强过线,等议员们自己坐了一趟,再投票就是把整个城市铺开,全票通过。

飞行汽车他不看好,理由都是具体的工程问题。噪音大,起降时的风压大。天气一差就停飞,大风、大雪、雨夹雪都上不去。天上掉东西砸到人的概率变高,一颗没保养好的轮子掉下来,谁都不想遇到。还有隐私,陌生人整天在你家后院上空盘旋。直升机除了急救基本被城市禁掉,原因是一样的。

他特意不用“飞行汽车”这个词,因为这些东西大多根本不在路上跑。

一堆短问题

最后一段是快问快答,信息密度很高,逐条记下来。

  • 比 Model 3 更小的车会不会做?这个场合不做产品发布,但有可能。
  • 会不会向其他车企授权平台?特斯拉的专利早就免费开放了,开放专利的目的是不让别人用专利雷区堵住电动车这条路。传统车企自己能搞定电动化,真正值得授权的是 Autopilot 和 FSD,大概一年后再看。
  • 哪家新公司最让人印象深刻?大众,虽然它算不上创业公司,从电动车角度看又有点像创业公司,是传统厂商里做得最多的。中国会跑出几家很强的公司,那里有大量相信制造业、肯拼的人,不是 996,是凌晨三点还不离开工厂。美国这边的人在想办法不去上班。
  • Cybertruck 会不会被福特、通用、Rivian 抢走皮卡份额?不会。订单已经排到投产之后三年。
  • 上海封控怎么看?跟中国方面沟通过,封控在快速解除,未来几周不会是大问题。当年在加州不满,是因为北美其他所有车厂都被允许复工,只有特斯拉不行,只因为工厂在阿拉米达县,既任意又不公平。
  • 软件为什么能领先?特斯拉既是硬件公司也是软件公司,他本人写了二十年软件,团队不把软件当附属品。他定期问团队,我们知道还有谁在做这件事吗,目前没看到别人在真实世界 AI 上有实质进展。
  • 超充会不会开放给其他品牌?欧洲已经开放,全球都会推。美国因为接口不同会麻烦一点,会把行业通用接口加进去。哪怕削弱竞争优势,也要做对电动化有利的事。
  • 会不会把旗下公司装进一个控股集团?不会。股东不同,目标不同。少数情况例外,比如顶尖的材料团队要求同时做火箭和车,那就让他们两边都做,反正不属于竞争领域。
  • 2020 年你跟 Kara Swisher 说,如果投资人知道你所知道的,他们会恐慌。指的是什么?记不清具体指什么了,但意思很明确,水面上的鸭子看起来很优雅,水底下的腿在疯狂扑腾。公司里永远有一串问题需要实时解决,解决了才有今天。上周日到周一他一夜没睡。
  • 特斯拉现在安全吗?没有债务,现金很多,上海有短期影响,未来极其强劲,会产出大量自由现金流。
  • 会收购其他车企吗?现在看极不可能。
  • 超级电容有前景吗?他在 Pinnacle Research 研究过两年,当时那里做出了全球能量密度最高的电容,用的是钌钽合金,钌极其稀有,无法规模化。他原本的博士方向是用芯片制造设备做分子级精度的固态电容,核心是压低电子隧穿的概率。他给出的结论是,整个汽车行业电动化,根本不需要电容。
  • 会不会买矿业公司?不排除。加速可持续能源没有禁区,如果买下一家矿企能明显改变它的轨迹,那就买。
  • 孩子会来特斯拉吗?老大们想自己做事业,他支持。最小的 X 对火箭的痴迷是另一个级别,如果这份热情从两岁持续到长大,他大概率会去碰火箭。
  • 氢能呢?这个问题他被问过一百到两百次。氢的体积能量密度太差。在宇宙里氢很丰富,在地球上完全不是。

整场访谈下来,时间表那些东西会过时,真正有长期价值的是他对失败的处理方式。他愿意把 Roadster 那段讲成一堆蠢事在燃烧,讲清楚哪两个前提错了、错了多少钱、错了多少年。这种复盘方式比任何成功叙事都有信息量。

可以带走的几条

  • 评估一家新造车公司,先追问它在小产量阶段犯过什么错、改过什么。没有这段记录的,等于还没开始。
  • 听到“把两块成熟的东西拼起来就行”这类方案,先检查两个前提能否同时成立。Lotus 底盘加 AC Propulsion 电驱,两边同时踩空,代价是六年返工和 94% 的零件重做。
  • 判断自动驾驶进展,去看 beta 用户的原始视频,别看发布会。十万人的测试池足够大,既藏不住问题,也藏不住进步。
  • 做能源测算时记住常数:每平方公里约 200 兆瓦。乘上当地日照小时数,再对比总用电量,很多争论会直接结束。
  • 把“这是愿望还是承诺”当成固定提问。两千万辆、火星时间表都适用,分清楚了,预期管理就不会崩。
  • 看社交平台的治理,只查三件事:算法有没有公开,人工干预有没有标注,永久封禁有没有被当成常规手段。
  • 做重资产制造,先接受“量产难度占 99%”这个比例,再决定起步团队是 100 人还是 1 万人。
语音识别自动转录(Whisper),含全部发言人,可能有识别错误。已按语句合并成段落,并去掉 um、uh 等口头填充词。

0:002022. We're very glad you can be with us whether you're here in the room in Central London or joining us on our global live stream. My name is Peter Campbell, I'm the global motor industry correspondent at the FT and it is a great pleasure to introduce and chair this session for you.

0:15Let me tell you how this is going to work. In a few minutes we're going to have Jamie Strabble back on stage and Elon Musk joining us remotely. But it's going to spend the first 10 to 15 minutes of the session talking about the history of Tesla. Now these two guys were central to the company and its growth and everything it has achieved to date and they have never before appeared on stage together to talk about the origins of the business.

0:37So this is a unique chance to ask them about that and hear about that. And then for the rest of the session, we're going to have time just with Elon on his own where we'll have questions about Tesla today and tomorrow and maybe one or two questions about some other projects he has going on.

