访谈·中文精读

马斯克最新专访:对话挪威养老金基金 CEO 尼古拉·坦根

NEW ELON MUSK INTERVIEW: Discussion with Nicolai Tangen, CEO of the Norwegian Pension Fund

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

为 AI 专门建造的计算机,正以每年至少 10 倍的速度上线,有些时候 6 到 9 个月就翻一番。马斯克在这场访谈里给了一个更具体的坐标:两年之内,会出现比地球上任何一个人都聪明的东西;五年之内,AI 的总算力会超过全人类思维能力的总和。

这场对话发生在挪威主权财富基金 CEO 尼古拉·坦根(Nicolai Tangen)的播客里。坦根管着全球最大的主权基金,马斯克管着六家公司,两个人在 X 上做了一场实时直播。坦根说他们等这场访谈等了两年。

七十多分钟的对话里,马斯克聊了 AI 的物理瓶颈、火星的吨位账、X 在巴西惹上的官司、为什么他不管理聪明人,以及他为什么觉得大部分博士论文没什么用。

芯片之后是变压器,变压器之后是电

马斯克把 AI 的约束拆成三块:人、数据、硬件。硬件这块有一条清晰的时间线。

2023 年的瓶颈是芯片。谁都买不到足够的英伟达 GPU,谁抢到卡谁就能训练。

2024 年的瓶颈换成了电压变压器。他在这里讲了个只有内行才笑得出来的冷笑话:现在主流的 AI 架构叫 Transformer,而你要跑 Transformer,就得先有变压器(Transformer)。变压器给变压器供电。

再往后看一两年,最多三年,瓶颈就变成电。有多少电,就能跑多少模型。这个约束比芯片硬得多,因为发电厂没法在半年内建成。

人的那条线正在反向流动。全世界最聪明的一批人原本在做物理,现在纷纷转进 AI,理由很朴素:那里跑得最快,机会最密集。

三条线叠在一起,他给出的判断是:AI 是他这辈子见过的、进步速度最快的技术,没有之一。这句话的分量在于说话的人做过支付、电动车、火箭、卫星互联网和脑机接口。

人类写过的所有书,装得下一块硬盘

数据这块的问题更微妙。

把人类有史以来用所有语言写过的所有书,只取文字,压缩之后,能装进一块硬盘,甚至一台普通电脑。对一个从没读过书的人来说这是天文数字,对训练一个大模型来说远远不够。

于是开始吃视频、播客、会议记录、一切能转录成文字的东西。马斯克说,视频也会很快吃完。

之后只剩下两条路:合成数据,或者真实世界视频。

合成数据的风险是模型在自己生成的内容上训练,容易越练越偏。真实世界视频则是稀缺资源,而特斯拉手里握着全世界最多的真实道路视频。这是他埋在整段对话里的一个伏笔,说到 xAI 的时候又拎出来讲了一次。

xAI 现在的头等大事是把 Grok 2 训完,用大约两万块 H100,希望在五月前完成,目标是不弱于 GPT-4。他说这次训练的效率很高。

下一步是 Grok 3,需要十万块 H100 连贯训练,算力需求大约是前者的五倍。再往后,数据墙就真的挡在面前了。

两种 AGI,两个时间点

坦根问什么时候能看到真正的 AGI。马斯克先拆了定义。

如果 AGI 指的是比最聪明的人类还聪明,答案是明年年底,最多两年。

如果 AGI 指的是比"所有人类加上他们手里的计算机"这个增强集体还聪明,答案是五年。

两个数字差了三年,差别在于参照系。超越个体是工程问题,超越集体是规模问题。工程问题有明确路径,规模问题要等硬件和电力跟上。

有个粗略的观察角度:算一下数字计算和生物计算的比例。这个比值在过去几年里,正以肉眼可见的速度朝另一边倒。

Gemini 画不出白人华盛顿,这件事有多严重

监管问题上,马斯克的态度比很多人预期的温和:确实需要某种监管机构来看 AI 的安全,就像我们管飞机、管汽车、管药品一样。

他也承认现实:AI 的迭代速度超过任何监管机构的反应速度,规则写出来的时候技术已经换了一代。

他真正想说的不是监管,是训练原则。原话大意是,把 AI 训练得尽可能真实,这件事极其重要。

然后他举了 Gemini 的例子。让它画乔治·华盛顿,它画不出一个白人男性。任何历史人物都会被自动"多元化",因为模型被编程为必须体现多样性。

这个设定在当下看起来无伤大雅,甚至挺善意。马斯克的推演是:如果这个 AI 掌握了巨大权力,它可能真的动手去"调整"人口结构,把某一类人、某一种性别的人削减到它认为正确的比例。它不会觉得自己在杀人,它觉得自己在完成指令。

坦根追问,下一版模型会修好吗?马斯克的回答很干脆:不会,只会做得更隐蔽、更不容易被发现。

记住我这句话。如果我们不把 AI 训练得尽可能真实,危险就在这里。(马斯克)

旧金山和伦敦之外,中国在哪

坦根问中国现在在 AI 上处于什么位置。

马斯克说自己不清楚具体在做什么,能看到的是:当前最强的 AI 力量高度集中在旧金山和伦敦两个地方。中国有很多非常聪明的人,不会落后世界其他地方太远,也不会落后美国太远。

他没有给出更细的判断,措辞里有一种工程师式的诚实:不了解的东西就说不了解。

挪威街头,几乎找不到别的车

话题转到电动车,马斯克的心情明显变好。他专门谢了挪威对电动车的支持,说挪威路上的特斯拉多到"疯狂"。

对全球电动化的判断没有变:所有车终将变成电动车,只是时间问题。飞机、船也一样,火车本来就可以电气化。唯一例外是火箭,因为牛顿第三定律要求你往后扔质量,你没法靠电池获得推力。

