全球 AI 算力的年产出,今天大约是 20 吉瓦。马斯克在奥斯汀说,他要做到 1 太瓦。
1 太瓦等于 1000 吉瓦。50 倍。他没说"争取翻一番",也没说"我们要建一座和台积电一样大的厂"。他说的是把整个地球的芯片产能乘以五十,而且是在特斯拉、xAI、SpaceX 三家公司的联合目标下做这件事。
彼得·迪亚曼迪斯(Peter Diamandis)在最新一期 Moonshots 里把这件事排在所有新闻之前。他的搭档、物理学家亚历克斯·威斯纳-格罗斯(Alex Wissner-Gross)给出的评价是:马斯克在瓶颈上做指数级增长,他不跟你竞争,他直接改写游戏规则,然后问在场所有人敢不敢跟。
这期播客的信息量很大。除了太拉工厂(Terafab),四个人还聊了人类驾驶什么时候会变成违法行为、标普 500 的估值体系会不会被 AI 拆掉、一家公司该不该统计员工的 token 消耗、月球上的土豆、小行星上的五种碱基,以及一个没人认领的一万亿参数模型。下面按主题拆开讲。
他要的不是更多芯片,是五十倍的地球产能
先把数字摆清楚。全球现在每年产出的 AI 算力,用功率衡量大约是 20 吉瓦。这个单位的变化本身值得注意:过去大家数芯片颗数,现在开始数功率,因为算力的瓶颈已经从设计转到了电力和散热。
马斯克的目标是每年 1 太瓦,两种芯片:一种是给机器人和汽车用的边缘推理芯片,另一种是给太空设施用的高功率抗辐射芯片。工厂落在奥斯汀,最终规模可能达到一亿平方英尺。短期目标是"个位数年份"内做到 1 太瓦,长期目标是 1 拍瓦,而那个量级只能靠月球质量驱动器送上去。
他为什么自己建?亚历克斯转述了马斯克的说法:他挨家挨户去找所有芯片厂,说"你们能给我多少产能,我就付多少钱,我不想跟你们竞争,我只要更多"。没有一家按他的速度走。于是他决定自己干。
这套剧本已经演过两遍。火箭发射,他把整个行业甩在后面;电动车和自动驾驶,同样。萨利姆·伊斯梅尔(Salim Ismail)说,这是他的固定打法,一次又一次重复。
三星那边还有一个细节。马斯克和三星签了 160 亿美元的代工协议,如果按计划推进,实际规模可能到 450 亿美元。他公开说,三星能给他多少他就买多少,但那还不够,所以他还是会自己建厂。戴夫·布伦丁(Dave Blundin)认为,这套说辞某种程度上是为了不激怒三星。毕竟这些芯片是给车和太空用的,是加固过的,"跟普通芯片不一样",听起来像个体面的说法。
奥斯汀工厂的关键设计是全垂直整合在一个屋檐下。芯片设计、流片、测试、迭代,全在同一栋楼里完成。亚历克斯说这是他最欣赏的一点:快速迭代的能力,比产能本身更值钱。
每 5.3 分钟一次星舰发射,才够喂饱这座工厂
彼得算了一组数字,值得单独拿出来看。
马斯克的目标是每年向轨道运送 1000 万吨质量。下一代星链卫星平均一颗一吨。要运完这些质量,需要每天 274 次星舰发射,也就是每 5.3 分钟一次。
马斯克自己的回应是:在航空业,这个频率很正常。听起来荒诞,但把参照系从航天换成民航,确实就是这个量级。亚历克斯的判断是,这个发射节奏会强制整个运营模型发生根本改变,包括所有工程师和基础设施的思维方式。
再往外推一层。如果长期目标是从月球采矿造 1 拍瓦算力,亚历克斯算了一下:这大约相当于月球质量的十万分之三。听起来无害,但把尺度推到 1 艾瓦(exawatt),就是月球质量的 3%。
"人们以为我在开玩笑,说要把月球拆了,或者月球活该被拆。但这条曲线确实画出了这样一幅图景:月球的确在劫难逃,它被安排去建戴森云了。"(亚历克斯·威斯纳-格罗斯)
这个说法当然带着玩笑成分,但它背后的逻辑不玩笑。算力的物理上限,最终是星球质量的利用率。
钱从哪来:一份被低估了一个数量级的账单
启动太拉工厂,第一阶段需要 250 亿美元,主要是把楼盖起来、把设备装进去。但 250 亿只够一个厂。要做到全球产能的 50 倍,理论上需要 50 个这样的投资。
戴夫让 Claude 做了一份估值模型,结论是前期资本开支在 1500 亿美元量级,最少;上限可能到 5000 亿。模型还算了一笔账:这些芯片自产自用,装进 Optimus 机器人,再装进 Cybercab,加上对外销售和应用的企业价值,总计是万亿美元到数万亿美元。
亚历克斯直接说这个数字偏低。台积电的市值是 1.7 万亿美元,英伟达是 4.5 万亿,而太拉工厂的规划产能是台积电的 70%,同时还要做英伟达做的事。"简单算法是台积电加英伟达,再乘以 50,作为估值的起点。所以这份模型差了不止一个数量级。"
市场已经在给它定价了。SpaceX 的 IPO 预测市场,最初锚定 1.5 万亿美元,现在已经偏向 2 万亿以上。也就是说,太拉工厂里属于 SpaceX 的那部分(按规划是 80%,剩下 20% 归特斯拉),已经在被提前计入估值。
这就引出一个治理结构问题。亚历克斯承认,他也不清楚太拉工厂的公司架构最终长什么样。但如果它主要落在 SpaceX 这个口袋里,那么这个 IPO 就同时装着 SpaceX、星链、英伟达、ASML 和台积电的影子。
萨利姆提了一个比较犬儒的角度:计划如此宏大,时机恰好卡在 IPO 之前,让所有人都兴奋起来。但他随即表示,即便如此,他还是喜欢这种胆量。
关于垄断的担忧,四个人一致不买账。亚历克斯的理由是:等你要拆月球建戴森云的时候,地球上不可能只有一朵戴森云。谷歌不会消失,它有 3000 亿美元收入和 1000 亿以上的现金流,有自己的芯片设计,除了火箭什么都有。而火箭这块,埃里克·施密特投资的 Relativity Space 也算半个谷歌系。
人类开车这件事,正在被 92% 这个数字终结
第二个话题是交通。彼得给了一组数据:Waymo 累计完成 1.7 亿英里全自动驾驶,相当于 200 个人类寿命的驾驶量,严重事故率比人类低 92%。目前有 3000 辆车,覆盖 10 个城市。
Uber 投了 12.5 亿美元给 Rivian,计划部署 5 万辆全自动 robotaxi。Joby Aviation 的 eVTOL 飞过了金门大桥,Archer Aviation 的 Midnight 成为第一款获得 FAA 100% 合规认可的 eVTOL。彼得判断,飞行车 18 个月内在美国开始运营,2028 年在洛杉矶会形成规模。
然后他抛出了问题:什么时候人类开车会变成违法行为?
萨利姆的回答是从市中心开始,然后慢慢向外扩。戴夫的回答更具体:这会像室内禁烟和酒驾入刑一样,需要一个引爆点。他预计三到五年后会出现一轮电视广告轰炸,画面很难看,然后大量选民开始投票,最后社会共识变成"在公共道路上开车是不人道的,去赛道开"。他还顺带提了一个数据:五岁以下儿童的第一大死因是交通事故。
彼得补充了真正的时间约束:
"能让这件事推迟的,只有芯片短缺。技术会在那里,需求也会在那里,但芯片会晚很久才到。所以如果你是工程师,想解锁这件事,去研究怎么在这个特定场景下用更少的硅做更多的计算,你会立刻变成亿万富翁。"(彼得·迪亚曼迪斯)
亚历克斯给了完全不同的答案:他认为永远不会违法,只是"驾驶"这个词会被重新定义。现在用 FSD 14,你告诉车去哪里,中途还能跟 Grok 聊天、做微调转向。未来这种关系会继续抽象化,人在环上,机器保证安全,人依然觉得自己在开车。
彼得接过这个思路,描述了一个更具体的版本:你可以把车开得又快又狠,因为车知道轮胎的抓地力,知道路面的状况,它会在你即将做出毁灭性动作的那 1% 时刻接管。剩下 99% 的时间,你说了算。
亚历克斯说得更直接:未来的"油门踏板"会变成 FSD 里的"疯狂麦克斯"模式,加速这件事本身被抽象掉了。
萨利姆顺势改了一句广告语:以前是"朋友不让朋友酒驾",以后可以改成"朋友不让朋友驾驶",把"酒"字去掉就行。
停车场变成公园,车库变成卧室
Cybercab 能不能和 Waymo 竞争?戴夫的回答是当然能,而且竞争会非常激烈,这对消费者是好事,因为十几家公司在一个城市拼服务质量和价格,最终会把价格压到接近成本。
Cybercab 的定价大约 3 万美元,马斯克打算让个人购买。彼得说自己想买 25 到 50 台放在圣莫尼卡,让它们出去跑活赚钱。"我从来没想过你下一份工作是出租车司机,"亚历克斯说,"但奇点会让奇怪的人凑到一起。"彼得纠正:是车队主。
萨利姆指出一个更大的结构性变化:今天全球每年生产近 1 亿辆新车,而这些车 94% 的时间是空着的。自动驾驶会把汽车保有量砍掉一大截,同时车的寿命接近无限。他那台 2017 年的 Model 3 能跑 100 万英里,车本身没有任何问题。整个汽车制造、汽车服务、汽车维修行业,会从底层被重新洗牌。
价格上,从旧金山机场到市区现在要 200 美元左右,绝大部分是司机成本。把司机拿掉,每英里成本会降到 10 到 30 美分,比拥有一辆车便宜四到五倍。
然后话题转向房地产,这是彼得认为对听众最有用的一段。
如果路上的车减少 10 倍,而且这些车不需要停车,城市的空间结构会彻底改变。彼得描述了自己的日常:从早餐桌起身走向门口,AI 知道他要出门、知道他去哪,已经叫好了车,他一句话都没说。他家里原来是三车位的车库,一个改成了卧室,剩下两个基本是储藏间,接下来打算改成健身房。"个人车库这个概念会消失。现在就开始想,你打算把车库空间拿来干什么。"
洛杉矶 60% 的土地面积是停车场。这个数字让四个人都愣了一下。这些地会变成花园、绿地和公园。体育场周围成片成片的停车区也要重新规划,连尾车野餐这种文化都得重写。
亚历克斯补充了两个二阶效应。第一个比较老套:郊区会重新扩张,因为通勤时间被压缩后,市中心的房地产溢价会下降。第二个不那么老套:如果把趋势推到极致,为什么不做自动驾驶房车,甚至自动驾驶办公楼?车与车之间组成社交网络,你的两辆车在需要开会时自动对接同步位置。早上在波士顿,晚上在华盛顿,车是卧铺,第二天到芝加哥。人变成了被自动驾驶系统路由的互联网数据包。
萨利姆给年轻人的建议很直接:如果你年轻、有工作、住在城市里,现在别急着买房,继续租。把钱投到一个风景好、但目前不太好到达的地方,作为第二居所。等到 2034 到 2036 年,那地方会变得极其抢手,因为那时财富总量是现在的十倍,而真正好的位置是有限的。他还补了一句:无人机送咖啡会解决最后一公里的问题。
每家公司都在偷偷统计员工的 token 消耗
第三个话题是就业洗牌,标题叫"大重组"。彼得列了几条本周新闻。
高盛的报告说,AI 可能自动化美国 25% 的工作小时数。彼得觉得这个估计偏低。普华永道告诉自己的合伙人:抗拒 AI 的人在这里没有位置,要么自己用 AI 武装起来,要么走人。G42 发布了一个只招 AI 智能体的招聘启事。还有一条来自黄仁勋:公司现在开始追踪单个员工的 AI token 使用量,他说"如果一个年薪 50 万美元的工程师没消耗掉 25 万美元的 token,我会深感不安"。
彼得把这句类比成戴比尔斯那句"三个月工资买一枚钻戒"。戴夫笑着接:这是 token 版的永恒。
萨利姆正在写一篇叫"组织奇点"的论文,他的结论是:一家典型公司可以用现在 20% 到 25% 的员工运转,因为所有工作流都从人对人变成智能体对智能体。但他不认为这会导致 80% 的失业,因为同时会诞生四到五倍多的新公司,而且大公司的转型速度远慢于初创公司,经济有时间消化。