0:54Just a reminder to you, please do ask your questions on the chat box or if you're in the room using Slither. But with no further ado, because of all our guests here's the one who needs least introduction, I'd like you to welcome back to the stage Jamie Strabble and joining us remotely Elon Musk. Good evening, thank you so much

1:20for being with us, both of you. I want to go back to something that was alluded to earlier in the session and the lunch that changed the world, as Patrick called us, when the two of you sat down and came up with a plan that led to Tesla. Take us through that? Who started it? Whose idea was the lunch? Where did it go? Elon, we'd love

1:42you to start. I'm sure. Well, actually, and JB, I completely, I think we could hear maybe speaking back to But yeah, I got a call someone out of the blue from JB and Rosen and I Think what I'm not sure exactly what the survey matter was but it was it's something involving space or maybe hydrogen airplanes or something and

2:15so we got together for lunch and at some point the conversation turned to electric vehicles and JB and Rose and I both worked on an electric vehicle company that I guess ultimately wasn't successful. And I mentioned that I actually had a strong interest in electric vehicles for a long time

2:41and I thought electric vehicles were the future of transport. And in fact, I had worked at a company developing high-energy density capacitors, super-altrow capacitors for potential use in electric vehicles. And that was going to be my area of study during graduate studies at Stanford, which I ultimately

3:04dropped out of. But anyway, that led to JB suggesting that I get a test drive in the T-zero prototype from AC Propaul short is a small company in Southern California. And because that sort of had a proof of concept electric sports car. And I said that would be great. I'd love to get a test drive.

3:27And that was in 2003. And then I did get a test drive in the T0. And I tried to convince them to commercialize the T0. I tried very hard actually to get AC propulsion to commercialize the T0 electric sports car. And then after they said they really did not want to do that, I said, well do you mind

3:58if then I do that? If you mind if I create a commercial electric sports car and they said, yeah, no problem. And so that my intent was to basically create a company to commercialize the T0 with JB.

4:20And then Tom Gage, AC propulsion said, well, you know, there's some other groups that also want to commercialize the T0 concept. And what do you think about teaming up with them? And one of the groups that was introduced to me was mock-topening in Wright and Martin Eberhard. But it's very important to emphasize that there was no actual company that existed in any meaningful

4:52form at that point. So it was really just three guys and then JB and me and then we decided to team up and create a commercialized, create a commercial version of the T-Zero electric sports car. Yeah, I mean, JV, what's your perspective? No, that's a pretty close history of how I remember it as well.

5:22My perspective on the thing was us trying to chat with you about this electric hydrogen airplane concept that we were, you know, I was at the time trying to work on and I drew a call that knack going extremely well actually for the first part of the lunch. I think, can't remember the colorful adjectives you used but they weren't very, you weren't very excited about the hydrogen airplane for good reason. So yeah then I think our topics, our

5:48conversation completely turned to talking about lithium ion batteries, you know, state-of-the-art of lithium ion batteries and what was possible, potentially possible, stringing together at the time, large numbers of small lithium ion batteries, which were not very mature in those days, but could be potentially

6:07connected into really big battery packs to make an EV that could potentially have hundreds of miles of range, which today seems kind of commonplace, but in 2003 was absolutely unheard of, it would literally set world records for range. And yeah, you understood that concept, I think, of the potential of it, better than anyone else that we'd ever talked to about it, and were immediately enthused about it and excited.

6:32So that was, I mean, shortly after, as you said, we met up with AC propulsion, and kind of we're off to the races. And there were a lot of early rejections when you talked to some car makers, obviously you settled on Lotus for the first Roadster, but you went around some others, was there ever a point between the lunch and getting the

6:52road straw for the ground where you almost gave up? You thought that actually you couldn't make it work, not from a chemistry point of view but commercially? Yeah, I can jump in maybe first on that. There were a lot of challenges in those early days. I mean, the technology was not a sure thing that all those pieces would work together,

7:13and some of the safety aspects around the battery were pretty thorny in the early days. And it was before, I think most people had figured out how to manage that. And I think in Tesla, I think we had an early chance to really figure that out in a very robust way and make them far safer than internal combustion actually has proved out to be through

7:34the statistics and through data. But that was not the common perception. So that was hard and that was risky. Obviously, there were a lot of financial turns and tribulations. It wasn't exactly smooth sailing from idea of an electric roadster through into commercialization. It was an insane nightmare, basically.

7:54And we screwed the pooch six ways to Sunday and made so many mistakes. It's embarrassing. Basically, the almost everything about the first design of the Tesla Roadster was wrong and wouldn't work. And the, first of all, we had no idea how to build a car. So, and then no one really knew how to build a,

8:25you know, a commercial lithium-ion battery car. No one had ever done it. So, like the original idea, sort of, it was sounds appealing, but was fundamentally flawed, was to use the Lotus Elise chassis, and then combine that with AC Propulsions drivetrain technology. And boom, you've got like, bury the two pieces necessary to create a car.

8:56It should be straightforward. And just do that, and great. Unfortunately, those were two fundamentally floor premises, the Lotus lease chassis did not work, meaning that the car ended up being 40% heavier and we couldn't fit the battery pack in the lease because it was a very tiny

9:20car. So we had to change basically everything about the lease. I think we kept maybe 6% of the parts were in common. All crash tests were invalidated, the structure was invalidated because of the greater mass and the mass distribution being fundamentally different from a gasoline car.

9:39And yeah, in retrospect, it would have actually been much smarter to start with a clean sheet design and not try to modify the Elise, because we ended up with a lot of the limitations of the Elise, but almost nothing to gain in terms of reuse of parts. So that was a staggeringly dumb decision.

10:04And then on the drive-free inside, it turned out that I think almost none of the AC propulsion technologies were viable in a production car. I mean, JB, did we end up using any of the AC propulsion stuff? I think pretty much nothing by the time it went to the first customer. Yes.

10:32It was a very helpful first proof of concept that maybe pointed the way that what might be possible, but none of it was commercializing all of it as Elon said. It was a lot of mistakes and trials and tribulations and it was very hard to build more than two of those motor controllers and the first days with analog components.

10:53Those were still analog motor control. I mean, it would get hot and the motor would perform differently. The throttle would respond differently and the hot day versus the cold day. So yeah, everything pretty much had to be redone from scratch. So go on. So what we're saying is that we're huge idiots.

11:14And we're totally wrong about the fundamental premises of creating the Tesla Roadster. And you ended up not being able to use any of the AC propulsion by trade technology using essentially almost none of the original Lotus lease, but we're still left with a lot of limitations of the lease being a car that was a bit too small,

11:38very difficult to get in and out of and very bare bones and independent on Lotus to make it and that every time Lotus would have financial difficulties, they would increase the price on us and charge us more money. So that made it very difficult to control the price of the car. And then the supply chain was also a complete mess.

12:06The initial idea was to basically outsource everything and then almost none of the companies to which we outsourced the various parts of the cars were able to succeed in making those parts. So we had to find a new supplier for the car body, because the supplier is just a supplier, isn't expect was flat out unable to do it.

12:31This is not a decision that I was building, but we had outsourced making the battery to a barbecue company in Thailand. This is a bit silly to say, okay, we're gonna make the most sophisticated battery ever made. And the barbecue company is going to make it. Obviously, they did not know how to make a company.