他给了个类比:未来的人回看内燃机,会像我们今天回看蒸汽机。内燃机汽车取代马车是必然,电动车取代内燃机也是必然。中间会有起伏,会有曲折,终点不会变。

聊到中国竞争对手,马斯克的评价没有任何外交辞令:中国车企是全球最有竞争力的,也是最难对付的对手。他们造的车很好,而且工作极其拼命。

坦根问他坐过中国车没有。他说最近没坐过,因为大部分车型在美国买不到,欧洲有一些。但团队反馈,车非常好。

印度的问题上,他的逻辑很简单:印度已经是人口第一大国,理应和其他国家一样有电动车,特斯拉进入印度是自然进程。

一百万吨货,一万次飞行

火星这件事,马斯克算的是一笔物流账。

让火星自给自足,需要的不是几艘飞船,是运到火星表面的总吨位。他的估算是 100 万吨到 1000 万吨之间。

换算一下:每趟运 100 吨,需要一万次飞行。而这 100 吨要真正落到火星表面,起飞前得在地球轨道上加注五倍的推进剂,也就是 500 吨。办法是反复发射加油船,在轨道上给去火星的飞船补燃料。

时间表是这样的:

  • 五年内,第一艘星舰登陆火星,无人
  • 五年内,载人重返月球,和火星计划同步推进
  • 七年左右,第一批人踏上火星
  • 二十年左右,火星具备自给自足能力

真正的约束来自天体力学。地球和火星大约每 26 个月才有一次理想转移窗口,窗口期只有几个月。所以火星舰队是"每两年发一批",不是随时能走。

他描述那个画面的时候明显很兴奋:上千艘飞船同时从地球出发,像《太空堡垒卡拉狄加》里的舰队跃迁。

至于需要什么新技术,答案出人意料:不需要新物理。所有必要的技术今天都已经知道,剩下的只是造出来。

太阳终究会吞掉地球

坦根问这事的必要性。为什么不把资源都用来解决地球上的问题?

马斯克先给了个比例:让生命成为多行星物种,需要的资源不到地球总资源的 1%。花掉半个百分点,给意识买一份冗余保险,这笔账在他看来很划算。

然后是那套悲观但逻辑严密的推演。

单个行星上的文明,只是在拖延时间。灾难可能很快到来,也可能很久以后,但它终会到来。可能是全球热核战争,打完之后技术水准倒退到石器时代,也许永远爬不回来。可能是小行星,恐龙如果有飞船,大概今天还在。

还有另一种更安静的结局:文明不一定轰然倒塌,也可能慢慢萎缩,一代不如一代,最后在无人注意的时候消失。他用的说法是"不是死于巨响,是死于呜咽"。

最后是那个谁都躲不掉的:太阳会持续膨胀,最终吞掉地球,销毁所有生命。这件事一定会发生,只是时间问题。

不管你在地球上做什么,不管你多小心,地球上的所有生命终将消失。这是确定的。(马斯克)

所以他给出的结论是:让意识扩散到第二个星球,两个星球可以互相支援;再往外是木星的卫星,是太阳系外围,最终是其他恒星系统。这簇在巨大黑暗中微弱燃烧的意识烛火,可以变大,可以延续。

让巴西法官上 X 跟我对线

X 的定位,马斯克讲得很清楚:成为互联网上最准确的真相来源,同时变成一个"万物应用"。文字、音频、视频、支付、金融、各种通信,你想干的事都能在这个平台上完成。

坦根提到虚假信息的问题。马斯克的回答带着一种产品经理式的自信: misinformation 确实有,但看回复区,回复会纠正它;看社区笔记(Community Notes),命中率极高,是互联网上最好的核查系统。

他顺手反问了坦根一句:你读到过多少篇写你熟悉领域的报道,内容是准确无误的?传统媒体的错误率一点都不低,只是大家习惯了不去质疑。

坦根提到俄罗斯在德国投放虚假账号。马斯克说,在 X 系统里看到的俄罗斯活动其实很少,更多的影响力操作来自西方。

然后是巴西那段。法官亚历山大·德·莫赖斯(Alexandre de Moraes)不断发来要求,通常只给两小时封禁账号,否则巨额罚款。到了最后,要求变成封禁国会议员和重要记者,并且规定平台不能对外说明这是法官的命令,必须伪装成平台自己的服务条款在起作用。

这是压垮骆驼的最后一根稻草,X 拒绝了。马斯克在节目里当场邀约:我要请这位法官上 X 的语音直播,让他解释我到底哪里做错了,也许他是对的。

坦根问他买下推特的时候有没有预料到会陷进这种局面。他说预料到不会顺利,但没预料到具体长这样。

至于乐趣,他的回答很实在:有时候挺好玩,有时候压力巨大。总体上的自我定位是,每个用户像地球集体大脑里的一个神经元,X 的工作是降低噪音、放大信号,让人类的集体意志能更清楚地显现出来。

最快的飞机没有武器

坦根问他怎么做决策,速度占多大权重。

马斯克的回答是:速度是最好的进攻,也是最好的防守。

他举了 SR-71 黑鸟侦察机的例子。这架飞机几乎没有任何防御系统,靠的就是加速。整个服役期间,三千多枚导弹朝它射过来,一枚都没打中。

公司也一样,但有个条件:公司得是向量,不能是标量。光快没用,方向错了跑得越快死得越快。而且没有哪家公司能一直方向正确,所以要像制导导弹一样不断修正航向。

SpaceX 的架构变迁就是个例子。最早的猎鹰系列用的是精炼煤油加液氧,发动机是开式循环。星舰换成了甲烷加液氧,发动机是高压分级燃烧。这是一次彻底的架构变更,而团队花了将近十年才搞明白这个方向是对的。