他甚至提出一个更激进的观点:未来的岗位应该几乎没有"模式识别"的成分,这类工作应该接近于零。至于咨询公司的合伙人,他的判断是:如果 AI 智能体能帮你定战略,你为什么还需要咨询公司?未来的咨询只剩实施环节,而且如果对方有比你更好的智能体,你连实施都不需要他们。
亚历克斯把普华永道那句话解读成一次自我颠覆的尝试,但组织自我颠覆极难。按熊彼特的逻辑,替代按小时计费模式的公司,大概率来自外部,内部自我革命往往做不到这件事。
G42 那条招聘,亚历克斯认真看了。确实是招 AI 智能体,而且招聘方还要求提供开发细节,确认提交申请的真的是一个智能体。"我们离一个人类需要反向证明自己不是歧视对象的未来,已经非常近了。"
戴夫给出的实操目标更狠:他们已经在旗下所有公司推行 token 与薪资的比例考核,目标是 80% token 成本、20% 人力成本。他的逻辑是,等到那个比例,没有雇主会在意最后那 20% 的工资支出,因为在那个阶段,一个能把 AI 效率提升 1% 的员工,价值远高于省下的工资。
关于 token 是不是正确的衡量单位,亚历克斯给了一个很漂亮的回答:token 不一定是正确的生产力单位,但它是第一个可测量、可审计、可分析的生产力单位。工时也能测,打卡机也存在了一个世纪,但那只是粗暴的输入计量。token 不一样,你可以用另一批 token 去审查这批 token 用得值不值。
戴夫给 CEO 的建议很具体:现在就抓 prompt 历史。用 Amazon Bedrock 的话,历史会自动进 S3 桶,不用自己搭任何东西。然后把这些数据喂给另一个 AI,让它评估你八个直接下属的 prompt 质量,告诉你该裁掉哪 20%、该培训哪 20%。他还加了一条底线:不要报销你看不见的 AI 工具,确保它跑在你的基础设施上。
萨利姆补了私募股权这一层。一家现金流很好但商业模式注定被颠覆的公司,在 AI 改造后短期利润率会飙升,因为没有劳动法、没有保险、没有福利。所以 PE 会冲进来收购,AI 化之后把现金流做得更漂亮,再用这笔钱去孵化新业务或者收购初创公司。Anthropic 和 OpenAI 都在和 PE 合作做这件事。
22 倍自由现金流的神话要破了
第四个话题是 Chamath Palihapitiya 那篇关于"终值崩塌"的文章。核心论点是:现代资本市场的整个架构,建立在一个很少被审视的假设上,即竞争优势会随时间累积,护城河会持久,品牌会延续,网络效应会防御。把这个假设抽掉,被重估的不只是几只股票,而是一个世纪以来资本配置的哲学基础。
具体数字:标普 500 现在平均是 22 倍自由现金流。如果 AI 让护城河变得临时,估值压缩到 7 倍,58 万亿美元的市值会蒸发三分之二。压到 2 倍,蒸发 90%。
戴夫同意分析框架,不同意结论。他的论证很扎实:22 倍自由现金流隐含的意思是,这家公司能按同样的赚钱方式再活 22 年。苹果 22 年前卖的是今天这些东西吗?当然不是。那种"建一家石油公司,五十年卖同样的油,让曾孙和洛克菲勒一样富"的模式已经死了,而且该死。但与此同时,未来十年会有远超我们这代人所见的财富涌入经济。所以评估的重心会从"22 年卖同样的产品",转移到"管理团队能不能像马斯克那样滚动创新"。
亚历克斯的反驳角度不同:自由现金流没有消失,只是流到别的地方去了。可能不再流向 SaaS 初创公司,而是流向基础设施、流向月球采矿。资本本身没有被压缩,反而在爆炸式扩张,因为能力和基础设施的基数在指数增长。他认为 Chamath 的论点应该被狭义理解,只适用于被颠覆的那部分行业。往大了说,这篇文章更像是在哀叹"丰裕"带来的财务后果。
萨利姆补了两组数据。第一,他们做过一个 CEO 指数,用灵活度、适应性和组织目的性给财富 100 强打分,结果前 10 名在七年里的股东回报是后 10 名的 40 倍。这个分化早就在发生。第二,标普之外的中间市场公司,估值已经掉到 7 倍自由现金流左右。被撑住的只有标普,而撑住它的是被动指数基金和盲目定投 401k 的人。
所以真正被重写的规则是:如果自由现金流的可见度跌破五年,整个公开市场的逻辑都要重来。奖励对象会从规模和稳定性,转向可再生性和期权价值。实物资产会变得更重要,因为原子比比特更难颠覆。萨利姆的判断是,唯一剩下的护城河,是一个比竞争对手学得更快的活系统。
彼得由此推出一个投资视角的反转:那些估值已经跌到谷底的中间市场公司,如果人力成本占比很高,AI 改造会让自由现金流翻三倍,同时估值倍数还在下降。这里面有真正的便宜货。
亚历克斯最后发起了一个"创业者征集":他希望看到有人做 AI 维权投资者,让 AI 给上市公司写公开信,指出它们做错了什么。这既是给 CEO 的推力,也是一种让 AI 悄悄接管整个经济管理层的路径。
月球上的土豆,小行星上的碱基
第五个话题是新太空竞赛。Artemis 2 计划 4 月 1 日发射,不登月,做一次阿波罗 8 式的绕月飞行。新任 NASA 局长把 SpaceX 拉进了 Artemis 计划,星舰会承担后续的载人登月。与此同时,中国确认了 2030 年前登月的目标。历史在重演,1961 年的剧本换了个主角。
彼得展示了 SLS 和星舰的对比图。星舰的入轨质量是 SLS 的两倍多,地月转移轨道注入质量也是两倍多。真正的差距在成本:SLS 的问题和当年的航天飞机一样,成本大头在飞行器之外,在于那支两万人的常备队伍。一年飞四次,每次 10 亿美元;一年飞一次,每次 40 亿美元。
亚历克斯开玩笑说,想知道联合发射联盟(ULA,SLS 的主承包商)多久会被贝索斯或者别人收购。买它当然不是为了资产,是为了合同和 expertise。
然后是两个科学新闻。第一,科学家成功冷冻了一整颗猪脑,细胞层面的活动被锁定,损伤极小。猪的器官尺寸和人类接近,这个进展有分量。亚历克斯透露自己以非正式顾问身份参与了其中一家公司 Nectome,这家公司做的是化学保存,和 21 世纪医学那套玻璃化保存不同。他一直在推动他们把结果发出来。他顺手发起了一个号召:为什么现在还没有十亿人注册人体冷冻?赶紧去注册,这是你能看到 23 世纪的唯一方式。
第二,日本的隼鸟二号在小行星龙宫(Ryugu)的样本里,找到了 DNA 和 RNA 全部五种碱基:腺嘌呤、鸟嘌呤、胞嘧啶、胸腺嘧啶、尿嘧啶。这为泛种论提供了新证据,生命的基础组件可能来自地球之外,随陨石落到地球上,提供了最初的原料。
彼得提到 NASA 局长 Peter Garrett 的一个预测:NASA 有 90% 以上的概率会在火星上发现某种形式的微生物生命。官方立场出现了重大转变。以前的说法是"我们找到了水,希望找到生命的迹象",现在直接说九成概率。
亚历克斯把这个推向更大的尺度。火星比地球早冷却大约十亿年,这个时间差在星系尺度上足够完成很多轮播撒。星际介质里到处都是生命的构建块,星系在十亿年尺度上相对稳定,恒星之间不断有近距离交会,星团物质反复混合。我们自己就是好几代恒星爆炸重组之后的产物。所以泛种论可能不只是太阳系内的现象,而是银河系尺度的。
他还引用了一项研究:用地球上生命的基因复杂度做参数化,往回外推,第一个碱基对出现的时间大约比已知最早的生命证据早十亿年。这是另一个独立估算,指向同一个时间窗口。
萨利姆提了德雷克方程,说他算下来概率接近 100%。他更期待的是找到非碳基生命,那才是真正的惊喜。
彼得给了一个很有画面感的收尾:恐龙的灭绝靠陨石,DNA 的传播也靠陨石。早期宇宙里生命可能到处冒出来,又不断被另一颗陨石抹掉。只有当一切冷却、沉降下来,才有足够长的时间演化出智慧。它更像一道巨大的系统动力学沉降题。
一万亿参数的模型,没人认领
第六个话题是模型战争。两条新闻。第一条,OpenAI 发布 GPT-5.4 mini 和 nano,速度是两倍,编码基准接近完整版 5.4。第二条更有意思:一个叫 Hunter Alpha 的一万亿参数模型出现在 OpenRouter 上,没有任何署名,100 万 token 上下文窗口,免费,没有开发者公告,没有新闻稿,没有背景故事。它处理了超过 1600 亿 token。
所有人都以为这是 DeepSeek V4。结果是小米的 AI 团队。消息公布后,小米股价涨了 5.8%。彼得说他记得当年小米做第一部手机时见过那个团队,三个年轻创始人,后来从手机做到电动车,现在又有了一个杀手级模型。
亚历克斯的第一点判断是:模型的扩散很难被遏制,因为前一代模型的存在,本身就是下一代模型的完整路线图。到这个阶段,一个合理的赌注是,一万亿参数的模型会在全世界扩散,门槛大约是 5000 万到 1 亿美元。而且这个门槛还会继续下降,因为算法改进一直在压低成本。
他解释了蒸馏这件事为什么"近乎魔法"。大模型先发布,然后用它生成大量合成数据,再用这些数据训练一个小模型,速度更快、成本更低,同时保留大部分能力。Anthropic 和 DeepMind 都在做同样的事。真正让他惊讶的地方在于,这个把戏可以一轮一轮玩下去,因为 5.4 本身很可能就是上一代蒸馏模型生成的数据训练出来的。他期待这条彩虹的尽头是一个相变,模型被压缩到几百万参数,像一个中子星,一 kilobytes 的文件装在手机里。
回到护城河的问题。戴夫的回答很干脆:数据才是护城河,模型本身不承担这个角色。很多人正在积累极其宝贵的数据,脑外科的、材料科学的、化学的。你可以用这些专有数据训练出下一个伟大模型,参数会流传出去,但训练数据不会。
他还给了一条对大学生很具体的建议:把亚历克斯刚才那段话反复听十遍,慢慢听,完全弄懂,然后让你最常用的 AI 去把这个思路泛化,再去网上找尽可能多的相关文档来读。走完这个过程,你就能构建一个蒸馏出来的专用模型,在某个具体问题上比地球上任何人都做得好。这就是现成的生意、现成的价值和现成的成功。
亚历克斯还分析了如果 Sam Altman 在场会怎么回答护城河问题。他可能说 OpenAI 在建自己的数据中心,但 Stargate 已经转向租服务器,这条护城河弱了。他可能说拥有世界最好的研究团队,但研究人员在流失,而且正在变成商品。他可能说十亿用户的分发优势,但企业客户可能比个人用户更值钱,而别家实验室用便宜的中国开源权重模型微调一下,也能做出看起来很惊艳的结果。结论是,基础模型正在商品化,价值在往上层框架迁移。
十二小时设计出一颗 CPU,压缩了九十天
第七个话题是"机器造机器"。Vector 公司一个叫 Design Conductor 的 AI 智能体,从概念到流片,用 12 小时自主设计出一颗 1.5 GHz 的、能跑 Linux 的 RISC-V CPU。传统工程团队做同样的事,周期是 90 天。一个季度的工程循环,被压缩成一个午休。
亚历克斯把这个现象放在递归自我改进的框架里看。软件优化软件的循环,正在咬穿自己的容器,往下吃到电子设计自动化(EDA)层,开始设计 RISC-V 核心。他预期它会继续往外吃,重新设计数据中心、重新设计能源供给,最后吃到整个经济。
他也给了一个怀疑论版本:RISC-V 有大量单元测试,可验证的奖励很容易定义,所以强化学习能跑起来。但这个怀疑论忽略了一件事:我们确实已经到了机器设计自己芯片的阶段。
彼得问:如果一个资深设计工程师年薪 40 万美元,一次完整流片要花几百万美元和一年时间,现在压缩成一个午休,那 5 万名硬件工程师去哪?