12:56It was supposed to be. It was pretty silly. So we had to insource battery production to California. We made it basically in our making the battery ourselves. Yeah, it's one thing after another, basically. Just a flat out burning dumpster fire of stupidity, I would say was the beginning of the company.

13:22And then we had to recover from there, and it was very difficult. So there are lots of other EV startups out there, potentially in the hundreds today, all wanting to do what you did. You guys, you blazed the trail 10 years before the industry really caught up with you. You had the advantage of having no one else playing in the same area, but the disadvantage that all the time you were trying to break through Virgin Snow, was it more difficult to do it 10 years ago than you think it is today?

13:53And if there's one piece of advice that you could offer to the various EV startups, what would it be? Well, I think most of them just completely forget about the first eight to 10 years of tussle. That's the kind of amazing part to me. They kind of see success. And then they're like, oh, that looks good.

14:10Let's do that. Let's do, you know, copy that. But, I mean, it was brutal confusion and tons of mistakes and a lot of money invested for a lot of years. I personally think it's a lot, actually harder today, even still, for a startup to compete because the first question is, what is the niche that even makes sense to have a new startup EV company relative to all the products from other OEMs or to Tesla?

14:38What are they gonna do better than Tesla? Tesla didn't exist. There was no Tesla in the world 17 years ago. So that was sort of the imperative and gave a runway to make a lot of mistakes and still make it work. You know, what might have been making? I think just, we're important to emphasize here, but you know, I think,

14:59I mean, the JBI, well, certainly I wouldn't have thought that JB would have wanted to go through the extreme pain of creating an electric car company if the traditional automakers were going to make electric cars. And, you know, it's just an important thing for the world that needs to happen to move to a sustainable technology future.

15:21And, but if at the time that we're created Tesla, There were no startups doing electric cars, and the big car companies had really no electric car programs going, and the few tiny, tiny electric car programs they had going, they were shutting down. So therefore, unless we tried to create an electric car, there was not one going to be

15:47created. It's not from the standpoint of thinking, hey, here's a super lucrative idea. makes Donner Cart Company. The history of Cart Company startups is horrific. They're almost all gone bankrupt. It's an incredibly big graveyard of car startups that all died. You've only heard a tiny number of them, sort of the tacos and the lorians of the world, but

16:16there are hundreds of others that people didn't hear about. So at this point in the United States, the only two American card companies that have not gone bankrupt are Ford and Tesla. And Tesla went back, almost went back, so many times I lost count. So basically it's a world of hurt.

16:35To start a card company is mega pain. It is not easy money. It's the furthest thing for easy money you can possibly imagine. So and what I see with some of these new card companies is that they are the jumping in at the deep end and trying to create a high volume vehicle that when they've never made a vehicle before.

17:00This is like, you know, not practicing your athletic sport and then going to the Olympics. You're not going to win. This is crazy. You really need to start out small, make your mistakes at a small scale, make sure you've got a lot of resolute capital and then gradually build up from the dumb things that you're at the beginning and be less dumb over time.

17:30Otherwise, what will happen is vast losses of money. The car industry is very competitive. It's the opposite of a natural monopoly. You have sort of natural monopolies in things like social media or say Google search. I'm going to get my card companies are naturally unnotable. They're hyper-competitive and they're used to being hyper-competitive throughout the world.

17:59They have entrenched customers, dealers, service, factories, existing expertise. These are veteran armies in fortresses. So, yeah, I mean, it's extremely difficult. Now, you happen to mention in that answer the word social media, which brings us very neatly onto the question of Twitter.

18:27So, JB, I'm going to say thank you so much. That was fascinating, a brief history there of Tesla in the early years and the mistakes that you made. So, thank you very much for being with us on stage. Thank you. Twitter front is not a lot that I can say because this is a Twitter's publicly traded company and

18:51you know, there are a lot of constraints in what I can say so I'm not sure what you would like to ask but most of the Questions like the answer is gonna be no comment. That's all right. I can have a go Let me ask you to put your product development hat on and if you pair Twitter all the way back

19:09What is it and what do you think it can be in 10 years time? Well, what I said is that I think Twitter is currently the best or looked at another way, the least bad public square. A forum for the exchange of ideas nationally and internationally. And I think it could be a lot better at that.

19:43In order to be better at that, it needs to really get rid of the bots and the scams and spammers. And basically, anyone tried to create sort of fake influence on the side by where one person or one entity operating 100,000 accounts or obviously scammers are not good. And Twitter really needs to do a much better job at that.

20:13It also needs to build trust, more trust with users. I think the way to do that is by open sourcing the algorithms so everyone can see how the algorithm works and can suggest improvements and changes. I would literally just put the Twitter algorithm on GitHub and say, hey, anyone want to suggest changes to this, please go ahead.

20:37And just really want transparency to both trust. And then any sort of adjustments to tweets or any human intervention with any account, a Twitter should be highlighted as a Twitter person took the following action with your account or with this tweet so that you're not sort of sitting then the doc wondering why did this tweet not get any attention or why did this one get a lot of attention?

21:05It's far too random. And then I think Twitter needs to be much more even-handed. It currently has a strong left bias because it's based in San Francisco. I don't think people that they're necessarily in 10, I would just have some of them don't attend, to have a left bias. They're just from their perspective, it seems moderate, but they're just coming at it from an environment that is very far left.

21:36So, but then this fails to build trust in the rest of the United States and also perhaps not the parts of the world because it sort of needs to be even handed and I think as I said publicly, it victory would be that the most far right 10% and the most bought off 10% are equally upset.

22:03I don't think this is a situation where you're gonna get necessarily a lot of praise. You're just gonna balance the anger. So how do you, because people will automatically associate you with Tesla and you with Twitter, is there any risk in your mind that the actions that you're gonna take at Twitter, which you've admitted freely, will upset some people, potentially lead to a commercial impact on Tesla?

22:32I'm confident that we will be able to sell all the cars we can make. I mean, currently the lead time for ordering a Tesla is ridiculously long. So our issue is not demand, it is production. But that's at the moment because of the global supply chains and the chip shortages. That's less around electric cardamom, which we are expecting to go absolutely through

23:00the roof and you obviously have very ambitious targets for that at Tesla. Yeah, I mean even before there was the supply chain issues Tesla demand exceeded production. So now it's demand is exceeding production to a ridiculous degree. We're actually probably going to limit just off taking orders for anything beyond a certain

23:29period of time because some of the timing is like a year away. So anyway, the frustration that we're seeing customers is under being able to get them a car, not are they willing to or interested in buying a car. So basically, I think zero about demand generation and a light about production and engineering and supply chain.