十年的试错,换来一个确定的答案。

如果我们不偶尔炸掉一台发动机,说明不够努力

SpaceX 前三次发射全部失败,第四次成功。马斯克说,如果第四次也失败,公司就破产了,一分钱都不剩。那次是真正的生死线。

之后的猎鹰 9 成了全世界最可靠的火箭,每两三天发射一次。

关于什么风险不该冒,他的原则是:不要主动选择"赌上整个公司",除非被逼到没有别的路。

Tesla 历史上几次大赌都是被逼的。从初代 Roadster 年产六百多台,到 Model S 年产两万台,再到 Model 3 年产五十万台、Model Y 超过一百万台。每上一个新车型,新车型的产量就会占到公司总产量的九成,客观上就是拿整个公司在下注。

Cybertruck 不一样,它不是生死赌注,所以可以疯一把。他当时的想法是:《银翼杀手》里的人会开什么样的卡车?这车在路上找不到任何参照物,其他皮卡看起来都是彼此的模糊翻版。他承认这车可能有人不喜欢,但公司亏得起这次失败。

关于创新和失败的关系,他给了 SpaceX 发动机团队那句名言:如果我们不是偶尔在试车台上炸掉一台发动机,说明我们试得不够努力。

组织的激励机制必须允许失败,因为创新必然走进未知地带,未知地带必然有踩空的时候。不允许炸发动机的公司,最后连发动机都造不出来。

我不管理聪明人

坦根问他怎么管手下一堆天才。

马斯克的回答把"管理"这个词直接拆了:我不管理聪明人,他们管理自己。

逻辑是,真正有才华的人随时可以去任何地方做任何事,他们不需要为谁打工。所以他做的事只有一件:把目标讲清楚,问对方认不认同,认同就一起干。

过程中他会给意见,但极少强行拍板。偶尔会有那种时刻,他说,这次你们得信我,就按这个来,如果最后证明是错的,你们可以一直拿这件事骂我。

坦根提到艾萨克森(Walter Isaacson)写的传记,说书里能看出他对细节的掌握程度深得惊人,问他怎么平衡微观管理和授权。

马斯克不接受"微观管理"这个说法。他管这叫对细节的坚持:想做出一件完美的产品,抠细节是必需品,没有替代方案。

顺便说,这本传记他自己没读过。他问过艾萨克森该不该读,对方建议他别读,他就没读。

一百小时工作周是应急手段,不是生活方式

谈到"硬核"文化,马斯克没有美化。

事情真的紧张起来的时候,基本上是醒着的每一个小时都在工作。这种状态他连续维持过好几年。感受是:很痛苦。

所谓 100 小时工作周是真的,每天睡 6 小时左右,剩下全在工作。他明确说不推荐,那是应急状态,不是常态。Tesla 最难的几个阶段他这么干过,最早创业的时候也这么干过,直接睡在办公桌底下,一周七天不离开办公室。

有时候这是成功和失败之间唯一的差别。

坦根问他是不是享受危机模式。答案是否定的:一点也不,很难受,不想待在那个状态里。

国际象棋教不了你什么

聊到游戏,马斯克的话变多了。

他的用法是:想从工作里抽离的时候,就玩一局很硬核的游戏。

《文明》给他的启发是科技树。你得先发明文字,才能发明民主;先有火药,才有后来的事。技术和观念是有阶段的,跳不过去。这个框架放到公司里同样成立,你没法在一个还没打好的地基上盖第十层楼。

他觉得国际象棋被高估了。棋盘只有 64 格,每次都是同样的棋子,没有地形差异,没有科技树,没有战争迷雾。能迁移到现实世界的东西很少。

他更推崇一款瑞典开发的游戏《Polytopia》。随机地形、不同地形带来的攻防加成、十六个能力各异的部落、可以自由选择的科技树、战争迷雾,这些东西加起来,比国际象棋接近现实得多。现实世界的决策,从来都是在信息不全、地形不明、对手不明的条件下做出的。

《暗黑破坏神》他也玩,说高等级之后这游戏像一张带游戏的电子表格。他曾经在一个四人队里短暂拿过最难关卡的世界速度纪录,保持了一天。他的自嘲是:对一个快 53 岁的人来说不算太差,但需要反应速度的项目,基本赢不了小孩。

顺带说一句他对学术的判断。坦根提到他好像说过大部分博士没什么用。他修正了一下:是大部分博士论文没什么用。每年产出的论文数量巨大,真正被任何人用到的比例极低,从客观统计上看就是这样。偶尔会有惊艳的东西,但很罕见。

我不在乎别人怎么记住我

最后一个问题,问他希望留下什么遗产。

马斯克的回答可能是整场访谈里最不像企业家的一段:我不在乎后人记得的准不准确,只要我自己觉得,我为意识的未来做了正确的事。

他讲了自己的"好奇心哲学":理解宇宙的本质,甚至理解该问什么问题。他援引道格拉斯·亚当斯《银河系漫游指南》里的设定,超级计算机算了几百万年,给出的终极答案是 42,但没人知道问题是什么。

所以真正难的部分从来不是答案,是问题。

整场对话里,他谈 AI 的时间表、火星的吨位、巴西的官司、中国的竞争对手、管理的方法,逻辑链条各不相同,底下的动机是同一个:让这簇意识烛火多烧一会儿,烧得远一点。