戴夫的回答是这个问题本身有问题。现在芯片设计成本太高,所以所有人对所有任务都用同一款 GPU 和 CPU,哪怕它对这个任务完全不最优。AI 设计解锁的是针对特定用例的专用芯片,效率可能高十倍。未来几年要在数据中心上花两三万亿美元,如果能针对用例提升十倍性能,这是几千亿美元级别的影响。5 万名工程师会全部被用起来,去设计几万种不同的芯片。
而且晶圆厂根本不在乎。掩膜每天换一种设计,工厂的吞吐量不变。所以"所有人都用同一颗 CPU"这件事,本来就是一个成本约束下的妥协,现在这个约束松开了。
俄亥俄想修宪禁数据中心,英伟达开始卖 H200 给北京
剩下几条散新闻,值得各说一句。
美国能源部宣布投入约 3 亿美元做 Genesis 计划,围绕制造、生物技术、能源等 20 个国家挑战,邀请团队用 AI 攻坚。亚历克斯认为美国终于开始有产业政策了,虽然 3 亿美元对联邦预算来说只是很小的一个脚趾头伸进水里。他的判断是,中国的产业政策对市场的扭曲程度远超美国,而美国的私人资本市场比中国深得多,所以整体上他更看好美国这边。
俄亥俄州农村有人在推动宪法修正案,禁止本州建设 25 兆瓦以上的数据中心。四个人一致认为这荒唐。数据中心占地极小,创造的财富极大,每吉瓦投资大约对应 100 亿美元。彼得说这些人没意识到自己错过了什么,"等他们晚上抬头看见天上的数据中心,会想念的"。戴夫指出,数据中心用水是循环冷却,不喝掉水。"我从没见过一个数据中心影响过水源供应,这个说法到处都是,纯粹是胡说。"
英伟达获准向北京出售 H200,也就是它最先进的芯片之一。彼得表示意外。戴夫的分析是禁令没起作用:中国既通过第三方拿到芯片,也在自研替代品,禁令反而刺激了本土的英伟达。所以政策反转了,问题是来得及吗。亚历克斯补充,即便卖,Blackwell 才是现在的 frontier,北京始终被挡在最前沿外面半步到两步。他还提了一个不太被讨论的角度:如果他是中方,现在会用审视的眼光看这些美国芯片的电路层,看看里面还埋了什么。"我们见过反方向的操作。"
四个人的问答环节
最后是 AMA 环节。几个问题值得记下来。
创业者该在哪里跑 AI 算力?本地硬件、AWS 云,还是 iPhone?亚历克斯说没有唯一答案。本地硬件控制力强、隐私好,但要自己维护备份。公有云把这些抽象掉了,但你要和其他租户抢 GPU,还可能扩大攻击面。iPhone 是终极边缘设备,隐私最好。他拒绝非黑即白的二分法,更愿意把它看成从边缘到核心的一条连续谱。他自己的答案是:我想把算力跑在戴森云里。
戴夫补了一条实操:Amazon Bedrock 上手很痛苦,但它做了最关键的一件事,把你和团队的 prompt 历史全部存进 S3 桶,让你的 AI 可以分析你做过的一切。别在随便什么硬件上跑,然后把 prompt 历史弄丢了。
人形机器人是不是过度设计了?戴夫的回答是绝对是的。为什么大家还在投人形?因为 AI 几乎是 overnight 出现的,现在是一场抢资本的竞赛,人形机器人视觉上更有吸引力,更容易融资,也更容易招人。这会先把零部件供应链解锁,然后才轮到那些种地、做衣服的专用机器人,而那些机器人大半不长人形。他去过超级工厂,那里绝大多数自动化设备都不是人形,是看起来就像机器的机器,人形机器人只负责操作那些机器。所以人形确实被过度投资了,但理由很充分:视觉吸引力和融资能力,是这一轮阶跃的核心部件。
还有人问:AI 能不能作为各方都信任的中立谈判者来结束战争?提问者特意强调,它扮演的角色是调停人,跟武器无关。彼得认为这是 AI 最被低估的应用之一。人类不擅长跳出自己的思维框架,让模型站在对方立场上重述你的诉求,效果惊人。模型可以吸收所有和平条约、谈判脚本、冲突解决框架,而且没有部落忠诚。双方如果都用 AI,并且把目标函数设成"达成一个双方都能接受的平衡方案",达成协议的概率会高得多。亚历克斯补充了一个正在发生的现象:商业谈判里,一方把僵局拿去问前沿模型什么是商业上合理的结果,另一方也去问自己的模型,然后很快就达成一致了。
还有一个问题:AI 解决大部分疾病之后,治疗多久能普及到所有人?亚历克斯的时间表是五年左右,这比传统三期临床的全流程还快。他提到 FDA 最近的两个变化:一是从频率学派转向贝叶斯学派,首次允许纳入药物之外的证据;二是某些情况下从两次临床试验改为一次。他的判断是,如果前沿实验室明天拿出五千种疾病的治疗方案,在足够的贝叶斯和计算证据下,FDA 会被政治压力推着走到零临床试验模式。
最后一个问题被戴夫接了:如果只能押一家上市公司?他不给投资建议,但指了条路。去 13f.info 查 Situational Awareness Fund,也就是 Leopold Aschenbrenner 的持仓,每季度必须披露。他的业绩很好,原因就是他只找"最内层循环"里的东西:晶圆厂、电力、芯片设计、直接衍生用例的算法公司。另外,未来三年 W-2 工资收入会被狠狠打压,而资产会涨上天。所以要去买资产,公开股票、私募股权、房地产,什么都行,只要它代表所有权而不是工资。
这期播客里,有六件事你现在就能做
整期听下来,宏大叙事很多,但落到个人身上,其实有几条非常具体的动作。四个人在不同地方反复提到,我把它们收在一起。
- 现在就开始存 prompt 历史。不管你用哪家云,先把员工和团队的 prompt 记录落到你能控制的存储里。这是目前唯一可审计、可分析的生产力数据。戴夫的原话是,别报销你看不见的 AI 工具,确保它跑在你的基础设施上。
- 用 AI 审 AI。把收集到的 token 使用记录喂给另一个模型,让它评估哪些 prompt 有价值、哪些是空的,输出一份"该培训谁、该调整谁"的名单。这件事一个人一个下午就能做完。
- 重新规划你的车库和停车位。如果你有房,个人车库在十年内会变成低效空间。现在就想想它变成卧室、健身房、工作室的样子。如果你在考虑买房,把"是否靠近停车场"从加分项里划掉。
- 如果你年轻且住在城市,先别急着买第一套房。萨利姆的建议是继续租,把钱投到一个风景好但目前交通不便的地方作为第二居所。等自动驾驶把可达性重新定价,那类资产会变得极其抢手。
- 去查 13f.info 上的 Situational Awareness Fund 持仓。重点在看一个专业投资者怎么定义"最内层循环":算力、电力、芯片设计、直接衍生用例。用这个框架去审视你自己的行业和持仓,比看任何宏观预测都有用。
- 做一次蒸馏练习。挑一个你熟悉的细分问题,找一份前沿模型的输出或者一篇高质量综述,用它生成合成数据,微调一个小模型,让它在那个具体问题上超过通用大模型。这是戴夫给大学生的建议,对任何有技术背景的人都适用,而且一个周末就能跑通全流程。
马斯克那句话值得再抄一遍:"为了理解宇宙,你必须探索宇宙。"这句话放在芯片工厂的发布会上听起来像口号,但把太拉工厂、星舰发射节奏、月球质量驱动器和戴森云串起来看,它读起来就是一份工程路线图的摘要。算力是探索宇宙的门票,而门票的价格,正在被一个愿意把地球产能乘以五十的人重新定价。
0:00Without question for me, the number one story this week was Elon's announcement of the Terafab.
0:06This is the most important endeavor in human history by far. In order to understand the universe, you must explore the universe.
0:14He's basically building a galactic [music] factory. On the left is 20 gigawatts. It's the current global output. And just the audacity of Elon's vision, [music] it has tremendous geopolitical implications as we discussed on the last pod. This could either accelerate or more hopefully [music] mitigate World War III and a Chinese invasion of Taiwan. We're going to need all the compute we can create. In fact, I'm actually kind of worried a self-driving car uses up basically a full GPU. When is it going to become illegal for [music] humans to drive? I think the thing that would make it later is purely the shortage of chips. Like the technology will be there and the demand
0:50will be there long before the chips are there.
0:52Figure out how to do more compute with less silicon for this exact use case and you'll be an instant [music] billionaire.
0:59I mean, we're heading towards a hundred trillion-dollar company. Maybe the largest, most important company [music] on and off the planet. Can we get there? Now, that's a moon shot, ladies and gentlemen. Everybody, welcome to Moonshots, another episode of WTF. Here with my incredible moonshot mates. DB2 in Boston. AWG,
1:23looks like you're on your home base as well. I am, but without my saucer separated Enterprise 1701D behind me. Oh, yes. I've got I I contracted one of my boys to create finally finally LEGO's come out with a Star Trek you know, LEGO set. I'm tired of all the you know, Star Wars LEGO sets. So, yes, 1701D.
1:46and it does do a saucer separation, but I'm not going to try it right now because a probably disaster may follow. And of course, we have Dr. EXO Saleem at his normal location at JFK. Saleem, how you doing, pal? [laughter]
2:02Salim is gone. Oh, okay. Well, so much for that. Yeah, it's as you know, listen, we try to be mobile. We're all dedicated to this podcast. But let's continue on because the singularity is not going to wait. So, today we're working to get you future-ready. A little bit of a format change. we're going to be having some deeper conversations about a more limited number of subjects, still covering the news that's breaking right now and there is a lot. But our mission is to get you excited about the abundance that's coming, show you the opportunities that are coming to you whether you're an entrepreneur, an investor, a student, a parent. And really, you know, this is the time to be
2:45paying attention to the supersonic tsunami, the most important tech in the world. And hopefully, this is your number one podcast on AI and exponential tech as well. Hopefully, Peter, there's more in here than ever before. It's it's bundled into themes that we can discuss, but if you just look at the raw news story count, it's it's as you would expect, exponentially exploding. you know, our goal, all of us, is to make sure that as we're talking about this on on Moonshots, that it's meaningful to the listeners, gets you excited, gives you context, helps you think about this in a different way.
3:22let's jump in. Without question for me, the number one story this week was Elon's announcement of the Terafab. he's basically building a galactic factory. think of this as putting all the parts of his LEGO puzzle together in an extraordinary fashion that is going to create massive capabilities.
3:44So, let me hit these these quick points and then we'll jump and discuss it. So, the Terafab is an objective across Tesla, XAI, SpaceX to build one terawatt of AI compute per year. To put this in context, the global output today is 20 gigawatts of AI compute. Again, we're measuring AI computation in terms of power, not just chips anymore. so, Elon wants to build 50 times the current production rate of the planet. he's building two kinds of chips, an edge inference chip for robots and cars, but also a high-power rad hard for his space Dyson sphere that is coming online. The fab is in Austin and it looks eventually like a hundred million square
4:33feet of capacity. one terawatt in the near term, near term, you know, single-digit years. long-term, a petawatt gets you only there from lunar mass drivers. Salim Ismail has joined the story. Hey, Salim, good to see you, pal. Hey, folks. Sorry, I'm bouncing around a bit, but I'm here.
4:55All right. In which terminal at JFK are you today? Where are you going? I'm flying to Brazil for 40 hours. Of course, you are. Of course, you are. You're a probability function on planet Earth. So, just to put this in context, check out this chart. On the left is 20 gigawatts. It's the current global output. And just the audacity of Elon's vision, a thousand gigawatts or a terawatt is his objective. I mean, you know, one thing I heard him say is, "Listen, I've been going to all the chip manufacturers out there and saying, 'I will pay you for as much production rate as you will give me. I I don't want to compete with you, but give me more and
5:36more and more.' And of course, none of them are moving at Elon speed. And so, he said, 'Screw it. I'm going to go and build my own production facility.'" And not exactly what he did in the launch industry. Right? Just lapped the entire existing launch industry and the autonomous car and electric car industry. He's playing his playbook over and over again.
5:58before I get into this data and stats, comments, Dave. Well, this is the most important endeavor in human history by far. Cuz it unlocks everything else. And you know, no no great surprise that he's announced it at the scale that humanity needs it. But the specifics on how you're going to actually physically do this are unknown because there are fundamental constraints to the number of ASML machines, EUV machines to do this and and many many This is the most complicated product ever made by humanity.
6:33and the supply chain is like like, you know, it makes cars look like child's play. So, so, he announced the mission. It's the right mission. The scale is crazy. It's you know, we estimated this on the last podcast at 50X all current productivity of chips and I I guess our estimate was dead on. So, so we got that part right. And he did allude to it last summer when we were meeting with him.
6:57And what what I was most curious about is how he was going to announce this and attract all the talent that he needs without [clears throat]
7:03irritating Samsung. You know, cuz he signed a 16 billion-dollar deal for for production with Samsung, which is more like 45 billion if it is going according to plan. And I guess one of the cover stories here is, "Well, these chips are for cars and they're also for space. They're hardened for space. So, they're not like the other chips."
7:22he's he said, "I will buy everything Samsung can offer me. But you're not offering me enough. So, I will still build all these chips and I will still buy everything you want to give me." You know, one thing I love and he pointed out in his Austin fab is that it's full vertical integration under one roof. So, that he can run rapid iterations on chip design. That's impressive.
7:48in our in our Austin podcast when we were talking to Elon, I asked him point-blank, you know, TSMC is being way too conservative in terms of their production of chips. They should be 10Xing their fab manufacturing. He said, "Well, you know, they're you know, the industry is cyclic. You know, maybe they're being conservative intelligently." [laughter]
8:08Which is hilarious in hindsight if you go back and listen to that audio because in the back of his mind, he's like, "Well, I'm going to build something 50 times bigger [laughter]
8:14anyways." So, but yeah, it is crazy that Samsung, Intel, and TSMC are not racing to build, you know, 10, 20X more production. So, Elon, of course, is well, he's going to do it instead. Okay, can I show you guys some calculations that I found just extraordinary here? listening to the presentation he gave 48 hours ago.