24:01I have two more questions on Twitter for me before we turn for the rest of the session to Tesla. How confident are you the deal will happen? And is there a risk? This is all one question. And is there a risk? Because you're putting a lot of your personal stake up for this to fund it.

24:19Is there a risk that if it all goes south, you're impereling your stake in Tesla, potentially hurting Tesla financially and maybe even SpaceX if it all goes to pod. That's the technical term. Yeah, sure. So I mean, I think there's still a lot of things that need to get down before the

24:41still concludes. Obviously, there's not yet even been a shareholder vote. And Twitter has not yet filed the proxy for a shareholder vote. So there are still some outstanding questions that needs to be resolved. And so it is certainly not a done deal. Just objectively it is not

25:07a done deal. The best case scenario is that this would be, I think, perhaps done in two or three months. Yeah. And the final question, and this is really the two-paid elephant in the room, are you planning to let Donald Trump back on? Well, I think there's a general question of should, what does it have, permanent bans?

25:42And I've talked with Jack Dosey about this, and he and I are of the same mind, which is that permanent bans should be extremely rare and really reserved for people where they're trying to, for accounts that are bots or spam scam accounts, where there's just no legitimacy to the accountable.

26:09I do think that it was not correct to ban Donald Trump. I think that was a mistake. because it alienated a large part of the country and did not ultimately result in Donald Trump not having a voice. He is not going to be on truth social as will a large part of the sort of the right

26:36in the United States. And so I think this could end up being frankly worse than having a single forum where everyone can debate. So I guess the answer is that I would reverse the Poma Band. Oh, say I'm not, I don't own Twitter yet. So this is not like a thing that will definitely happen because what if I don't own Twitter.

26:58But my opinion, and Jack Dorsey, I want to be clear, shares this opinion, is that we should not have Poma Band's. Now, that doesn't mean that somebody gets to say whatever they want to say. If they say something that is illegal or otherwise Otherwise, just destructors the world,

27:19then that should be perhaps a timeout, a temporary suspension, or that particular tweet should be made invisible or have very limited traction. But I think pro-avans just fundamentally undermine trust in Twitter as a town square, where everyone can voice their opinion. It was a fun, I think it was a morally bad decision to be clear and foolish in the extreme.

27:58Even after he egged on the crowd who went to the US Capitol, some of them carrying nooses, you still think it was a mistake to remove him. I think the if there are tweets that are wrong, they should and bad, those should be either deleted or made invisible and a suspension, a temporary suspension is appropriate, but

28:29not a permanent ban. So if the deal completes, he might potentially come back on but with the understanding that if he does something similar again, he'll be back in the simpen. He has publicly stated that he will not be coming back to Twitter and that he will only be on truth social.

28:52And this is the point that I'm trying to make, which is perhaps not getting a cross, is that banning Trump from Twitter didn't end Trump's voice. It will amplify it among the right. And this is why it is morally wrong and flat out stupid. Okay. Thank you. Let's turn to Tesla again.

29:18I'd like to ask you about your ambitions for the future. You've said you want the company to be able to make 20 million cars a year by 2030, which would make it the same size as Toyota and Volkswagen combined today. Give us a sense. What does the business look like by 2030 to make 20 million cars?

29:41Plants, footprints, models. Yes, well, this is not a foreign for announcing new products, new Tesla products. The $20 million by 2030 is an aspiration, not a promise. And the reason for anything from something like that is there are approximately 2 billion cars and trucks in the world.

30:06And for us to really make a dent in sustainable energy and electrification, I think we need to replace at least 1% of the fleet per year to really be meaningful. And that's where the 20 million units come from, is let's try to replace 1% of the global fleet of 2 billion cars and trucks per year.

30:30And that's our aspiration. It's not a promise, it's an aspiration. I think we've got a good chance of getting there. and people will see based on the products that we unveil, it will be able to judge for themselves whether that's realistic or not. Now you've gone from zero plants to four plants, that would require you to get to possibly about 20 plants, depending how they're ramped up.

30:59Is that phase more difficult, what you've got ahead of you by 2030, than what's gone behind, given you've already done it, or do you still think it's more difficult to ramp to that level? I think it's close. It's roughly, the share amount of stuff that Tesla's done is I think quite mind boggling.

31:28We've been an incredible team at Tesla and executing very well and our annual growth rates are faster than any large manufacturer product in the history of Earth. But I think the next fastest was the growth of Model T and we're faster than Model T. So, you know, if that growth rate continues, then obviously we will reach 20 million vehicles

31:52a year. But we may stumble and not reach that goal. So, but I'd say it's roughly equally difficult to have gotten to this point as we'll be to get to 20 million. And what's the biggest uncertainty with getting to 2030? Is it manufacturing ramping? Is it the raw materials? Is it something else?

32:29There are some raw material constraints that we see coming in the production probably in about three years. And in cathode production, the cathode, the two main cathode choices are nickel and iron phosphate. Obviously, iron is extremely plentiful. The earth is 32% iron by composition.

32:56So a little bit of trivia, if someone says, what is earth made of? The single biggest element that earth is made of is iron. The second biggest element that earth is made of is oxygen, which is about 30% of earth mass. So clearly, iron is not much in short supply. The phosphate is slightly more of a challenge,

33:22but still quite common. So I do not see any fundamental scaling constraints, and lithium is also quite common. So lithium is practically everywhere. So this is not a question of a shortage of like, as though it's some rare element, it's really just that the lithium mining, and especially the refining capacity, and that of taking iron and phosphorus and turning it into battery grade iron phosphate

33:52or nickel and turning it into battery grade nickel is, it's really the equipment, I think the single-axis trait would be the equipment necessary to convert the ore into battery grade materials. And we're working on that with suppliers, so I'm not saying that this is an impossible thing to address.

34:12it is simply a one of the problems along the way to getting 20 million vehicles. LFP batteries are much less recyclable than the other batteries you use. How much of that is an issue in terms of overall environmental impact of what the company does if you end up using lots and lots of them?

34:31No, that's not true. Iron phosphate batteries are fully recycled. Good. Well, that's easy. And so our nickel paste. I'm not sure we're getting your information, but Ironfoss fate is a really recyclable and so is a historical base matrix. Thank you. I stand corrected. Can I ask why you think Tesla needs to grow so much by 2030?