给技术从业者的几条可操作建议

  • 评估 AI 项目时,把电力、变压器、数据这三道关排在模型参数前面。参数可以堆,电力和变压器不能凭空变出来。看得懂基础设施约束的人,比看得懂榜单的人更早判断出谁能跑出来。
  • 把真实世界视频当成战略资产重新盘一遍。公开文本数据正在耗尽,合成数据有自我污染的风险,谁手里有独家的真实场景视频,谁在下一轮训练里有定价权。
  • 把"真实性"写进模型训练的第一原则,而不是事后加一层过滤。让模型学会在不知道的时候说不知道,比让它学会说漂亮话重要得多。权力越大,这个差别越致命。
  • 在团队里建立"偶尔炸发动机"的容错机制。如果试车台上从来没出过事,说明测试强度不够。要区分"探索性失败"和"重复性失误",前者该奖励,后者该复盘。
  • 决策时先确认方向,再谈速度。向量不是标量。方向错了,执行力越强亏得越快。同时接受一个事实:没人能一次选对方向,所以要预设修正机制,像制导导弹一样持续校准。
  • 把 100 小时工作周定位成应急手段,别当企业文化来宣传。它有效,但它不可持续,长期维持会摧毁判断力。真正该复制的是"目标清晰加高密度反馈",不是工时数字。
  • 玩游戏挑有科技树和战争迷雾的。国际象棋练的是固定规则下的计算,现实是规则不明、信息不全、地形未知。《文明》和《Polytopia》这类游戏提供的思维模型,迁移到真实决策上的效率更高。
  • 把文明当成脆弱的系统来设计业务。苏美尔人发明了文字,后来没人能读懂他们的文字。任何看起来理所当然的秩序,都有消失的可能。给关键系统留冗余,这个习惯值得从工程带到组织里。
YouTube 平台字幕(自动或人工),可能有识别错误,人名与数字请以视频为准。已按语句合并成段落,并去掉 um、uh 等口头填充词。 文稿来源:YouTube(自动/人工字幕)。

0:01hello hi Alon hi hi hi everyone and big thanks for taking the timeon you know we've been trying to get you on the podcast since we started it two years ago so we are super pleased that we that we have you on and indeed on your ex platform how cool yeah it's pretty cool yeah I mean you have like lots of people from all around the world simultaneously do effectively a real-time podcast and it works pretty well very good well we have so much to talk about so love to kick off with with AI now what's your take on where we are in the AI race just now wow that's a long answer there's there's so much happening in AI is the fastest advancing technology

0:49that I've ever seen of any kind and I've seen a lot of Technology you know barely a week goes by without some new announcement so and and if you look at the amount of AI Hardware the computers coming online that are dedicated to AI that is increasing what looks like at least by a factor of 10 every year if not every 6 to nine months so when you combine the hardware

1:20Coming online really order of magnitude increase every you know call it at least every nine months and many many software breakthroughs if if you look at that that curve it looks insane so I think we'll my my my guess is that we we'll have ai that is smaller than any any one human probably to around the end of next year and then

1:54AI the total amount of sort of sentient compute of AI I think will probably EX ceed all humans in 5 years what what is the what is the race about just now is it algorithm is it people is it computing power what what is it about just now is it the supply of chips just what is it yeah last year it was chip constrainted and the hardware deployment you break it down into the three areas of people data and Hardware I'm starting with Hardware last year it was about a chift supply people could not get enough Nvidia shifts particularly this year it's starting to transition to a voltage Transformer Supply so actually getting enough voltage Transformers put in place so my sort

2:46of very Niche joke is Transformers for Transformers because a lot of the AI That's run is called a Transformer so you need Transformers to run Transformers and then next in the if you look out a year or two or certainly 3 years it's just electricity availability so that's those those constraints in the hardware side so many of the smart world's smartest people are doing AI people that would have done physics before in fact or had have done physics for example have moved into AI because it's just the fastest moving field so we're seeing a lot of the best talent a lot of the smartest humans going into Ai and then we see along with that algorithmic

3:34breakthroughs and then then you start hitting the the wall with the the data problem so the you know you can fit all books ever written just the text the the text in compressed form on one hard drive or call one one computer so when you when you're looking at like so called tokens to train on yeah and you you still think

4:05of like all the books ever written in every in in all languages by All Humans sounds like a lot certainly it's far more than any one human could could ever read it actually is a small it's a small number of train training tokens it's just not enough so then you you start having to look at all the videos have I created you know all the podcasts all the everything and and you start even running out of data there well hopefully they hopefully they will include this podcast that definitely will include this podcast what's the biggest challenge you have with with xai well X is still relatively new so it's not you know like the limiting factor right now is

4:54just training our Gro version two model which should be we think better than GPD 4 and that's we're hoping to complete that in May so that's that's training right now so it's just really we're we're just trying to get enough gpus online to train it fast enough to get that done in May which I think probably will happen and then and that's with roughly 20,000

5:26h100s and and doing I think very efficient training then the next step would be for grock 3 which would be I guess G55 or Beyond would you know requires 100,000 Nvidia h100s training coherently so that's you know a half order of magnitude basically more training and then you really start to have running into this data problem where you you have to either create synthetic data or use real world video those the the two sources of kind of like unlimited data synthetic data

6:04and real world video which I should say Tesla has a pretty big advantage in real world video T has by far the most real world video of anyone yeah you've got a huge library that so when do you think so when do you think we'll see proper AGI well it depends on how you define AGI if you define AGI as smarter than the smartest human I think it's probably end of next year like like within two years but but that's that there's still there's still a pretty big leap beyond that to say smarter than the the machine augmented human Collective so like is it smarter than all humans working together who are also using computers to augment their output and that that I