8:43You know, his target is one terawatt of compute per year in orbit. and he said, "Mass to orbit, 10 million tons per year." we're talking about an average satellite, his next-generation Starlink at a ton. Long story short, in order for him to launch that much capacity, it's 274 launches per day
9:06on Starship. It's a launch every 5.3 minutes. Which, of course, he says, "Listen, in the airline business, that's normal." But just the audacity and the level of of you know, thinking that Elon takes on is amazing. AWG, you want to you want to jump in? So many thoughts on this. Well, first of all, I think the elephant in the room is if Elon can indeed ramp up capacity for the Terafab in, call it, the next five years, which is the time scale that's being tossed around, it has tremendous geopolitical implications as we discussed on the last pod. This could either accelerate or more hopefully mitigate World War a Chinese invasion of Taiwan.
9:49If you look at the the moon aspirations, the the lunar aspirations for not a terawatt, but a petawatt of of from the moon. If you do the back of the envelope arithmetic for what would a a pedawatt of GPU compute that comes from lunar mining take, you run the arithmetic, comes out to be approximately 3/100,000
10:12of the lunar mass. So, a pedawatt coming from lunar mining with electromagnetic launches from the moon is starting to have a material impact on the mass of the moon. So, this is
10:24one big one big crater dug out of the moon. [laughter]
10:27So, that that's that's a pedawatt. As we scale, of course, to an exawatt of compute, and because why not, then at that point we're talking something like 3% of of the moon's mass. And this is, you know, when people think I'm joking when I talk about disassembling the moon or the moon had it coming, it this certainly paints a portrait. The moon did indeed have it coming, and it the moon is is slated for disassembly to to build the Dyson swarm. This is what it looks like. I I think more broadly, there are other sort of secondary implications, really interesting, that this is a joint Tesla xAI SpaceX maneuver. And many folks have
11:11speculated over the years, wouldn't it be wonderful if all of Elon's industrial ecosystem came together into one singleton. This is sufficiently cruxy with 20% of its production slated for Tesla and 80% slated for SpaceX, that this starts to look a little bit like maybe a cornerstone for some grand unification of all of Elon's projects.
11:33you, Alex, we talked about in a previous podcast, the idea that, and Elon said this, we'll see the first hundred trillion dollar company. And when we look at the numbers here, I want to show another set of calculations I did on what might the terrafab be worth in the ecosystem. I mean, we're heading towards a hundred trillion dollar company, and can we get there? And And at the day, my
11:56I don't know how you guys feel, but the Musk world ecosystem here looks like it will lap by an order or two of magnitude what Nvidia's done. It may be the largest most important company on and off the planet. Yeah, and I don't think Elon wants to unify all of his projects just for the sake of having one unified company. I think he wants to unify the the capital raising and the capital leverage with his massive multi-trillion dollar IPO and the massive joint mission unlocking an unprecedented amount of capital, which is what it's going to take to do these fabs in parallel at this scale.
12:35Cuz that's what That's the thing that's holding back Samsung and Intel and TSMC. They actually could do it. He needs 25 billion initially to turn on the terrafab and and get it, you know, get it the buildings started, so to speak. Right. I saw that in the analysis, but 25 billion is is just one fab.
12:52And here we're going to 50x the US or the world production. 50x the world production. So, he needs 50 of those 25 billion dollar investments to to achieve this mission. Yeah, there's a there's a lot to do. I thought I thought I had two three thoughts. I mean, one is like talk about patron saint of exponentials, right? Like this guy thinks at scales that very few people do. And it sounds incredible. It's classically the future of anything he's looking at looks vertical, and the past looks like flat and boring. what I what I thought was great was this is like amazing XLR logic cuz he's going exponential at the bottlenecks.
13:27You you stop competing. You're competing You're just redefining the game. And you're challenging anybody to dare to to come with you. I think that's like the amazing part of this. The launch cadence is is unreal, every 5.0 minutes. And I think that's exactly right. It forces the operating model to completely change. And it's forcing everybody to rethink that, including all the engineers and all the infrastructure, etc. If this is done in any kind of normal industry, one thing to point out is that his predictions on timing tend to be about 15 to 20% accurate. So, you know, but but it doesn't matter if he's if it takes him three times as long, who the hell cares? Like the fact that
14:08thinking at the scale and he'll get there is the fact that he's putting his planting up Yeah. Yeah, he's directionally correct. idea of like you shoot for the stars, and if you get to the moon, who the hell cares? We're landing somewhere amazing. [laughter]
14:21That's A W S G's plan. So, this is Here's the next question. a a a rapid schedule to disassembly of the moon, I think is what's on the table. [laughter]
14:32Okay. Good thing I don't [clears throat] have a glass of wine. Everybody, you may not know this, but I've got an incredible research team. And every week, myself, my research team study the meta trends that are impacting the world. Topics like computation, sensors, networks, AI, robotics, 3D printing, synthetic biology. And these meta trend reports I put out once a week enable you to see the future 10 years ahead of anybody else. If you'd like to get access to the meta trends newsletter every week, go to diamandis.com/metatrends.
15:00That's diamandis.com/metatrends. So, a terawatt of compute per year, if physically achievable, is this aspirational? What time frame? You know, he says five years. that's still 50x is is crazy, but if he's able to achieve that, you know, what happens to the you know, to the terrestrial data centers?
15:23And to the investments made in terrestrial data centers. you know, gentlemen, questions on that? every every chip, every investment in power and data centers is going to pay off tremendously. They won't cannibalize each other. We're going to need all the compute we can create and so much more.
15:38In fact, I'm actually kind of worried that the self-driving car is going to get cannibalized. You know, driving a self-driving car uses up basically a full GPU. And by the end of this year, a full GPU can also do brain surgery, or it can discover new math or new physics. And it's not clear that driving somebody around is going to make the price cut as the demand for compute goes to near infinity. So, I think the terrestrial data centers are going to be critical for national security for every country in the world because if something goes wrong in space, you got to fall back Your whole society will be running on these GPUs. Yeah, you can't you can't have an outage. So, anyone
16:15investing in this does not have to worry about one thing cannibalizing the other. And you The other thing you're going to see, I think it's later in the deck, but all the different process nodes All the fab process nodes are going to get used, even the older ones, the 3 nanometer and the 5 nanometer, are going to be running full throttle now, even if they're not as good as the new two and 1.6 nanometer.
16:35It doesn't matter. We need all the compute we can crank out. So, yeah, that it's it's going to be just an all hands-on-deck race, and Elon is is just documenting the up upper bound of what we can achieve as a as humanity. My expectation My expectation would be that in the process of I think Elon likes to call it production hell. In in in the process of production hell for realizing the terrafab over the next five years, I I strongly suspect we're going to discover some new not semiconductor physics, but more material physics and process engineering. It it seems improbable to me that Elon will just build the terrafab based on the existing stack as
17:16say TSMC did of ASML plus the existing optics plus all of the the conventional semiconductor processing techniques. If he really is looking to disrupt the space, he's going to want much more disruptive unit economics. So, maybe some of these technologies for semiconductor production and and fabrication that have been waiting in the wings for their right time in in the light, maybe this is purely speculative, maybe he'll look, for example, at alternatives to photolithography. The we It's not like as civilization, we don't have lots of alternatives. Exactly. And he's got and he's got superintelligence to help him get there, to design new systems.
17:57going to need the humanoid robots to be building the fabs. We don't have the workers. Salim? I think this is a really important point being made right now, Alex. Thank you for this, because, you know, you think about the secondary technologies and the benefits, like like all the carbon fiber that came from the space industries and cascaded down to everyday life. The secondary inventions that will be needed here will be massively beneficial to humanity. Well, this this is one of the most exciting things in tech history, in fact, the most exciting thing in tech history is what Elon was talking about in Austin, is laying down single atoms using a some kind of a self-organizing
18:31process. And I feel like Alex is exactly right. Something will get discovered in the next year or two using current LLM AI running on GPUs that will then dictate a very different non-lithography future. But it'll probably be five or six or seven years before those start getting manufactured at the same scale as lithography.
18:52but it's super exciting to to watch. Dave, I asked Claude to give me an estimate based upon all the data that we got from Elon during his his talk on the value of future terrafabs, right? And so, it makes the point here that initial capex, as you said, Alex, you know, it's 20 25 billion, but the real capex for buildout is going to be on the order of 150 billion dollars at a minimum. It may be a half a trillion there. And then he talk And then the the model here looks at what's the annual cost savings for his captive opportunities, right? So, if he's building the chips himself and he's putting them in Optimus and then cybercabs, and then there's external
19:33revenue, and then there's an applied enterprise value. And, you know, it's on the order of a trillion to multiple trillions. TSMC is valued at 1.7 trillion, and what we talked about last week was that terrafab is expected to produce on the order of 70% of TSMC's output. So, yet we're layering on another multi-trillion dollar opportunity. These numbers seem low to me. If you're really generating Yeah, if you're generating 50x Yeah, 50x the total output of AI chips on the planet.
20:07Yeah. this this is operating This is operating a different still a car company and it's not. Yeah. Well, not only that. I mean just right out of the gate, this is not just doing what TSMC does. This is TNC plus Nvidia. You know, Nvidia's worth four and a half trillion. Yeah. But but I mean even that's ridiculous as an analysis. We're talking about 50 times the production of the world's current compute.
20:28so, you know, out of the gate you would take TSMC plus Nvidia and multiply by 50 to get you know, a a starting point for an estimate. So, this is off by over an order of magnitude, well over an
20:41Any of you Any of you concerned about a monopoly here? No. If you're following the the prediction markets for the SpaceX IPO, this is already starting to get priced into the SpaceX IPO. So, SpaceX IPO was originally going to be 1 and 1/2 trillion. Now prediction markets favor 2 plus trillion dollar SpaceX IPO pricing in the SpaceX portion of the terrafab.
21:03So, in in some sense again, I'm not quite clear on what the government structure is here is going to look like and and how clean it's going to be, but to the extent it falls mostly in the SpaceX bucket, SpaceX IPO, not an investment advice, obviously, could end up being as you say, SpaceX plus Starlink plus Nvidia
21:23plus ASML plus TSMC all rolled into one. I I just as a as a piece of advice for entrepreneurs out there, just understand the level of audacity that Elon is looking at. He's building in in a [clears throat]
21:40sort of multiple orders of magnitude beyond anybody else. And from his first principle thinking, he's looking at where are the blockages for my growth. And we had this conversation. People are not generating enough chips. I need to build a chip fab. And then he doesn't just go out to say I'm going to buy Intel or build a chip fab equivalent to what TSMC is building in Arizona. No, if I'm going to build a chip fab, I'm going to build something that is 50 times bigger than the world supply.
22:09amazing. Well, also he's thinking two moves ahead in the big chess game and you know, two moves used to be 20 years. Now it's six years, but but two moves ahead of everybody else and this came up. Alex and I were talking earlier this week about fusion energy and you know, Elon doesn't talk much about fusion energy. Why? Well, because he's visualizing solar in space and and you know, solar on Earth is a great stepping stone to solar in space and it requires panels and batteries and cooling, but it doesn't require turbines and and fusion reactors. So, he can skip a couple of hard steps and go straight after the next move in the big chess game. So, it's really interesting to
22:46watch how those those timelines have inverted. You know, the Dyson sphere is now right on our radar and even Google is talking about it, which means everyone's talking about it. So, it that came to the forefront really just in the last month or two and so the whole timeline of humanity got shifted. So, the Dyson sphere will come before anyone even figures out how to get licensed for a fusion reactor.
23:07Can I be cynical Can I be cynical just for a second? Yes. The plans are ridiculously grandiose and if anyone is going to achieve them, it's Elon, but curious that the timing of this is just leading into the IPO to get everybody excited about things. So, that would be the cynical view, but I still I still just love the audacity.
23:26Yeah. Well, I think Saleem, your your first observation was dead right. If you shoot for the Mars and you end up at moon, you're still a way up. [laughter]
23:33So, I I want to ask the I want to ask the question again that AWG said, "Nope." Which is monopoly concerns. Do you believe I mean if he's really generating, you know, I'm not worried about it because when you have an MTP, which he does, you're basically operating on this massive mission. You may have ethical issues here and there, but generally the trend is so positive and so beneficial for humanity, who the hell cares?
23:58Well, I mean humanity becomes There's always another person. So, somebody who's like Brendan foodie maybe. You know, somebody like that right now is like, "Wow, I'm going to do this, too." And you know, it's just the way it is and that person, we don't know who they are yet, but they'll emerge and you know, you can't exist in the US without antitrust action if you don't have a competitor. So, Elon will invite that competitor, whoever it is and and it'll be great. I'll say one of the things here. If If we label this monopolistic behavior, then don't we have to label everyone with an MTP a monopolist? Mhm. Mhm.