34:54Because what you set out when you in the early years was that you wanted Tesla to help tilt the world away from fossil fuels. And to some extent, it looks like it's already succeeded. Yes, I think you're sort of putting too much emphasis on this 20 million vehicles by 2030 as though this is some grand promise

35:17Some hill upon which we will die That is it is simply an aspiration and we may we may achieve it or we may not We our goal is to accelerate the advent of sustainable energy and and so That's why we want to make a lot of cars and also a lot of stationary battery packs because the three pillars of a sustainable energy future are

35:43electric transport stationary battery packs and solar and wind and Geothermal and hydro basically sustainable energy sources with solar and wind particularly are intermittent And so you need stationary battery packs to store the energy when the Sun doesn't shine So, you know, you have you constant electricity provision or when the wind doesn't blow.

36:06So, but you can have a, you can completely convert or be very clear about this. All of Earth, all of Earth can easily be powered by solar and wind, stationary battery packs, and electric transport. Well, you could power several, even if it goes to economy, you could do 10 times what Earth's

36:31economies with solar wind batteries and electric transport. It's not a close thing. I think most people just have not done the basic math of just how much energy hits the earth from the sun. It's a kilowatt per square meter or set another way, a gigawatt per square kilometer. So if you have a 20 cent efficient solar system, that's 200 megawatts per square kilometer.

36:57It's a lot, basically. You could power all of Europe with a section of Spain, a relatively electrified section of Spain. You could power all of the United States with a corner of Utah or Texas. Now, obviously, it would make more sense to spread this out. And I'm just using this as a figure of speech, or to show that it is a small amount of area that is necessary to power the United States or Europe.

37:26It is not some vast area. And I invite anyone to do the basic math. It's not hard. 200 megawatts per square kilometer. Okay, so how many square... What's the power needed? How many... That'll tell you how many square kilometers. Then you will need batteries to store that. I think our calculation of the amount area of the batteries needed to store only energy to probably the United States was roughly one square mile.

38:02Literally. So do you think Tesla has succeeded in its goal then? Or do you think there's still much more to do? No, I think that we've not succeeded in the goal. If you consider the goal to be getting the automotive industry to move strongly towards electric vehicles, I think that part of the goal we have succeeded in.

38:30And that was explicitly by our goal is to get the industry moving towards electric vehicles, because they were doing nothing in that direction when we started. And for longest time, they were dismissing the concept of electric vehicles. And then Tesla started taking market share away from them.

38:50And that changed their mind. How long do you think you're likely to stay at Tesla? As long as they can be useful. And what else is there in terms of potential future projects that piques your interest? You're obviously going to be quite involved in Twitter if the deal goes through.

39:11But you've got Tesla, you've got SpaceX, you've got boring, you've got Neuralink. As you're looking around thinking you potentially have capacity, what else is there in terms of the kind of, I don't know, betterment of the human condition or improving Earth that you feel you might want to turn your attention to in the future?

39:32Well, I'm trying to take this out of actions that I think most likely make the future good. And hopefully not pave the road to hell with good intentions. So I think Tesla's about accelerating sustainable energy because we're already near sustainable energy future on Earth, for Earth to be good.

39:52Then SpaceX is about extending life beyond Earth so that we may become a multi-planetary species and with Starlink providing internet coverage to the least served in the world. Those are the rural communities or places that just don't have good internet, or it's very expensive. And I think it's important that we become a multi-toned species

40:19and a space-brained civilization, because eventually the sun will expand and destroy all life on Earth. So if one is a true environmentalist or cares about the future of life, it is obviously important that life become multi-planetary and ultimately a multi-stellar. So we must make that first step.

40:43What do you think the next goals are then for SpaceX? And do you have a date in mind for when you think they will get to Mars? I think we should be able to, I don't know, maybe get Starship to Mars uncrewed in three to five years. And then I think if that's accessible, then we may be able to send a crew admission to Mars

41:23before the end of the decade. The windows for going to Mars occur every two years. of the planets align for a total transfer every 26 months. So that's, you get a kick at the can every 26 months. So we're making a lot of progress with Starship. We will hopefully have our first launch attempt in the summer or the next two or three months. And we're

42:00We're building up the factory for scholarships. So there's a whole bunch of startups coming behind the one that will attempt launch soon. So if that one doesn't work, we'll have many more behind that. And we're continuously improving the design. So each one is better than the last.

42:21And a game changer to an extraordinary degree with respect to access to space, because it's the first orbital rocket that is designed to be fully and rapidly reusable. There's never been a fully reusable orbital rocket. Falcon 9 is the most reusable rocket since we're able to fairly rapidly reuse the blue stage and the ferrying, and that's maybe 70% of the other cost

42:53a permission is saved. But we still lose the upper stage on every mission. So with Starship, both the booster and the upper stage, both the ship and the booster will be recovered. And they'll be recovered very rapidly. And in theory, you could launch the booster 10 or 12 times a day and the ship, because of

43:21orbital limitations, maybe you could launch it three times a day. And it's also a very big vehicle. So on the order of roughly 100 tons, 100 tons of useful out-to-orbit. So it's Saturn V scale payload to orbit, but fully reusable. If it was expandable, it would probably do under 250 tons to orbit.

43:48And I think with further improvements, even with full reusability, you can probably get to 150 tons to orbit. To put this into perspective, if you have a fleet of starships, launching with the ability to launch every day or multiple times a day, Starship will do more than a thousand

44:09times the payload to orbit of all other rockets on Earth combined. Can I thank you, that was fascinating. Can I bring us down to Earth gently and ask you about China in relation to Tesla. How important is the market for you? Do you think China contributes most of your growth in the future?

44:36No, I think China is obviously a very significant market, but it is probably 25% to 30% of our market long term. The rest of the world is probably three quarters of it. Do you see, I promise I didn't have any more Twitter questions, but this is sort of a Twitter question. Do you see any risk at all that China uses your ownership of Twitter, potentially to interfere or block Tesla's operations in the country, because obviously China has banned Twitter.

45:13I've seen no indications that effect. Okay, how close would you say you are personally to the Chinese government? Because when you set up the factory in Shanghai, the wolves were rewritten around joint ventures to allow you to do that? Well, yeah, I was suddenly asked many times

45:36by the government of China to do a factory in China. And I said, well, we're not going to do one, which is 51% locally owned. And so if they're willing to change the rules, not just for us, but for everyone, then we would move forward. And so they did. And I think it's been very successful so far.

45:56And the government's very happy about it. And so I don't know if things are proceeding fairly well. How many other plants are you expecting to open in China in the near future? Well, we're not expecting to open any additional plants in China in the near future. We will be expanding our Shanghai factory,

46:21but our focus on production is gonna be on the two new factories that we've recently completed in Berlin and in Austin, Texas, and getting those to high volume production is the near-term challenge. We had Herbert Deese, Volkswagen chief executive, who you know well at the summit yesterday.