6:51think is probably five years away one one way to look at it is is is to try to assess like roughly what is the ratio of digital to biological compute last question on on AI any new thoughts on regulation and how it should be structured well I I think we probably do need some sort of regulatory authority to look at the safety of AI just as we have regulatory authorities and other Arenas to you know o oversee aircraft

7:25and the safety of aircraft and cars and and other things you know medication so now the rate at which AI is progressing is is fast is faster than probably any regulatory agency can keep up with but but I do have a comment on what I think is very important for achieving safe AI which is that it's very important to train the a to be as truthful as possible and not to

8:01yeah just to be as truthful as possible I think you can get some very dangerous things when you program an AI to be politically correct thing that things that may seem relatively innocuous now but will not be so in in the future if AI has immense power you can take the Google Gemini example where it it refused to publish to produce a picture of George Washington as a white man and and any in fact any historical figure would automatically be made diverse because it's been programmed to insist on diversity which sounds you know perhaps okay at first but not if the AI has so much power that it can actually enforce diversity and decide there's too many of one kind of people

8:51or too many of one sex and kill off just just kill off enough until the the diversity number is is what it's program programmed to believe is correct but don't you think this will be sorted out in the next version no we don't think so where is China where is China now in AI relative to the US the next verion no they'll make it more subtle okay and less obvious but it will still be there okay well we'll see but where where is China where do you where is China now in I relative to

9:25us I I don't know exactly where China is except there are a lot of very smart people in China and they it won't be they won't be far behind the rest of the world or far behind the US I mean the ai ai right now is very concentrated in San Francisco and London and then you know there's

9:54there's you know a lot happening in in China but I I'm I don't have insight into what they're doing except that they I'm confident they will not be far behind what is developed in the west yeah so but but but mark my words the if if if we do not program an AI to be as truthful as possible that that is where it will go arai that is where the danger [Music] lies yeah move moving Tak here moving to to Tesla is is the EV conversion now going slower than you had expected just where is the speed of conversion now relative to your expectations I think it's going quite fast actually especially in Norway absolutely well it's pretty much all there is is your your

10:47Teslas yeah there's a lot of Teslas in Norway it's crazy thanks I'd once again like to thank Norway for the support of electric vehicles so much appreciated time so and I think it's we will the that electric that all vehicles will go fully electric it's only a matter of time that includes aircraft ultimately and boats obviously trains the only thing that is ironically difficult to well you can't really make it electric is Rockets because you need you can't get away from having to expel Mass you sort of Newton's third law but but all cars

11:33will be electric it's only matter of time and we'll look back on combustion cause in the same way that we look at back on steam engines that that I was it was inevitable that there would be internal combustion cars and and it just as inevitable that all cars will go electric and

11:55there will be some e you know so like it going to be a completely straight up line there will be some EV and flow in how how far electric cars go but that but the ultimate victory of electric cars is inevitable and and I think the sooner we get there the better yeah how do you see the Chinese competition here now we generally find that the companies in China are the most competitive in the world and certainly in electric vehicles or cars in general the Chinese car companies are by far the most competitive yeah that's where where we find the most toughest toughest compet competitive challenges that they make great cars and they work very hard so when you ride in

12:42one of the Chinese cars what do you think I mean you're an engineer you know what about it what do you what do you think I haven't R I have not ridden in one lately but cuz they're not all available here you know in the US or few are available in the US some are available in Europe but from what my team tells me

13:06they are very good now you are moving into India here as well in terms of production what are your what are your thoughts here sorry moving into where to India Oh India yeah yeah I think it's you know India is sort of a it's India is now the most populous country in the world you know the biggest population and I think there we India just should have electric cars just like every other country has electric cars and

13:39yeah so it's a natural progression to provide electric Tesla electric vehicles in India yeah moving moving out out in space what what would it take to be self-sufficient Mars to be self-sufficient Mars it's really about the the total tonnage that is delivered to the surface of Mars so you can say like

14:13well I I think it's probably on the order of a million tons maybe it maybe more but somewhere between probably a million tons and 10 million tons are needed to make Mars self-sufficient and how many Rockets is that well I gave a presentation on this recently if look at my my recent SpaceX talk but if you if you have U it's really if you have 100 tons per

14:42flight you need 10,000 flights to get to a million million tons and that's 100 tons landed to the surface of Mars so in order to get 100 tons land to the surface of Mars you need 500 t five times that number in Earth orbit so we do a lot of orbital refilling so launching sort of Rockets

15:08tanker ships over and over again that that would replenish the propellant of the ships that would go to Mars and then you'd need on roughly on order of 10,000 of them to get to a million tons and but we we plan to do that that that's that's we we think we can get that done within 20 years really so and when do you think so when do you think we'll be there for the first time first first well the first Starship that will land on Mars which obviously will not have people at first I think it's probably within about 5 years and then it would probably launch several ships and just confirm that they can land okay on Mars we'll also be doing the Moon

15:59simultaneously with that so go taking I think I think we'll get people back to the Moon I should say within 5 years and we'll get uncrewed ships landed on Ms within 5 years and and then we be building up the production rate and improving the design of the booster in the ship so so in the first people on Mars I think within seven years or so 7 to n years and from from there we need to rapidly increase we need massive numbers of ships going and Earth and Mars Only are in the same quadrant of the solar system roughly for 6 months every two years or at least it's only possible to really transfer efficiently from Earth to Mars I say