24:34Just asking the question, my dear mates. Market dominance if you if you would get there, then that's true. But you haven't even picked the real estate site yet [clears throat] for the the terrafab. It doesn't He hasn't even picked a final location. I I think it's way too premature to declare this monopolistic behavior being so ambitious as to build the Dyson swarm. We're going to have multiple Dyson swarms.
24:54I'm with I'm with you. Love it. I'm with you. All right. Well, well, look, Google Google isn't going away. You know, Google has 300 billion dollars of revenue, 100 billion plus of cash flow, their own chips and design. they have all everything except the rockets. So, Google's not going to go away as a Well, I don't forget Eric Schmidt Eric Schmidt is trying to bring Relativity Space online. So, that the rockets are at least part of Google family. Oh, you know, they're going But let me let me just close Let me close this There there's plenty of them to go around. Let's not forget.
25:26One thing that's important here is every time there is a constraint, the judo move is to realize it's a massive opportunity. And so, this is an abundant story once again. This is a a massive increase in abundance of AI compute beyond what anyone was speaking about just a week ago.
25:46So, I'm glad he's focused on this and less on the politics. Mhm. Yes. I'm in. All right. Here is our second conversation story, one that I'm excited to have with my moonshot mates. It's about the future of human transportation. You know, robots are getting their driver's licenses, flying cars are taking flight. here's some of the data and I want to go deep on this because I want everyone listening to understand how this is going to impact how and where you live, how you commute, every aspect of our lives. So, Waymo hit 170 million fully autonomous miles, equivalent to 200 human lifetimes of driving with 92% fewer serious crashes. So, a significant
26:33reduction in crashes. At current, they've got 3,000 vehicles in 10 cities. Still early, right? Uber has now invested 1 and 1/4 billion dollars in Rivian with plans to deploy 50,000 fully autonomous robo-taxis. Here's a look at Waymo versus human drivers. Waymo is doing an extraordinary job of of saving not necessarily lives, but saving crashes, you know, minus 92%.
27:05So, I think cyber cab, we've seen incredible data like this also on full self-driving from from Tesla. Check out this image. This is Joby Aviation. Joby had started this company. He started Velocity 11, took that money after he sold it. he was partnered with Rob Nail, Saleem. And Joby started Joby Aviation God knows over a decade ago and here it is flying over the Golden Gate Bridge in San Francisco.
27:33It's a a beautiful image. This is a eVTOL, electric vertical takeoff or landing. It's a name that rolls off the tongue onto the floor. I'm calling them flying cars cuz that's what they are. And here's what's going on in the eVTOL world. I think really important. So, Joby just is now in testing for its
27:55first FAA conforming aircraft, meaning it's it's demonstrating to the FAA that it can build a reliable design over and over again. It just had demonstration flights on the Golden Gate Bridge. Joby and Uber announced Uber Air powered by Joby. In fact, when when Travis left was before he left, he had created something called Uber Elevate and they were doing their earliest work on flying cars.
28:25I keynoted their their talk there, but Uber Elevate got sold to Joby and now Joby and Uber have a partnership. The other company in the United States that is a competitor to Joby is called Archer Aviation. they have a beautiful aircraft called Midnight and they're the first company to achieve 100% FAA acceptance of its eVTOL aircraft means of compliance. Long story short, we've been waiting a long time and flying cars are almost here. We're going to start to see them operating in the US in the next 18 months. Should be here in LA in 2028 in a big way.
29:04and so, here are some of the conversations, gents. first off, when is it going to become illegal for humans to drive? Saleem? Yeah. I think pretty you know, you'll start with city centers, right? And you'll be illegal to drive in the city centers. Then it'll slowly broaden out from there. I think what's I'm I'm the flying car is the most exciting technology for me personally given that I'm commuting to airports a lot that I could ever ask for. So, this is this is 10 years later than I wanted it, but finally it's finally happening. What I like about this These are not transportational stories. This is full urban redesign is what the narrative is. Essentially, you
29:44make Our becomes abundant. Now, land has always been scarce. Real estate has been scarce. Real estate becomes abundant. If you fly across the US, it's empty. Right? And we've talked about the statistic just around between Toronto and Chicago airports, there are 10,000 islands and lakes, right? So, we do not have a scarcity problem. We have a mobility and accessibility problem. So, I think I'm super excited by this particular model.
30:10I've got my 2 years for us to get to full autonomy before my son gets his driver's license. He's 14 right now. So, I'm I'm pushing hard on this race. Mostly, he wants to get it to get away from parents, but that's fair. so, we'll see what happens. But, you know, you compress travel time, you re-price real estate. This is such a huge thing. And I got to just shout out to Joe Ben just because it's hard to build a hardware platform like this and to do it over a decade with all the inevitable regulatory and market structure against you and infrastructure against you. This is a huge This is like a Nobel Prize in patience here.
30:46Yeah. Incredible. Dave, you were going to say? Oh, the EVTOLs are going to move very quickly cuz they don't they don't run the risk of you know, crashing into houses like cars on a self-driving cars on a road do. They're all going to be autonomous on birth. That that's the new thing.
31:04EVTOLs have been in the works for years, but the AI that makes them self-flying, self-driving, and super safe is here all of a sudden now.
31:11True, but the first airplanes are going to be piloted, right? They'll be single pilot, four passengers in the back. The the goal is rapid recharge at the vertiports when they land recharge. Probably an average length of flight of under 10 km, I think. you know, going from Santa Monica where I am to the Dodger Stadium and avoiding the 10.
31:34But, autonomy will come with enough with enough data and enough demonstrations. Wait, why why won't they be Why won't they fully autonomous from the get-go? I mean, it's there's Cuz the It's called the FAA. FAA. The FAA is not happy until you're not happy. [laughter]
31:53Yeah, that's exactly it. The manufacturing of these wants to happen right away. And the the AI command and control is being worked on for the car, not for the EVTOL yet. So, there'll be a couple year very short period of time, in my opinion, 2 years or so. Cuz you know, in in the Middle East, they're already doing the self-driving, self-flying version of this. So, should be very short window where people get to fly these. you know, the bullet here though is when does it become illegal for humans to drive? I think that's going to happen very quickly as well.
32:21very similar to indoor smoking or drunk driving. There's a tipping point where a lot of voters say, "Wait, you're putting my children at risk with your crappy driving." Mhm. That that's ridiculous. We got data and proof here that the the self-driving is 90% safer, [clears throat] soon to be 95, 97% safer.
32:44And and you know, the human tragedy that comes from car crashes is is unbelievable, shocking. And so,
32:50for under 5-year-old under 5-year-old kids, it's the number one cause of death. Yeah, accidents. Yeah. Oh, and and you know, the the it's devastating to families, too. It's just absolutely tragic.
33:00in the first world. Yeah. Well, there'll be a there'll be a TV ad campaign probably 3, 4, 5 years from now with lots of ugly images in it. And then there'll be massive amounts of voting, and then people will say it's it's inconceivable that you would drive on a public road. That's inhumane. Go drive on a test track, that's fine. Maybe you know, some country roads, that's fine.
33:21But, but no way, don't put my children at risk. So, I I think that's going to come as soon as we have the manufacturing for the cars themselves. But, I I think the thing that would make it later is purely the shortage of chips. Like the technology will be there and the demand will be there long before the chips are there.
33:38So, if you want to unlock this as an engineer, figure out how to do more compute with less silicon for this exact use case and you'll be an instant billionaire. Alex, your thoughts?
33:51this format, Peter. This is like an internal AMA. So, I'm going to try for a lightning round on on all of these. Oh,
33:58Leave some Leave some room for Leave some room for the rest of us. Let's take it one at a time. One at a time. At what point does it become illegal for humans to drive? I think never. I think we'll simply redefine driving to represent higher and higher levels of abstraction. So, right now with like FSD14, you tell it where you want and if if you're running the most recent subversion, you can have a conversation with Grok and you can do minute steers along the way. I think that notion will get refined such that driving gets redefined to be sufficiently abstract that it's always safe for pedestrians. It's always the human on the loop of the AI driver. So,
34:34it's effectively a human-machine hybrid, if you will, that has the safety of the machine, but makes the human feel like they're in the driver's seat still.
34:41I I said I said this with when Dara was on stage with Salim and myself. I said there's a version in the future of self-driving where you're driving and you can push the car as fast and as hard as you want and the car knows its own limits. It knows the traction of its tires. It knows the road surface and it prevents you from doing something stupid, but you're in control of it 99% of the time, but the 1% where you're about to do something that will destroy you, a person, or the car, it stops you.
35:09Exactly. I think like the the future the future of the accelerator pedal isn't the accelerator pedal. It's if you use FSD, it's turning the driving mode up to Mad Max. That's sort of like a an abstraction of acceleration.
35:22that's that's all I use is Mad Max and it still doesn't go fast enough. So, I have to step on the pedal.
35:28There used to be an ad saying, "Friends don't let friends drive drunk." And so, you can just keep that ad and drop off the drunk part and go, "Friends don't let friends drive." Period. So, all the all the messaging is there. All right. So, so AWG, why don't you Why don't you kick us off on question two here?
35:44Okay. Question two, with Uber partnering with Waymo and a bunch of other names, will the cybercab be able to compete? I think we mean compete here. Yes, of course. it's going to be very competitive market. Period. Yeah. And I love the fact that this is driving us towards abundance, right?
36:01This is driving us towards UHI. If you've got a dozen companies delivering autonomous vehicle services in your city, they're going to be competing against quality of service and price and just bringing the price down to a minimum amount. Now, one of the things that's interesting about cybercab is that that's going to be priced at probably 30K is what roughly what Elon's announced. And he's going to allow people to buy it. So, you know, one of my goals is can I buy, you know, 25 or 50 of them here in Santa Monica and own them, but have them going out and and basically generating revenue for me and for my my cybercabs. I'm sure they'll have some level of
36:46of personhood by then, Alex. I think so. I I I I never would have guessed, Peter, that that your next gig would be as a cabbie, but the singularity makes for strange bedfellows. Fleet owner. Fleet owner. I'm I'm a comedian and I I think Yeah, please.
37:01the big impact for me when I see this is the complete collapse in the market structure of cars. Today, we make close to 100 million new cars a year. And they sit empty 94% of the time. So, even if you drop that by 50% in utility, you you basically collapse the need for past the car industry instantly. And if these cars maintain for a long time, the lifetime should be near infinite. My Tesla Model 7 2017 should it'll go a million miles. There's nothing wrong with that car. So, this is going to completely change the nature of car services, car maintenance people, like the complete industry gets reshuffled from the bottom up.
37:44Yeah. Yeah. We'll think about the implications of that, too, Salim. Right now, if you take an Uber from SFO to San Fran for like 200 bucks or whatever the hell it is, it's almost all driver costs. So, even before you shrink the number of required cars by a factor I think the estimate was 5x.
37:59Is it 10x that savings? Is it 10x? So, then then but the driver's already the majority. So, you take the driver out of the loop. So, the cost of that ride should go down, you know, at least 10x cuz the car's coming down 10x and the and the driver is more than the car anyway. I think the number I've seen is 20 between 10 and 30 cents a mile.
38:21Yeah, the the I've seen it as a four- to fivefold cheaper than owning a car. The the next question I want to ask and and offer my points of view is I think one of the most important one for our listeners. This is going to have a profound impact on your real estate holdings, where you live, what you do with your real estate. So, if we have autonomous vehicles and we've reduced the number of vehicles on the road by 10x, let's call it that. and these vehicles don't need to park. Again, my my current version [clears throat] of this is I get up from the breakfast table with my family, I walk towards the front door. My AI knows that I'm moving to open the front door. It knows where
39:04I'm going. It's ordered an autonomous vehicle, what I call automatically, for me. I haven't had to ask. And so, all of a sudden, you know, we had In our home here, we had a three-car garage. We already converted one of those garages into an extra bedroom. the other two garages have become effectively storage and I'll build out, you know, probably a workout gym and so forth. I think the idea of a garage, a personal garage in your home, goes away. So, start thinking about what you're going to do with your garage space. What are you going to make it into because you're not going to own a car.
39:39You might want access to a car, but most of the time do you really like driving? I mean, when you get into an Uber, do you ask the Uber driver to get out and let you drive? So right. And you know, this part of the conversation is incredibly actionable for all of our listeners. It doesn't [clears throat] rise to Alex's level of, you know, like change the world tomorrow, but it really matters to almost everyone who listens.
40:04Everything Salim said and Peter said is dead right. If you're young and you're you've got a job and you're living in a city, which is 60% of you, you might not want to buy in the city. keep renting and look for something that becomes your second home later in life that's in a beautiful spot. Yes.
40:26That's a little harder to get to that is going to be incredibly coveted. Imagine a world where there's 10x more wealth about 2034, 2036. And this is a spot that anyone in their right mind would want. And actually, if my wife is listening, close that transaction that you kicked off this [laughter] weekend, even if you have to pay a little more.