46:43We asked him what his big question was for you, and it was how do you scale production in China, Germany and Texas simultaneously because for any experienced area, and even the ones with 100 years of doing this, that's a challenge. Yes, I think generally the difficulty of manufacturing and scaling-age factoring is very much under-appreciated.

47:08I've said many times that prototypes are easy, production is hard. We can easily whip out a prototype of any kind of card you could possibly imagine with a small team in six months. Now, we have a team of 100 people in six months. Now to bring that to fruition with high volume production is 10,000 people in two years and

47:31that's three times faster than the rest of industry. So overwhelmingly the problem with cars is production. Overwhelmingly it's 99% of the difficulty. So he's certainly correct that the difficulty is scaling. But if you look at our track record thus far in scaling production, we've scaled production

48:01at, like I said, a rate faster than any card company on the history of Earth or any company making a large sophisticated product of any kind. So provided we continue to execute at that rate, we'll do fine. even if that rates slows down a little, it'll still be faster than any other company on earth.

48:22Now I want to ask you about two specific rumors that have swirled around future production plants for you. Are you gonna build a plant in Indonesia? As I said, this is not the foreign price to make major company announcements. And finally, have you looked at building a plant in the UK?

48:42Do you think you might ever build a plant in the UK in the future? I would throw it back to my prior answer. It was worth a try, right? Can I ask? Can I ask about autonomy, right? So you have this way that you think about autonomy using cameras on the vehicles and you want to get to almost a kind of human level AI to get the cars to drive themselves

49:09fully in the real world. But a lot of people who work in AI think human level AI is a long, long, long way off. How do you square that circle about the approach that you're taking and the speed at which you want to deploy full self-driving technology? Well, I don't think you need full human level intelligence driver car.

49:33You need sufficient human level intelligence to drive a car, but not, you don't need like deep conceptual understanding of esoteric concepts or anything like that. need sufficient intelligence to match what human neural nets do when driving a car. And I think, you know, anyone who's driven a car for any length of time, once you know

50:01once you have some years of experience driving a car, the cognitive load on driving a car is not high. You're able to think about other things, listen to music, have a conversation, and still drive safely. So it's not like matching everything a human does. But it is still matching enough of the neural nets

50:23to what, you know, the silicon neural nets need to at least be on par with the biological neural nets to enable safe driving. And I think we're quite close to achieving that. But I recommend that, don't take my word for it, just sign up for our beta program and try it out. Or look at the videos that people are posting who are in our beta program.

50:49We have like, I think at this point about 100,000 people in our beta program. So it's not exactly top secret. How hard is it as a problem to solve? Because you've made predictions in the past about autonomy. And, you know, some of those haven't come to fruition at the same time.

51:04Has your understanding of the problem of autonomy changed over the last few years? Yes, I've seen that self-driving is one of those things where there are a lot of full stones, or where you think you're getting there, but then you end up asymptoting, your progress is initially linear and then looks logarithmic and sort of tapers off because you're in a local

51:38maximum that you did not realize you were in. Now at this point I think we are no longer in a trapped in a local maximum and obviously I could be wrong but I think we are actually quite close to achieving self-driving at a safety level that is better than human and It appears, my best guess is that we will get there this year.

52:08But we're really not far from it. And like I said, the best way to assess this is to be in our beta program. We'll look at the videos of those who are in the beta program and look at the progress that has occurred and the progress is dramatic. And I'm confident we will not really get to the safety level of human.

52:30will get far in excess of the safety level of human. So I think ultimately probably a factor of 10 is safer than a human, as measured by the probability of injury. Given where you're trying to get to with this, the factor above human driving, the potential lives that could be saved with that, I need to ask you about some of the accidents and fatalities that have happened with vehicles previously.

52:57the people who died, do you consider that that was a price worth paying to get to the level where we want to be to save more human lives in the future? Well, it is important to note that we have never said ever that Tesla autopilot does not require attention. We have always made that extremely clear repeatedly.

53:25You can't even turn it on without acknowledging that it requires a supervision. We remind you of that every time you turn it on to ad nauseam. So this was not a case of setting expectations that the car could simply drive itself in the past and then not meeting those expectations.

53:45That is completely untrue. Now, it is also the case, however, and I knew this would be true from the beginning, that people don't, the people whose lives are saved with autopilot or autonomy don't know that the lives were saved. And so, you know, if you look at, say, deaths, annual automotive deaths every year, it's

54:15around a million people per year die from automotive accidents, maybe 10 million per year are severely injured. And so with autonomy, the cause of driving, it's assisted driving right now, but it will be fully autonomous in the future. Like I said, those who didn't realize they would have crashed or hit a pedestrian or cyclist, they don't know that.

54:42So basically even if you, let's say, save 90% of the people that otherwise have died,

54:55the remaining 10% who did die will still sue you. Now, but I think in the grand scheme of things, what is the morally right decision? And I am a strong believer in doing the reality of good over the perception of good. And I have utter contempt for those who prefer the perception of good over the reality of

55:21it. And so we're just going to take the heat. But if we believe that the polity of injury is reduced, and we are very confident with that and but we also know that we're going to be sued despite doing the right thing, we will do the right thing and get sued. Thank you. We have, it will not necessarily surprise

55:49you to hear a lot of audience questions that have come in. In the last few minutes I'm going to try and rattle through as many as we can. What is Tesla's approach to smaller and more affordable end of the market? You're going to go smaller than the Model 3 but could you get into a scooter micro mobility something else? Scooters are very

56:10dangerous I would not recommend anyone drive a scooter if there's ever an argument between a scooter and a car the scooter will release. What about smaller than a Model 3? How far can you go down the entry level of the market? Well like I said this is not some forum around about the product

56:33announcements, no matter how, no matter how cleverly the question is asked. So, you know, will we, there's some probability that Tesla will do a smaller called the Model 3. I don't know what's in one note. Okay, that's fine. That wasn't my question. That was an audience question.

56:57Would you ever consider licensing your platform to other OEMs? Presumably that would help switch the industry towards electric mobility in your opinion? Well, we've already opened source oil patents. Anywhere can you use patents for free? We only patent things in order to prevent others from creating this minefield of patents

57:21that inhibit progress with electric vehicles. But several years ago, I came from the conclusion, we're never going to really prosecute anyone for using a patent. So let's just say you can use any Tesla patents for free. So I think hopefully that's helpful to others. But I think the regular car industry, the traditional

57:45car makers, will solve electrification. It's not fundamentally difficult at this point to make electric cars. The thing that I think they may be interested in licensing is Tesla the autopilot full self driving. And I think that would save a lot of lives. But I think we still have, we have to prove ourselves for I don't know maybe another year or so.