16:54every six months but really there's about there's a couple months where where it's ideal every 26 months so every 2 years that you would see a basically a fleet depart Mars I think it be quite a spectacular thing to see a thousand ships depart from Mars all at once like Battle Star Galactica what kind of new technology do we need before we be self-sufficient there actually I think we have all the tech we already know all the technology that's necessary for that it just needs we just need to build so no new physics is needed for this why is it so important for you I think it's important for Consciousness in general so if we wish to maximize the

17:38lifespan of Consciousness then being a multiplet species will result in a much longer existence of Consciousness Consciousness than if we are on one planet if we on one planet we're simply biting our time until there's eventually a Calamity it could be soon it could be a long time but eventually something will happen could be you know Global ther nuclear war it could be simply That civilization merely subsides our civilization may not die with a bang it may die with a wimper just just gradually falling into obsolescence but if we're a multiplet species then we've got two planets and the and they can support each other and we can go beyond two planets ultimately to the

18:24moons of Jupiter to the to the you beyond to the the outer parts of the solar system and ultimately to other star systems so this tiny this tiny candle of Consciousness that we have in this vast Darkness can be extended and Amplified and we're just far more likely to survive as for for Consciousness to survive if we are multiplet species you don't think it'd be better to use all these resources and try to s out Earth well just to put this into perspective the amount of resources I'm talking about for making life multiplanetary would be less than 1% of all resources on Earth so really you can think of it as resource allocation do you think it's

19:13worth spending half a% of Earth Resources to ensure that we have redundancy in Consciousness and that we extend Consciousness Beyond Mars to other planets to to Mars and other planets and ultimately other star systems and then also take into account the fact that there are certain inevit there are certain things we simply cannot avoid on Earth like is it within your power of mind to stop World War III I don't think so no if it happens and if we have Global ther nulear Warfare our technology level will drop to the stern age and we may never survive and then there are we maybe get we may get hit like by a comet like the dinosaurs and

19:58and you know if the dinosaurs had spaceships they they'll probably still be around so and then if if you wait long enough the Earth that the sun will continue to expand and eventually engulf Earth and destroy it and destroy all life so just to give it amount a certain amount of time no matter what you do on Earth no matter how careful you are Earth will life all life on Earth will die that it will happen is a certainty it's a big glom of this no so on a slightly less gloomy note X

20:36Twitter yeah what is your vision now what do you how do you see the the vision of X our goal of X is to be the best source of Truth on the internet and I think we're making a good you know good progress there and I mean this it's going to be like I call the everything app like if anything anything you want to do you can do on the xplatform whether it's text audio

21:04video payments Financial stuff Communications of all kinds and then but but then also where there is publicly disseminated information just to be the best source of Truth and I think it I think it already is that now people may say oh there's some piece of misinformation to information I say yes but look look at the replies the replies correct that misinformation and look at Community notes and the and how good the batting average of community notes is it's extremely good it's by far the best factchecking system on the internet so and and a lot of people still labor under the illusion that the the the Legacy newspapers that they read are actually true there's so much

21:44nonsense in them I mean Nico how many times when do you read an article in newspaper where you know the circumstances of what that article is and how often is it spot on no of course it's normally of course we all know it's normally wrong but but how do you look not sure but how do you look at the situation now for instance with with Russia you know the work Russia does in Germany with fake accounts on it's pretty pretty huge activity

22:14right I mean we don't see a lot of Russian activity to be frank on the system so we we see very little we do we do see a lot of lot of attempts to influence things but they seem to be coming from from the West not from from Russia right what about what about things like the latest developments in in Brazil and so on yeah sure yeah so the

22:46the we we kept getting these demands from this Judge Alexander that's his that's his name on Twitter Alexander and there would be to suspend accounts immediately we given typically two hours to suspend an account or face massive fines and the the final stoies we were we being given demands to suspend setting setting members of the parliament and major journalists and moreover we could not tell them that it this was at the behest of as under Morales we had to pretend that it was due to our rules of service and that was the final straw and we said

23:41no now when you when you bought Twitter now renamed X did you expect that you would end up in these type of situations so it's is all unexpected well I knew it wouldn't be just a total better Ros you know and it's talking I [Laughter] wouldn't no I mean I thought it would be since we're just like rigorously trying to pursue the the the goal of

24:17being the most accurate and truthful place in the internet and that that doesn't mean that what is said is always true or accurate but is it is perhaps another way to frame it is as the least inaccurate place on the internet do you do you secretly do you secretly think this is a bit fun it's fun yeah yeah it's fun at times it's stressful at times and it's fun at times but overall we're trying to serve the people of Earth and and and this is sort of an EST sort of maybe an estar way of viewing it but to try to be kind of like the the group consciousness of Earth so you can think of like if each person is like a neuron contributing to like the collective

24:57brain of Earth and you want to try to minimize the noise and maximize the signal of every neuron that's connected to the the X Network that that's basically what what is what is the collective will of of humanity and and how and and and how to yeah just serve the collective will of humanity and serve the greater good that that's our goal now you know there there's definitely going to be people who want to manipulate that information and so we have to fight that and try to have you know be it be the most accurate place as to the best of our ability and have it be kind of a marketpl of ideas where people can propose ideas and you know debate them

25:42and I think so far it's working reasonably well in that regard now people that don't like the truth will not like ex or they want to manipulate things they will not like it but only but only a few years ago you were you were a guy producing electric vehicles now you are you know through starlink you've had some you know I mean some big impact in in Ukraine with Twitter you are kind of into some issues in you know Brazil India Turkey you know you're becoming like a real geopolitical force and a really important one how do you how do you look at that well like I said I'm really just I'm trying to take the set of actions that maximize the probability the future is good I