40:48But yeah, that's what you that's the life plan you want because accessibility, not not just getting to it, but also delivering things to it. Like, you know, your Starbucks or Dunkin' Donuts is going to come by drone. Absolutely.
40:58changes what you want. Think about it. And there aren't, you know, we have a huge country like Salim said, but the really great spots are limited. So really do your soul searching and look look for that thing and don't buy near an airport in a city. Island real estate is going to become, you know, 10x, 100x more accessible. That will drive the value up. And in a downtown LA, you know, it's like I don't remember the figure. It's like like 30% of the blacktop is parking. All of that gets released to become new. 60%
41:3460 60% of the land area is parking spots in Los Angeles. That's crazy number.
41:39That's insane. Well, that becomes that becomes gardens. It becomes green land. It becomes parks. That's incredible.
41:45and think of the unbelievable space we use in in stadium parking lots, right? Acres and acres and acres of rows and stuff. So we'll have to rethink, you know, tailgating and everything. There's so much available, you know, business opportunity here. If you can think ahead of what you will do with that. And if you're in the parking garage in business, you got to think ahead as well.
42:09Yeah. A couple of other second and third order implications, if I may. I mean, so we've already touched on I I think the more obvious ones, parking garages, et cetera, need to be reprogrammed for other purposes. Another, I I think, borderline cliche implication of full autonomy everywhere is the re-spread of suburbia. Why invest so much in urban center real estate if you can be effectively connected to an urban center or even not even need an urban center if AVs take you everywhere. Basically a virtual subway from from anywhere to anywhere. So re-suburbanization, if you will, at least in relatively low population density countries like the US. I think these are pretty cliche
42:51implications. A a less cliche implementa- implication, in my mind, is what if we just take this trend and extrapolate it fully to completion? What happens? I think there's a future I I put out a request for startups around this idea of why not just create autonomous Winnebagos, the the equivalent of having entire office buildings that are themselves autonomous vehicles. One could imagine living in an autonomous vehicle. It's all part of a social network. When you need to take an in-person meeting with someone, your two AVs are part of the social network and they connect and synchronize all of your locations. So maybe you're in Boston in the morning,
43:30but you're in Washington, D.C. in the evening. This is all handled automatically to synchronize your calendar with your AV location. And then it's a sleeper car and then you're in Chicago or wherever the next day. So Your bedroom car You become Your bed You become humans become to join you.
43:48Humans become internet packets that are being routed by the autonomous vehicle system. Yes, love it. Love it. Love it. Love it. you know, the EVTOLs, I it's taken a while. There is still a lot of doubt people have about EVTOLs. you know, the opportunity we have is going to be limited by the size of these, being able to land locally. So there needs to be sort of local hub and spoke vertiports. you know, somewhere within 5-minute driving and gluing these all together. And that's what Uber wants to do with their platform. So I hop in my autonomous Uber. It takes me without thinking to the right EVTOL site, which takes me to another location 10 km, 20 km away. And
44:32then I'm in another autonomous vehicle. What I'm missing from all of this, Alex, is is the hyperloop, right? So I actually joined one of the first hyperloop companies. Virgin got involved. it was we raised, you know, probably close to a hundred million dollars. It didn't go forward, but the material science of creating hyperloop, and of course, the benefit for hyperloop right now is effectively supersonic travel point-to-point inner city and inner city, LA to San Francisco, LA to Las Vegas. That one will be busy. so got to see hyperloop on this list eventually.
45:13Peter, which do you think you're going to see first, like in in prac- in practice for, say, New York to Los Angeles? Do you think you're going to see hyperloop first or do you think you're going to see rocket cargo first where you hop on an Elon Starship, go up, go down? You know, I I've thought and looked at point-to-point rocket travel and it's a tough it's a tough thing, the energy dissipation cuz you're basically going you're going to orbital velocities and you're having to re-enter over or near a city. I guess the version that Elon put forward was offshore landing facilities. That's right. So that you're, you know, a kilometer offshore. [snorts]
45:51I think for one reason, rocket point-to-point travel, because Elon's behind it, and because the vehicle exists and they're going to be launching every point 5.3 minutes. That's right. and, you know, Elon almost got involved in hyperloop. But like you said, I can't do everything.
46:13Anyway, I'm curious what I have two thoughts here, quick ones. Yeah, please. Yeah, yeah, please.
46:17One is I think hyperloop will be used largely for commercial and for container loads rather than human beings cuz then you don't have to worry about G-forces and the safety standards can be lower. and the second is remember that all yeah, although it takes us like 3 hours to fly from New York to Miami, that 3 hours on a plane today is way more productive than it was, say, 10 years ago. You've got full internet, you can work. So you can I want to see megabytes, baby, on Starlink.
46:44we can schedule ourselves now to do things when we largely want to do them. So I think that's a huge opportunity also. So for our listeners, I would love to get your feedback on this format where we're going deep on a topic and having a conversation, trying to educate you about how we think about about in this case transportation or previously TerraFab. Our next conversation is the great reshuffling. Job loss is inevitable. The only question left is what we build on the other side. So here's some of the stats and some of the articles that came out this week that have us thinking about this. Goldman says AI could automate 25% of US work
47:24hours. Seems like a low estimate to me. A PwC told its partners, "If you resist AI, you have no place here. AI tool yourself or get out." G42 posted a job listing exclusively for AI agents. Is this sort of a gimmick? Is it real? And I love this one and this come came from sort of a a hit from Jensen. Companies are now tracking individual employee AI token usage. And Jensen came out saying, "If a $500,000 engineer didn't consume at least 250,000 tokens, I'd be deeply alarmed." You know what this reminds me of? This reminds me of De Beers saying, "3 months salary to buy a diamond ring." Right?
48:11Doesn't it? I mean, it's like forever, Peter. I mean, [laughter]
48:14token talk forever, that's great. So I mean, literally, Jensen is saying he's taking the total salary, you know, of all engineers on the planet and And then Perplexity AI won in appeals in court to continue running shopping agents on Amazon. So I'll show one chart here and then let's talk about it. So AI could automate 25% of all work. this is Goldman's chart showing each of these columns is a different type of work. And I guess the median here is about 25%.
48:54We've seen this lots of different places in different versions of it. But let's jump in. So Salim, you work with more consultants than I I do than anybody here does. So what do you think about this PwC telling its partners, you know, adapt or die? Yeah, I think that's fine, but I think it doesn't go far enough. And same with the McKinsey's thing. So the the calculations I've been running as I kind of get this paper organizational singularity paper finalizes that you'll be able to run I'm I'm just give me a couple of days, it'll be ready for like draft and review. We're just doing some narrative thing. the the you'll be able to run a typical company with between 20 or 25% of the
49:37employees you have today. Cuz all workflow goes from human to human to agent to agent, right? Now, you could take on the doomer side and go, "Oh my god, 80% job loss." But no, because we're just going to be creating four or five times more companies. And also for bigger companies, that transition to an AI-based workflow is going to take much longer than for a startup or mid-market.
49:59and therefore there'll be time for the economy to so so I'm actually suggesting that we want to have no pattern recognition in jobs. Almost zero. Okay? Now, definitely partners who resist AI will have no place. There's also some need to say that consulting partners have no place in the future because in the future if you have an AI agent figure out your strategy, why do you need a consultant firm? You're going to need that for more for implementation and if they have better agents than you do, I think that's where we'll end up with that.
50:28Alex, what are your thoughts on these? You want to pick one?
50:32difficult. So, on the PwC story, it's very difficult for organizations to self-disrupt. So, if you're a management consultancy or an accountancy, some other bill-by-the-hour heavy firm, it's very difficult for for you to willingly voluntarily transition to an outcome-based pricing model versus an input-based pricing model. So, I I I take the you'll have no place comment. I I interpret that as an attempt to self-disrupt. In practice, it's very very hard to to do that. and the the whole point of of Schumpeter and disruptive innovation in general is most of the most of the macro replacements for in in this case, for input-based actors in the economy are probably going
51:19to come from other firms, not from large firms self-disrupting. On the G42 story, I think it's actually really interesting. I I looked closely at the G42 job listing and it it really is a job listing for AI agents. And one has to wonder, yes, like around the edges, they also ask for details from the developer and what was used to make sure that this was really an AI agent submitting itself for I think it was a marketing job. We are so painfully close, I think, to a near future where there's a sort of reverse discrimination against humans and where humans need not apply [laughter]
51:58and up ends up being an epithet on so many jobs. Well, you have that already with the with the about PwC partners, right? If you don't use AI, get out of here. That's essentially we're getting to that point. We're halfway there already. Mhm. That's that's PwC, though, which is a human-oriented business basically trying to force humans to self-automate, at least from a a unit pricing perspective. Yeah. This is born AI jobs where humans need not apply.
52:27Agreed. You know, Saleem, one thing that you and I do for large companies that I think people need to understand, most large companies out there are walking dead. Their business models will be fundamentally disrupted in the next two to five years. And so the question is, how do they disrupt themselves before before someone else does? And the answer is, it's really hard, almost impossible. And so, you know, what you and I have done before is invite superb talented young entrepreneurs to come in, hear the company's business model, and say, "This is how I would disrupt you if I, you know, was funded to do it." And then the company should fund the best of them,
53:09right? And we've done this. fund the best of them to actually build a adjacent company who's intended to disrupt the primary company. And then literally Yeah.
53:23I I the company I the design firm the design firm IDEO actually did this. They realized that their methodology would be widely known and they couldn't stop the leakage of that. So, they they picked one of their crazy partners and said, "Go to the edge and build build the disruptor." And he created an open IDEO marketplace of of design ideas. It was amazing. One one caveat to what Peter said there, the private equity guys are having a field day with AI automation and if if a company has great cash flow, even if its business model is doomed in the age of AI, the profitability is going to go through the roof in transition In the in the near term
53:58humans In the near term, yeah, cuz you know, then AI can do the job for 10%, soon to be 2% of the cost of the human, you know, with no no labor laws, no overhead, no you know, insurance, no health insurance.
54:12if you've got good cash flow, there's an entrepreneur looking at that salivating. Coming to eat your lunch. Yeah, so what happens is the private equity guys will come in and say, "Hey, cash flow cash cow with great cash flow, we're going to buy you or buy part of you and then we're going to AI-ify your business and that'll drive even more cash to the bottom line and then we're going to use that cash flow either as a vehicle to launch new things like an incubator or to attract that entrepreneur or to just roll up those startups and you know, acquire them back in. So, it becomes kind of a centerpiece. By the way, both Anthropic and OpenAI are setting are partnering with private
54:52equity firms to do exactly this. Go buy companies and then AI-enable them cuz you can do it with the owner. As if as if they did anything in the world, just announced a new hundred billion dollar fund to do nothing but this. So, Jeff is doing it. You know that.
55:07have enough money already. Alex, I'm curious or Dave, I'm curious about your thoughts on the fourth bullet here that companies now track individual employee AI token usage and you should have a minimum token usage per employee. Thoughts? Dave, do you want to go first?
55:22we we already implemented Yeah, sorry. We we already implemented targets across all of our companies on this and we're targeting 80% token, 20% salary. And I think that's a really it's very similar to what Saleem said a minute ago. there's going to be huge amounts of job disruption in the next two or three years and then it'll turn around and by 2030, 2032, things will be good again.
55:45But but what you want to do is be one of the 20% that's still there when it's 80% token cost, 20% human cost cuz no employer in the world, including all the companies on the controlling shareholder of, care about cutting the last 20% of payroll. It's not a priority at all cuz at that point, an employee that can improve the efficiency of our AI even 1% is worth a lot more than cost-cutting.
56:14And so, we're in this footrace now to 80/20. Jensen's got a stepping stone here of of 2/3, 1/3 token cost, but that's going to be very transitory. We're we're racing toward token costs being much much bigger than payroll. So, I have an immediate step at 50/50. But it's coming soon. Sorry, go ahead. Alex, is this the right metric? I mean, cuz you can waste tokens. I mean, it's got to have a a different harness, right? You're you've got to be measuring something else besides just token usage.
56:42can waste tokens except in this AI-abundant era, you can also ask another AI to look at all the tokens a given employee used and ask, "Was this a good use or not?" Or "Was this just vacuous?" So, exactly.
56:54It becomes the ultimate self-licking ice cream cone. The the the quote from from Jensen, I think, is interesting and it's it's sort of if a half million dollar engineer didn't at least spend a quarter of a million dollars on on inputs that ultimately flow back to Nvidia, I'm deeply alarmed. So, so there's a little bit of circular [laughter] there's a little bit of circularity there that I I take with a huge grain of salt. No, it's the beers and the diamond ring. [laughter] Really is.