58:14And then perhaps there will be some other comicers who may wish to license Tesla autopilot. It would be very open to that. Thank you. Of all the other EV startups, which one has impressed you the most? Well, I think the company making the most progress besides Tesla is actually VW, which is not a startup, but could be viewed in some ways as a startup from an electric vehicle standpoint.

58:48So VW is doing the most on the electric vehicle front. I think there will be some very strong companies coming out of China. There's just a lot of super talented, hardworking people in China that are strongly believe in manufacturing and they won't just be burning the men out at all, they'll be burning the

59:163 a.m. So they won't even leave the factory type of thing, whereas in America, people are trying to avoid going to work at all. We've got a question on the cyber truck. Is there a risk that you lose the pickup segment to Ford, GM and Rivian, if you don't launch the cyber truck or get it to market soon?

59:43No. Good. We've had another question come in. What orders the first cyber truck then we could possibly fulfill for three years after start of production? Okay, thank you. We've had a question coming about China. been critical of lockdowns in the past, particularly when it happened in the US, Shanghai is

1:00:04currently locked down. Most of the Western world is able to carry on functioning at the moment because of vaccines, but China is going towards a zero-covid approach. What do you make of the Chinese government's actions? Well, we've had some conversations with the Chinese government

1:00:24in recent days and it's clear that the lockdowns are being lifted rapidly. So I would not expect this to be a significant issue in the coming weeks. In the past where I was sort of upset with lockdowns is where those lockdowns deferentially affected Tesla but not others. So in the case of California and the Bay Area County specifically, every other car factory in North America was allowed to start, but not to Tesla.

1:01:00Even though there was no basis for that, it was simply because we were located in Alameda County in California. And but they had no rush. It was arbitrary and unfair. And that's the reason for why we're quite upset about Tesla being singled out as the only car company in America that was allowed to start even though I think our health care

1:01:27practices are probably better than anyone else. Thank you. We've had a question on software. What point did you realize that software was going to be key to Tesla's success and particularly as others are coming into software, how do you maintain leadership in it? Well, Tesla is, I think as much a software company as it is a hardware company.

1:01:49I personally wrote software for 20 years. So I have great respect and admiration for software engineering. I think it's incredibly important. And I think that's why Tesla's been able to recruit some of the world's best software engineers is because we value software engineering so highly and do not regard it as an afterthought.

1:02:18So our autopilot, full self-driving AI team is incredibly talented and some of the smartest software engineers in the world. And I do not give compliments lightly. So we have an incredible team. We have the best real world AI team on Earth. And we're seeing more and more incredibly talented people join our team.

1:02:52So, and we're really not seeing anyone else being close to solving real world AI. They may exist, but if so, they're being very subtle about it. And I periodically poll our team and say, like, do we know anyone who is doing this because We know who a lot of the top people are in AI and what they're doing and we just don't

1:03:18see anyone else that we're aware of making any significant progress in real world AI apart from Tesla. So yeah, I am quite confident that we'll not merely maintain our software lead, but it can increase it. Thank you. We've had a question come in on the Supercharger network. What are your thoughts about opening it to other motorists who don't drive Teslas?

1:03:48And doesn't that risk what is currently one of your great competitive advantages? We've already opened Tesla supercharges to other electric cars in Europe, and we intend to roll that out worldwide. It's a little trickier in the US because we have a different connector than the rest of industry, but we will be adding the rest of industry connectors as an option to superchargers

1:04:17in the US. So we'll be fully, we're trying as best as possible to do the right thing for the advancement of electrification, even if that diminishes our competitive advantage. Do you think you'll ever wrap all your various operations, Tesla, SpaceX, etc., under one umbrella group or do you want to keep them separate for the time being? I think there's sort of

1:04:45separate objectives with different shareholders of ACES and I think you know I don't see a ton of merits in combining them. At times there are people you know where we have some very talented people who actually are willing to join but they want to do things both at SpaceX and Tesla. So for example, we've got one of the best advanced materials teams in the world.

1:05:19It might be the best, but it's certainly one of the best. And a lot of people on that team were willing to join, but only if they could work on both rockets and cars. And so it was like, great, let's do that. And so we can share some of the ideas between rockets and cars with trust, you know, they're not different competitive segments. So, you know, if somebody's

1:05:47a really incredible technologist, innovator, engineer, they want to work on interesting things. So the more interesting, you know, like, sure, like, monies, you know, they can get money from, you know, anyone would hire them for a lot of money. So then it's not a money thing, really just how interesting are the projects. And so there are just a few cases where we

1:06:15can recruit some of the smartest, you know, engineers, scientists, technologists in the world, but they want to work on both rockets and cars. And there's a few cases like that. Thank you. I know we've run over time, but one more I'm going to try and get in and then if you need to go, go. But if you're happy to keep talking, we'll keep you. What is the

1:06:37What do you think is the next big innovation in personal transportation? Well, I think tunnels are underrated. I'm underappreciated. Tunnels are never let you down. Oh, well, there. Here's the question on tunnels, right? So if you look at, say for instance, Robert Moses in New York built loads of highways,

1:07:01they were supposed to solve congestion, and all they did was lead to more congestion. How do you avoid tunnels doing exactly the same but just being very expensive in the process. I have to say this this notion of induced demand is one of the single dumbest notions I've ever heard my entire life. If if if you know if adding roads

1:07:24just increases traffic why don't we delete them and decrease traffic and I think you'd have an uproar if you did that. The the real problem is that we have not if you take say congested cities, which really almost all large cities are congested, you have a fundamental dimensional problem. You have say these 12 buildings or multi-level

1:07:47buildings where you've got people living in 3D and then you want to take them in and out of those buildings on a 2D road network. Like how would you possibly expect that to work? So especially if they want to all arrive and depart at roughly the same time. This is just a recipe for traffic, obviously.

1:08:11So now if you go 3D, which you go 3D up or you go 3D down, now you have, you're matching the dimensionality of the buildings. The buildings are 3D and if the road network or you could potentially have flying cars, if 3D then you will completely alleviate the traffic problem. So think of tunnels not as a single layer of tunnels, but as many layers as you want.

1:08:35And whatever layer count is necessary to drop traffic to play mind that people wouldn't think it would work. And we've already, we have a proof of concept of this in Las Vegas with the tunnel going from the Convention Center to the Strip. And that we'll soon be connecting all of the hotels and the airport in Las Vegas.