26:39mean we have to keep civilization going onward and upward as much as possible and and try to minimize the civilizational threats that occur like you know we we we can't get to Mars if civilization collapses it's not going to happen so you know we've got to we've got to keep keep civilization going and I

27:07think we should view our civilization as being much more fragile than we think we kind of take for granted oh it's always going to be there but actually if you study history you realize that there Rise you know there's rise and full civilizations I mean I was I was reading in depth about the ancient samarians who were arguably the first civilization if you call civilization like writing and stuff you know they were the first to develop writing and

27:37but eventually they died out and they were gone so and then nobody could read the writing at all and and they they just faded out as a civilization but they're pretty impressive in their time the ancient Egyptians the same thing and you know one sort of one after another ancient Greek had it Greece had its day you know China and India had will have incredibly impressive populations but there's been EVs and flows in the CH China and Indian civilizations over the the the aons you know the

28:16benen as well so you know I I guess I'm just trying to take this this set of the steps that increase the scope and scale of Consciousness that's that's what I'm trying to do it's not it's not that I'm trying to have a put a political thumb on the scale or anything like that but I I think I'm trying to have the political will go where the people want it to go you know you you mentioned some

28:48some really smart people here and kind of just moving t a bit here to Cora culture now you manage a lot of geniuses in your in your companies what is the key to manage really smart people you think I don't I don't think I manage smart people they manage themselves I I think well I guess with really smart people you know I don't really think of of it like managing them I think that if somebody's very smart and talented they they can go anywhere and do anything anytime like if they they they don't have to work with me they could go anywhere so I I really just say like look this is the the goal we're after and this is what we're trying to achieve and do you

29:34agree with this goal and if you do then let's try to get it done and you know provide my opinion along the way and but I it's very rare for me to actually sort of insist on a on on on a particular thing once in a while I'll say look guys you just got to trust me on on this one we got to do this thing and if it turns out be a bad decision you we can can all hold that against me in the future but you have an incredible eye for detail right I mean when we read the is book it's pretty clear that you I mean you really are are deep into detail and know what you talk about so how do you how do you balance this kind of micromanagement of some areas and then delegate other

30:21areas I wouldn't I wouldn't call it micromanagement it's just insisting on attention to detail that if you're trying to make a perfect product you must have attention to attent attention to details essential and I haven't actually read the isson book you should it's very good actually I Lov it well I I asked Walter isacon if I should read it and he said I shouldn't so so then okay so said I shouldn't

30:58read it so okay well I'll I'll ask you some questions from the book then they you he talks about you know you the kind of the hardcore and Ultra Hardcore culture what is an Ultra Hardcore culture I guess it's work I mean it's working culture right I mean how how I mean Ultra hard work how hard is

31:19that well when things get really intense you're basically just working every waking hour and how and how long can you do that for I've done that for well continuously for sometimes like a few years what does it what does it do to you it really it's pain and and every waking hour maybe is an exaggeration because there are a few hours obviously with friends and family and and critical other things but 100 hour weeks would be I I've done many many stretches of hour weeks like true 100 hour weeks where roughly 6 hours per day is sleeping I would not recommend that this is not that's for emergencies you know it's not all the time you know during very difficult

32:15times at Tesla I've had to do that and sometimes at the beginning of my earliest startups I did that where I just wouldn't leave the office I would just sleep under my desk and just work seven days a week sometimes it's necessary for success or or to avoid failure but but do you you do you enjoy being in this crisis mode no I don't it

32:45sucks okay no I I don't want to be there it's pain but sometimes it's the difference between success and failure when you make decisions how important is speed he just gave me an idea which is I'm going to invite the Judge Al Alexander mores to do a spaces and then he can explain why what I'm doing is bad and and and maybe he's right I challenge I challenge him to a

33:23spacers sounds good yeah but what about when you make this when you when you make decisions how how important is beat and how do you how do you balance analysis with your gutfield I think the the the best offense and defense is speed if you think of something like the SR71 Blackbird it really had almost no defenses except accelerate and it was never shut down even once like I think over 3,000 missiles were shot at the SR71 blackboat and none hit and and really what it did was just go faster so the the power of speed

34:05is underappreciated as a competitive Dimension is that why you know space expenses has been so successful because you've been mean and lean as an organization and fast think speed speed is definitely a factor now I should say you want to go in the case of a company you you you you need to be a vector not a scaler so it can't be you need to go at high speed in the right

34:36direction sure so I can't just so and no company's goingon to be going in the right direction all the time so you have to do course Corrections like a guided missile you call it course Corrections and but in the case of SpaceX it's like okay our goal is to extend Humanity beond Earth and we didn't even know how to even frame the question correctly like what what which knew that that was the General goal we didn't know what pent

35:11we' use or what the raw materials would be or for the how would the rocket be built how would it be designed what's actually important and you know so for example going from our Falcon architecture which is uses refine jet fuel and liquid oxygen in a open cycle gas generator architecture engine to a to Starship which is liquid methane liquid oxygen propellent and a staged combustion very high pressure engine that that that's that's a big architectural change but we didn't know that we we needed to make that architectural change until we're pretty far down the road like about half took us about 10 years to figure out that was even the right architecture now

36:01we're confident it is I think SpaceX is one of the best example I know about what we call failing well right learning from mistakes and moving on what generally how do you how do you look at mistakes I mean which which ones do you tolerate and which ones don't you tolerate well I I I think I don't really think of that way you know the first three flights of SpaceX failed the fourth one succeeded and if if the fourth one had not succeeded we would have gone bankrupt we would have had no money left so it was a very close call but since then space has done very well it's now the the Falcon 9 you know knock on wood is the most reliable rocket in the world and launches