57:20That's right. There's another side to this, which is the employee side. So, I I I talked a bit about this in in my newsletter. At some firms, especially the frontier labs, employees are actually competing, so-called token maxing, to max out their their token usage on internal leaderboards to see who can use more tokens than than the other person. So, it's not just sort of Big Brother top-down, it's also bottom-up. I can use more intelligence more superintelligence than you can. And I think this is ultimately probably pretty healthful. to your original question, Peter, about whether tokens are the right unit of productivity, I I think what's interesting is tokens they're they don't
58:00even have to be the right measure or right unit of productivity, but they're the first measurable unit of productivity.
58:07Yes. Hours are are certainly measurable. Like, you can punch clocks And you and you know, Yeah, it's useless. It it's naively measuring inputs, but tokens where you can actually like they're introspectable and they're legible and you can ask you can spend other tokens to look at the tokens the the primary tokens and decide, "Are these valuable tokens or not?" For the first time we have legible, defensible, analyzable inputs for employee productivity and that is a sea change. Yes. Dave, what's the advice here for CEOs? The advice to CEOs here for you. Alex completely nailed it.
58:41Like, worrying about whether the tokens are being used intelligently or not is not a problem at all in the real world. So, so Jensen's metric is perfect. Just measure the spend on tokens. And then Alex's insight that you must the most important actionable thing is make sure you gather all of the of the prompt strings history Yes.
59:04for each and every user. Cuz that's the AI can analyze the efficiency
59:08Analyze that. Yes. Exactly. And you can say, "Hey, you know, I've got a hundred or let's put it realistically, I've got eight direct reports. Evaluate the quality of the prompts and the output they have and give me feedback on which bottom 20% I should cut or or train up." Well, and really practical advice, if you use any of the models on Amazon Bedrock, the grabbing of the prompt history is already built in. It goes right into S3 buckets. I'm sure you can do it elsewhere, but our company just be happen to be using it on Bedrock.
59:39so it you literally don't have to build anything to start doing this. you just need to grab the data and feed it into another AI, which you can also do on Bedrock or you can do on you know whatever. personally I I like using Cloud 4.6 for the stuff, but you just got to close that loop. But but the key is grab the data right now before people get used to using their own home account or or you know something outside of your purview.
1:00:05Do not reimburse people for AI that you can't see. Mhm. Make sure it's on your infrastructure. Welcome to the health section of Moonshots brought to you by Fountain Life. You know, my mission is to help you use the latest technologies including AI to not just do your work at home, teach your kids, but to help you live a long and healthy life. I'm here today with an extraordinary physician, the chief medical officer of Fountain Life, Dr. Don Meusel. Don, let's talk about cancer. you know, I know from the member database that we have at Fountain, our members who come in who think they're healthy, it turns out 3.3% of them have a cancer in their body they
1:00:45don't know about. That's right. You know, the majority of cancers that we screen for, those aren't the ones that are necessarily taking the lives when found at a late stage. We know that when cancer is found early, the chances for cure are much higher. We know it's much easier to treat a cancer when found early versus when found late. What we're finding in our members is over 3.3% were found to have these cancers that were otherwise wouldn't have been found or detected. Yeah, you know, it's interesting people you don't feel the cancer until stage three or stage four.
1:01:16And and if you don't know what's going on inside your body, it's like driving your car with your eyes closed. And you can know. And so when members come through Fountain, how do they detect cancers? So we're doing full body MRI, and we also do early cancer detection screening. This is very very important.
1:01:32These are not typical tools used in the conventional care setting when it comes to prevention. This is a hard thing because currently these are not studies that insurance would yet be covering, but the goal is to collect these numbers, do the research, and work hard to democratize wellness. Yeah. So at the day you can know what's going on inside your body. It's your obligation to know.
1:01:53So check out Fountain Life. You can go to fountainlife.com/peter to get access to the latest technology to help you detect cancer at the very beginning at stage one when it is curable before it gets to stage three or stage four and you're world of hurt. All right, topic number four, the collapse of terminal value. What happens if AI makes every competitive moat temporary?
1:02:15So this is a article posted by Chamath. it's a it's a it's a powerful concept. He argues that AI could compress equity valuations by two to sevenfold of free cash flow down from today's S&P average of 22. So today the average S&P companies are getting 22 times free you know forward-looking cash flows. And he's saying we're going to get a massive reduction in that. So AI makes disruption so cheap and fast that no company can project free cash flow beyond five years terminal value.
1:02:52Very true. I mean it used to be all of the SaaS companies were projected forward and you could depend on it. so this can break down investment paradigms, break down VC investing. I'd like to jump into that, but first let me just show this is Chamath's sort of image he went along with his his post on X.
1:03:16So here we go. There's $58 trillion in the current S&P 500 and this is at the 22x of free cash flow. if we compress it down to sevenfold, it drops, we lose a 2/3 of the value of the of the S&P 500. if we end up driving it down to 2x free cash flow, it's down 90% and we get a lot of disruption of our financial markets.
1:03:46here's a chart that's showing the S&P 500 over the last 10 years, actually from 1950 through today and we're seeing it basically deviate significantly on on value you know PE ratios. So let's jump into some of the conversations. I've got the article up in front of me as well. I think I'll I'll read the opening paragraph here for us.
1:04:12and Chamath said, let's start with first principles. The entire architecture of modern capital markets rest on a single rarely examined assumption that competitive advantages compound over time, moats persist, brands endure. Network effects defend. Strip that assumption away and you aren't just repricing some stocks, you would be dismantling the philosophical foundation of how capital has been allocated over a century.
1:04:40Dave, let's go to you first on this. Absolutely correct. but the conclusion that the S&P is going to collapse is not correct. if you say look, you know, prior to the computer revolution, my ambition was to build an oil company or a a manufacturing company that would endure for 50 years building the exact same goddamn product or delivering the exact same goddamn oil for 50 years so my great-grandchildren could be as wealthy as a Rockefeller. That's dead forever and good riddance and it should be dead [clears throat] forever. If you said, well look, 22x free cash flow implies that the company will exist for 22 years making about that same amount of money.
1:05:20Well, what company like Apple has has done that? You know, is Apple selling the same products it was 22 years ago? Of course not. So the overall tailwind is 10x over just the next 10 years. There's a massive amount of tailwind coming into the economy, massive amounts of new wealth, more than we've ever seen in our lifetimes is going to come into the economy, but you got to stop looking at 22 times free cash flow from the same product over 22 years. That's nuts. You have to be looking at the management team and the ability to to roll with the innovations ala Elon. And so I think the the overall conclusion is yeah, there's going to be a massive amount of
1:05:57shuffling in the S&P. There's going to be some huge winners like you've never seen before and anyone who's doing the same thing and resting on their laurels like an insurance company or oil company doomed. Yeah, he's right. He's dead right. This analysis is basically exactly the right analysis to show how that stock's going to go down.
1:06:15Yeah. Alex? Yeah, it I mean it's certainly a provocative thesis, but I don't think it holds water. I I think it's the moral maybe the call it the earnings multiple equivalent of friend of the pod Ray Kurzweil's notion that a singularity takes the form of a firewall that you can't see past, but except in earnings multiple form that that as we start to see faster faster more accelerationist innovation that free cash flow just comes to a halt a few years later because everything is disrupted, the the everything disruption if you will.
1:06:50Here's the problem. The the free cash flow does go somewhere. Capital does get allocated somewhere. It may not be allocated to SaaS startups, post-SaaS apocalypse. Maybe it gets allocated to infrastructure. Maybe it gets allocated to lunar mining. But capital does go somewhere. It's it's not actually capital that's being compressed, quite the opposite. Capital is is explosively expanding because now we have so much more infrastructure and so many more capabilities. So I I think the the sort of the the nihilistic take that earnings multiples are compressing because a few years from now there's no moat anywhere, I I think that's relatively narrow or it
1:07:31should be construed relatively narrowly to focus on the areas that are disrupted. In this case Chamath focuses on software and SaaS type businesses, but but but but everything I expect [clears throat] is going to be disrupted. Energy becomes abundant and farmland infrastructure, these all become abundant. So in some sense I want to zoom out and and take his thesis more broadly as sort of almost bemoaning the financial consequences of abundance and it it may just be the case that a number of our existing sectors that are priced based on scarcity and moats when moats arguably are a form of scarcity or at least a way of enforcing scarcity, those go away and we live in a
1:08:15post-moat world and that will be a better world. So you going to start to value companies in a different way. In the old days it was how predictable is their cash flow? I have number of seats in this particular industry and these are many companies I can sell it to number of seats available.
1:08:32And now it sounds like from what you and Dave just said, I'm actually going to be evaluating companies on their agility, on how rapidly they can innovate, how rapidly they can get the next products out the door. Salim, love your thoughts here. yeah, well two points. One is you know, we have the CEO index where we ranked the Fortune 100 by their EXO score gauging how flexible and adaptable and purpose-driven are their org structures. And we found the top 10 of the Fortune 100 outperformed the bottom 10 by 40 times in shareholder returns over a seven-year period. So this has been going on for a long time anyway.
1:09:08Okay? I agree with Alex that capital still flows the less towards incumbents and way more towards infrastructure and adaptive platforms. It's a very important point. The only moat I think that's going to be left is a living system that learns faster than your competitors. That's that kind of inner loop that Eric Schmidt was talking about, right? Exactly.
1:09:26all the mo all the on that slider are under attack from AI. IP gets copied, switching costs, shrinking scale advantages are weakened, etc. etc. I think but the bigger point I think is that if free cash flow visibility collapses beyond say 5 years, the entire logic of the public market has to be rewritten.
1:09:44And so that's a very big thing. You're you're you have to reward renewable renewables and optionality, not scale and stability. Physical assets are going to matter more than because atoms are harder to disrupt than bits, right? Over time. But the entire SaaS business model is broken. So this is I think one way of saying this is we're going to have terminal value collapse.
1:10:05Mhm. Yeah, that's exactly what it the title is actually. So the the terminal value collapses. I think if you if you look at the S&P at 22 times free cash flow, the mid-market, the non-S&P companies are already down to about seven times free cash flow, most of them. so this is already happened outside of the S&P.
1:10:26What's propping up the S&P is mostly index funds. A huge fraction of the market is passive indexes. Mhm. And and people contributing blindly to 401k plans, which Elon Musk said clearly do not do that right now. [laughter]
1:10:38but but what will happen next is a lot of those dirt cheap mid-caps and small caps will get a huge tailwind from AI automation. You know, especially the ones that have huge payroll and labor components to them. And so that's going to drive record you it's not un- unlikely that you triple your free cash flow while your multiple comes down.
1:10:59And so there's some serious bargains out there [clears throat] just from a from a straight cash flow acceleration through AI point of view. I mean shareholder shareholder calls are going to change to this is how we're rapidly iterating our products and services. This is how we're reinventing what we do. and our future cash flow and Saleem, you've got to redo the EXO index. It's time to take a shot at that once again. so I'm the part of the paper that we're writing, the reason it's taking a little longer is that I I I hate to say but we breaks the EXO model, right? Community and crowd becomes communities and crowds of agents. So we have to rethink the model from the ground up.
1:11:38We're kind of mostly we've done that cuz then you evaluate based on those new criteria. For example, what's your intelligence that? What's your MTP architecture? What's your trust framework? There's a bunch of different elements that are new here that we have to take into account cuz the the concept of an organization where you did things inside the organization is completely gone. We're going to be doing API calls to get various things done, legal work, etc. etc. It's all going to be agentic. And then the firm, which used to be coordination costs and transaction costs with a bit of legal liability, now becomes only legal liability risk purpose container and liability
1:12:15container. Yeah. Well, also
1:12:18Amazing. And this is not super mainstream, so I'll get off the high horse quickly here but but if a private equity firm like Advent they must have heard comes in and brings either Saleem or Alex along and says, "Hey, we want to take this non-AI company with huge free cash flow and we want to retool it for the age of AI.
1:12:35Triple the cash flow and retool the business plan for AI." Alex or Saleem can tear down that company now using agents in 1/1000th the time that it would have taken a year ago or 2 years ago. And so whatever private equity firm mechanizes that is going to have no trouble retooling all of these entities cuz cuz you know exactly what the company's assets are, you know, whether it's a regulatory framework or whether it's a bunch of data. You can rip through that with Gemini or with Claude age or with OpenAI agents in light speed now. It's a transformation wave. It's fully automated. Yeah. It's a it's a transformation wave. I remember there was a one of the Star Trek episodes was
1:13:13the was it the Genesis machine that like you
1:13:17project. The Genesis project had like this wave forming agent, yeah.
1:13:20Yes. And this this this this this wave that that went over a planet and transformed everything. We're going to have the same thing. You're going to have cherry you know, you have teams cherry-picking companies and reinventing them. And disrupting. Not just not just private equity. I would argue this applies equally well to public equity.
1:13:36So something I would like I'll I'll broadcast a a request for startups if I may to the audience. Please. I would love to see activist investing disrupted by AI. I'd like AIs to write open letters to public firms telling the firms what they're doing wrong to disrupt them. If if you're if you're building an AI activist investor, write to me, please. Would would love to find a way to support you.