1:08:59and people just try it out for themselves. It's working really well already in Vegas. And there was some skepticism among the county and the city as to whether it would be effective. And I think the test tunnel just barely succeeded in a vote with the local government in Clark County

1:09:22and Las Vegas. But once they saw the initial test tunnel and wrote it themselves, we got a unanimous vote in favor of expanding it to the whole city. So that should tell you something. Could you ever potentially go in the other direction three-dimensionally and look at V-Tolls and potentially flying vehicles?

1:09:47Well, I like the idea of V-Tolls, but we already had V-Tolls in the form of a helicopter. helicopter. So the problem with going 3D in the air is that you now have things that make a lot of noise and the wind force that they generate when taking off landing is very high. I mean, if you just say like look at a little drone and say, and I imagine that

1:10:20that thing was big, what a record it would make and how much wind force would it generate. And now they're going all over the place, like a giant beehive of super noisy bees. I don't think people, this is what people want. And most cities, helicopters are actually banned except for emergency purposes because of the reason.

1:10:42So then there's also the weather dependency. So if there are high winds or heavy snow, rain, sleet, now you can't fly. So now you're shut down and can't go anywhere. Then there's the probability of something falling on your head is much higher if there are all these, you know, beetles flying all over the city.

1:11:11I mean, I think people's comfort level be quite reduced. Should someone have perhaps not properly serviced their flying car and drops a wheel on your head? I know that I think would be discomforting to most people. And also having them fly over your backyard and having strangers stare at you all day is

1:11:33probably also discomforting. So I think these are all reasons why I think details will not succeed. I deliberately avoided calling them flying cars because most of them don't drive on the road but no you're right but apart from that they're great. You told Cara Swisher in 2020 that if Tesla investors knew what you knew they would panic about the

1:11:55state of the company that was two years ago so what were you referring to at the time? I'm not sure what I was referring to at the time. 2020. I mean, I think it's not that Tesla investor should panic. It's just that the sheer number of problems being sold behind the scenes, I think, would blow their minds.

1:12:25And it's like that analogy of you see that the duck is serenely moving along the pond. the, meanwhile, the legs are patting like crazy underneath. This is really what's happening pretty much all the time. So there are a nonstop series of issues that need to be solved with Tesla, or with SpaceX,

1:12:51or really with most companies, I suspect. And now we do solve them. But I think it would be very scary to most people to realize just how many things have to get solved on a real time basis of the company to function well. Is that it? Just last night, I basically got no sleep. I was up on Sunday night until, well, I really just didn't get any sleep until last night.

1:13:21So what is... I was socially working all night just Sunday through Monday. What is paddling under the water at the moment? something going on that makes you worry about the future of Tesla? Do you think the company is now completely secure? No, I think the future of Tesla is extremely strong. So Tesla

1:13:49has no debt, it has a lot of cash, and there's a short term pick up with the COVID restrictions

1:14:03in Shanghai. But as for the best of my knowledge, the future of Tesla is incredibly bright. And I think we will throw off a tremendous amount of free cash flow. Do you think with your cash and obviously your market cap at the moment, you would ever consider buying another car maker, whether an established area more in new business?

1:14:27Well, I think it's highly unlikely now. Great. Finally, we've had a couple of questions coming on supercapacitors. Do you think they might be potential, have potential as an alternative energy storage solution to lithium ion batteries? Well, it's interesting. Supercapacitors or ultracapacitors are oddly enough.

1:14:53I did research in that for a couple of years while doing my physics degree. out of company in Silicon Valley called Pinnacle Research, which at the time at least made the highest energy density capacitors in the world. Now they used a rhe opinion tantalum alloy which is very rare, rhe opinion especially

1:15:18is very rare. So that would not scale to high volume. This simply isn't not enough refi-fennium. So, but I thought about the problem quite a lot and in fact, my primary goal had I continued as a graduate student and done at PhD at Stanford, would have been to try to figure out how to build ultra high-energy SD capacitors

1:15:49And the sort of theory I had at the time was to use advanced chip making equipment to build solid state capacitors that would be accurate at the molecular level. And this is a little bit of, not new physics, but exploring some areas of quantum mechanics quantum mechanics because you have to minimize the probability that electrons can sort of tunnel from

1:16:24one, you know, essentially tunnel across the insulator from the conductor to the insulator. So I guess things get quite complicated and rather weird at the sort of single or when you're getting to get to a small number of atoms. But anyway, along to kind of long story short, I think the capacitors are not needed at all

1:16:52for electrification of the entire auto industry. Thank you. We've had a question on mining again. I know what I'm talking about. I know that. Don't worry. Kurt, we've had a question on mining. So obviously, you've done a number of mining deals, trying to secure raw materials. Do you think you could ever go a step further and actually buy a mining company in the future?

1:17:20It's not out of the question. We will address whatever the limitations are on accelerating the world's transition to sustainable energy. It's not that we wish to buy mining companies, but if that's the only way to accelerate the transition, And then we will do that. There are no arbitrary limitations on what's needed to accelerate sustainable energy.

1:17:49We'll just tackle whatever set of things are needed to accelerate sustainable energy. And doing mining and refining or mining company, provided we think we can change that mining company's trajectory significantly are possibilities. Yeah. Do you, this is a slightly personal one.

1:18:12Do you ever think your children could end up working at Tesla, obviously not the new one. She's only a few months old, but the older ones. Yeah. Well, my older children have said that they wish to do their own thing, which I support. That none of them has said that they wish to work at Tesla or at SpaceX.

1:18:35They want to do their own thing, and obviously I support them doing that. I think they wanna do something for themselves and make their own way in the world and that's fine. And it's possible at some point in the future that perhaps they do something of themselves and then decide that they'd like to do something at Tesla SpaceX or that, but currently they wanna do their own thing.

1:19:02I would judge by a little act. I know he's seen a kid love rockets more than he does. It's next level. He likes it. I mean, it's really next level. So if his love of rockets continues from age two to when he's older than he may be interested, he will be interested in working on rockets.

1:19:25Perfect. Thank you. Do you have time for one more question? Sure. Excellent, good. You talked earlier about things that are abundant in the world, iron's abundant, oxygen's abundant, and hydrogen's also very abundant in the universe. In the universe, but not on Earth. Not on Earth. Have you ever considered, do you think that it has a role to play if

1:19:54we really want to accelerate the transition away from fossil fuels? No. I really can't emphasize this stuff. The number of times I've been asked about hydrogen, it might be like, it's well over 100 times, maybe 200 times. It's important to understand that if you want, a means of energy storage, hydrogen is a bad choice.

1:20:21It is as extremely low energy density. Sorry, it's extremely low density. And so it's actually, it's volume metric energy density is poor.

资源与出处

相关视频

同频道、同主题或相邻年份