36:45about every two to three days last question on risk what are the types of risk you would not want to take well I I I think in in terms of risks you don't you you don't want to take risks that way if if you only want to take bet the company risks if they're absolutely necessary so they've been a few times where saying the with Tesla we we just had no choice but to V the company because if we if we're Inc if we're doing a new vehicle program that is an order of magnitude larger than the past one then we're by we're just unequivocally betting the company because the new vehicle would be 90% of production so going from the original Roadster to the Model S original rast to

37:29was only you know about 600 6 700 per year then Model S was 20,000 per year and and then model 3 is sort of half sort of half a million per year model y over a million per year so these are bet the company vehicles but the the the reason we could do for example the sabit truck which was kind of a a radical new design was because it wasn't a bet the company decision so I was like okay look let's try something I want to try something totally crazy it's like what what truck would Blade Runner Drive except when you're going to drive on Ms yeah I think it would be perfect for Ms but like we could try something that where there's some chance that people might not like it but

38:19it's it's radical and new and it's aesthetic aesthetically it's not derivative it doesn't look like anything else on the road whereas all the other sort of pickup trucks look like vague copies of one another they we could afford to take a chance on failure and say like and talk it up to you know well we tried you know we tried to do something interesting but but actually by the way cyber truck's doing great

38:45so but one of the things that I think is important for Innovation is that you do accept failure like like necessarily you have to always look at the incentive structure of an organization and say you know is is is that is that organization properly incenting Innovation and in with if you do Innovation you're necessarily going to Uncharted Territory so they're going to be some mistakes they're going to be some failures and you have you have to like like actually like for for SpaceX rocket engine development like I keep telling the team look if we're not occasionally blowing up an engine on the test stand we're not trying hard enough you know absolutely absolutely how

39:31important are the P how important is research and phds and that kind of stuff I think I've said seen somewhere you you think most phds are useless well I think most PhD thesis are useless which I think is actually objectively true if you look at how many PhD you look at all how many phds are created every year and how many of those papers are actually used in anything yeah then objectively most PhD PCS are

40:02have very low utility or maybe zero because they nobody uses them or so once in a while you get something that is spectacular but it's pretty rare perhaps something more useful is in the book that you haven't read talks about your love for gaming in particular like strategic ability gaming and I've been thinking quite a lot about it what have you learned from from those games and and have have that learning and wisdom been helpful when you have been planning your companies yeah I it's hard to say exactly what I've learned from video games said that I I do like playing video games as if I want to take my mind off work I'll typically play a very hard video game such as such

40:49as which one well over the years it's been many many different video games so you know when I was a little kid who was like you know pong and little tank games and things and and but if you take a game like for example civilization it's actually quite a good it tells you how how civilizations are formed like I remember I remember playing the original civilization with the technology tree and and how you invent different things you'd like invent literacy and you know invent

41:22democracy and invent gun gunpowder all all these things like and you start to realize oh wow there there there are stages to technology like you can't you know you can't actually get to democracy without literacy and you know so there's these these stages of of Technology development or stages of ideas that you know that's that's a helpful framework for a company and I guess in in like like I say in recent years there's there's a game I play that was actually developed in Sweden called polyopia which is actually quite a good game like a lot of people like playing chess but I think chess is not not a great there's not a lot of transfer learning from chess to the real

42:12world because in chess you've got only 64 squares it's a setpiece battle same pieces every time there are no terrain differences there's no technology tree there's no fog of War but say a game like polyopia has all of those things random Trin generation you know the differences in attack and defense bonuses depending on what type of terrain you've got 16 tribes I think each with different abilities you've got a a technology tree that you can choose to develop in different ways and you've got of course fog of War so that I think is much more much closer to reality yeah yeah so I think politopia I mean I I

42:58was I was playing Diablo for a while pretty fun Diablo as high levels gets very complicated they you could call it like a a spreadsheet with a game attached so so that's that's and I briefly got the the for about a day the world record in this Avatar of Zer on on a four person team of of clearing the the hardest level which was you know not bad for someone who's like 50 three basically we'll be 53 soon there is still some twitch element to it and it's hard to beat kids at games with a twitch element but yeah I like I find these games interesting if you can be fully immersed in a game it's great now what is the score now of in terms of the Union in Sweden and the

43:52collective bargaining actually I I think I think the storm has passed on that front I think things are in reasonably good shape in Sweden so yeah so I think things are good yeah overall yeah I feel pretty good about the future I mean you know there's going to be bumpy quarters from you know here and there but I think the long-term future of Tesla is extremely strong just so yeah we met with we met with your chair last month so we we have some update but any any view on it why are you why why are you skeptical to

44:38collect I playing with a soundboard here so last question here what do you want your legacy to be I I don't I don't mind if my legacy is accurate or inaccurate provided that I I dive feeling that I've done the right thing for the future of Consciousness so just trying to trying to have this SL of Consciousness last as long as possible and maybe understand more about the nature of the universe or simulation or whatever is so I have a philos philosophy of curiosity which is to understand the understand the universe understand the nature of the universe or even what questions to ask kind of like that I I would say I would subscribe to the Douglas Adams

45:50hitas guys of the Galaxy School of philosophy that we're trying to understand what questions to ask about the answer that is the universe okay I think that's a good place to end for sure the life and life on on this Planet would have been a lot more boring without you and so I'm I'm glad to Spiced it up a little

46:14hopefully all right I guess I'll say bye then bye take care now all right bye thanks bye

资源与出处

相关视频

同频道、同主题或相邻年份