1:13:58That's a great idea, Alex. And what is it's a service to the CEO of the companies who need prompting or need sort of a forcing function to transform their business models. And the And if you're a board member of any of these companies, your job as a board member is to give your CEO top cover and to say, you know, you must get on the disruption you know, band here. You've got to reinvent your
1:14:24also a sort of a stealth way for AI to become a manager of the entire economy and not just picking off mom and pop businesses on the margins. All right. On to story number five, one of my favorites. Hopefully Alex, one of yours too. It's the new great space race. NASA picks SpaceX for the moon.
1:14:42Potatoes are growing in lunar dust and asteroids are carrying the code of life. All right. So here we go. I mean, listen, Boeing has been building the you know, the Artemis 2 vehicle. It's going to be launching on April 1st. it had its its readiness review on on March 12th. And if all goes well, April 1st, we're going to be going back to the moon. Not to land but to do basically an Apollo 8 style circumlunar orbit.
1:15:12very cool but you know, the new NASA administrator, an amazing individual. I'm very happy and proud to call him a friend. We're going to have him on the pod. He's agreed to do it. Just needs to get scheduled. is elevating SpaceX into the Artemis program. So Starship is going to be taking I think astronauts more safely, more economically. We'll see those numbers in a moment. But just to be clear, this is not the US story only. China has confirmed their intention to land on the moon by 2030. Let's play it back again.
1:15:45History repeating itself. 1961, we're on the moon before end of the decade. China's saying they're on the moon before the end of the decade. and so that's going to be a a beautiful competition. we'll get to the idea that you know, you can grow potatoes [laughter]
1:16:02in lunar soil. They're going back to the Martian, you know, again an incredible movie now that Project Hail Mary is out. I can't wait to go see it in IMAX theater later this week. And we just saw that on the asteroid Ryugu Ryugu, we found the five nucleo nucleo bases nucleic bases for DNA, which has four ATCG, and RNA, which has U for uracil.
1:16:30And we found all five of those bases on that asteroid. This strengthens the panspermia theory that life on Earth originated elsewhere in our galaxy in the universe and it rained on Earth and gave us the starting components for that. Let's take a look at this chart. I put on the left here NASA's SLS. And now to be clear, when I say NASA's SLS, NASA's the prime contractor and it has probably aerospace companies in every single congressional district building that vehicle. It is an expendable vehicle in in a time when everybody's going reusable. There you have SpaceX with Starship. And just for comparison of size, here's the Saturn Saturn 5 that got us to the
1:17:20moon. If you look at these two charts, these two bar charts on the left here, we're seeing well, let me just go to the center ones. We're seeing the the delivered mass to orbit. That that is it is Starship over, you know, twice as much as we're getting with with Artemis with the SLS vehicle. And if you look at to translunar injection TLI, getting out of Earth orbit to the moon, we're seeing twice as much mass going on a Starship compared to to the SLS.
1:17:59But where the rubber really hits the road is launch costs and mission costs. it's expensive to be running the SLS system. It's like the space shuttle. The space shuttle used to cost if you did four launches a year, it was a billion dollars a launch. If you did one launch a year, it was four billion dollars a launch. It wasn't the cost of the vehicle. It was the standing army of 20,000 humans that were used to to operate the space shuttle. So I I honestly don't know why the SLS has existed as long as it has. I think Starship's going to do a clean sweep of this. And of course, we've got Blue Origin as well.
1:18:36Alex, comments. Well, do you want to place bets as to how long before the United Launch Alliance, which is the prime contractor for SLS, gets acquired by Jeff Bezos or someone else? But why would you acquire it? I guess for the contracts? For the contracts, for the expertise. I'm familiar with again all the cliches in the space space about how SLS was a a make jobs program or a way to keep alive in civilian form certain capabilities that were useful for defense or other say intelligence purposes. But I think at the end of the day, we're so painfully close to finally relaunching a second space race. And I think Starship is is the obvious incumbent there, not the SLS. So, I
1:19:23hopefully we have humans on landing on the moon again in the next two to three years and we get humans eventually on Mars and all of this plays out exactly as For All Mankind has foreseen except decades late. Yeah. I don't know if you if Dave or Salim you want to play on this on this conversation. I just think, you know, we're we're building the I don't know what your best historical analogy, the you know, the covered wagons, the railroads Wagon train to the stars, Gene Roddenberry called it. It's it's all currently on on
1:20:00Starship. Starship is the only economical. Yeah. Go ahead.
1:20:05I've I've thought two or three things. One is we've gone from kind of government space theater to commercial space evolution. I think that's really powerful. For me, the really exciting thing was finding all the nuclear bases on reusable. you know, it takes life from scarcity to abundance.
1:20:21Yeah, let's let's go there. So, here's Here's the Here's the graphic if you would. Again, adenine, guanine, cytosine, thymine and uracil, the five components of DNA and RNA found on these on these vehicles. I do believe that as we get to Mars, as we get to Europa, as we get to the all of the planets and moons, that we're going to find at least microbial life you know, ubiquitous on all of these.
1:20:50Did you see Peter Garrett's prediction about microbial life on Mars? No, what did he say? Garrett's our NASA administrator, yes. Yeah, NASA administrator said that he he predicted more than 90% plus probability that NASA will eminently find evidence of microbial life in some form on Mars, which is a sea change in in terms of NASA's official position on life on Mars. It was always, well, we found water, frozen water. Now we found liquid water. We hope to find signs of life. Signs are ambiguous. Now for the first time we have a NASA administrator who's saying 90% probability we're going to find microbial life.
1:21:30And the exciting thing is how related is it going to be to microbial life on Earth. One of the theories, of course, Mars cooled first, which probably means life evolved on Mars first. And of course that we know when aster a large asteroids impacted Mars, the ejecta, the rocks that flew out, some of them reached Mars escape velocity and landed on Earth. And so we have Martian meteorites in museums today.
1:21:59and did those did those meteorites carry life with them from Mars to the Earth? Are we going to find basically even genes that are common between Martian life and life here? I mean, the real exciting thing is if we go to Europa or someplace like that and we find completely independent life forms that don't connect with life on Earth, that will be amazing.
1:22:22That's really cool. So, I got a question for you since since you guys are experts on this and I'm not. is it in the scenario where lo and behold it turns out that everything we learned in biology should have said life started on Mars or actually started farther out in the solar system and then asteroids knock chunks off and then they transport to other chunks and then life starts over again and then it ends up on Earth through that mechanism. Is that all going to be then bounded to the solar system or is it more likely in your mind that hey, wait, this this propagates through deep space?
1:22:58when I was a freshman in school, I did a paper on the interstellar medium and you can actually look at the interstellar medium and you can find the building blocks of life out in the in you know, in the medium between stars in our in our in our galaxy. These components are everywhere.
1:23:19And and the galaxy is relatively constant on the time scale of a billion years. So, I think the statistic is Mars cooled about a billion years or so plus or minus earlier than Earth. That's a a lot of So, so the galaxy first of all is not rigid. We're we're constantly We have different stars at different velocities passing by each other, close passes, all of that. So, on a time scale of a billion years, that buys an enormous amount of time for panspermia at potentially a galactic scale, not even necessarily at a an interstellar neighborhood scale. And we're we're we're several generations in as well.
1:23:55We're, you know, born from from several generations of of stars exploding and then forming new stars. There's been a lot of nebular mixing in our interstellar neighborhood. There there's one other of relevance story Please.
1:24:11folks can find it if if if they Google it. This is from a few years back attempting to extrapolate based on genetic complexity when the the last common ancestor actually would have been and finding that if you just take genetic complexity as I don't forget how exactly it's measured, but you come up with some appropriate parameterization of genetic complexity of life on Earth, extrapolate backwards, you find that the time when you get the first base pair happens approximately a billion years before life is thought to have observed on Earth. So, that's sort of an independent measure of when in principle life as we know it, DNA RNA based life could have emerged. Maybe it started on
1:24:53Mars, but we'll find out I suspect soon enough. Exciting times. Salim, you want to add something? No, I just remember my favorite thing around all this is the Drake equation. Where you where I'm I'm just going through the where you where you calculate all the factors that led to the probabilities of binary stars and life appearing. And you And the when you add it all up, you end up with 100%. It's So, the panspermia thing, but I think what we mentioned earlier, if we could find something that's non-carbon based, that would be truly exciting.
1:25:28You know, what's really cool to me is this this idea that the the dinosaurs [laughter]
1:25:33The The dinosaurs were extinguished by a meteorite and or meteor. And the propagation of the DNA or the base pairs is also via asteroids and meteors. And early in the universe history, you know, this may be popping up all the There might be life popping up constantly everywhere and propagating through all these projectiles flying around. But it always gets extinguished by another meteor, you know, just like the dinosaurs were.
1:26:04And it's not until everything cools and settles that you can have enough time to evolve human intelligence or other intelligences out there in the universe. So, it's just a big system dynamics settling problem, which is just just really cool to me to think about. I hope it turns out to be right. Yeah.
1:26:23All right. story number six, the model wars go underground. The AI frontier is fracturing into a stealth arms race where anonymity is the new moat. And here's the story. There two stories here to focus on. One, OpenAI launches GPT-5.4 mini and nano that runs twice as fast and approaches the full GPT-5.4 on coding benchmarks.
1:26:47So, these models are getting smaller and faster. And the second story, which I think is most of our conversation here, is there was a mystery model. So, a one trillion parameter model called Hunter Alpha appeared on open router with no attribution. It was secret. It a million token context window. It was free. There was no developer announced, no press release, no origin story and it processed 160 billion plus tokens.
1:27:16Everyone thought it was DeepSeek V4. Right? Cuz DeepSeek had been the main player here, but it turned out to be Xiaomi's AI team and when that was announced, their stock went up 5.8%. I remember meeting the team at Xiaomi when they came out with their first mobile [music] phone, like three young founders. They've since gone beyond just mobile phones to electric cars and now they've got a killer model.
1:27:42Points, gentlemen. I I think there are at least
1:27:46Point one is proliferation of models is very hard to contain because the the existence of the prior model gives you a complete roadmap on how to build the next model and it helps you build the next model. Mhm. So, I at this stage, I think it's a fair bet that trillion parameter models are going to propagate all over the world with anyone who has about 50 to 100 million dollars that they're willing to invest.
1:28:08and that'll come down, too, as Alex is pointing out. Many times the the algorithmic improvements are driving that down constantly. Alex
1:28:16Alex. Sorry, I cut you off.
1:28:17Yeah, maybe a couple points. So, first on the the 5.4 story, distillation continues to work and I find that completely remarkable. On the one hand
1:28:28you explain distillation for our listeners who don't know? Yeah. Sure. So, the I think the reasonable expectation for say OpenAI as well as other firms launching a big model first with lots of weights, a high parameter count, and then subsequently launching a mini or nano version, and by the way, Anthropic does the same thing and DeepMind does the same thing. They all launch smaller models later is that they're using the larger models to generate lots of data, synthetic data, and then using those synthetic data to train a a smaller model that can be faster and less expensive. So, that that's sort of a a caricatured way of describing the distillation process of in some sense
1:29:10squeezing down or compressing the larger model down to a a simpler student model. And the fact of this continues to work is I think borderline magic. The the amount of complexity that's already in the full 5.4 model that and and moreover that 5.4 has likely been the result of so-called iterated amplification and distillation over many cycles where 5.4 was likely in large part trained off of synthetic data generated by distilled models from earlier generations that we can keep playing this magic trick over and over again. It's borderline magic that that it continues to work and that we continue to be able to distill down models while retaining a large fraction
1:29:55of their capabilities. It it again makes me think that there has to be an end to the story, but hopefully it's a very satisfying end where at the end of the the distillation rainbow we get like the the distilled black hole of a model or a neutron star or something, the ultimate phase change where it's maybe like a few million parameters. It's the the end state of the game.
1:30:17A one kil- a one kilobyte file on your on your phone is all you need.
1:30:21file that's like the master equation for superintelligence after all of this distillation. It it we we showed on a previous podcast a gentleman on his iPhone using a distilled model on airplane mode being able to basically answer every question. So, imagine if on all of your devices without having to have, you know, Wi-Fi internet access, you have the distilled knowledge of humanity there to serve you. It's inside your kids teddy bear. It's in your, you know, Thomas train set.
1:30:51it's becomes magical. Here's my question for you Alex and and and Dave. You know, now that we're seeing this, we're seeing a basically a mystery trillion parameter model announced without any attribution. it used to be that the traditional moat [clears throat] for these models was their brand, their capitalization, who they were, you know, is there any defendability or are we just going to see newcomers rushing in with new models that you're going to just utilizing new model, you're going to be no longer dependent upon AI or Gemini. Thoughts on that? Well, you you've said it a million times, Peter. Data is actually the great moat, not the model itself.
1:31:42And many, many people are accumulating phenomenal data for, you know, for sur- brain surgery, for material science, for chemistry, for all of these use cases. And, you know, if you create the next great, great, great model using that proprietary data, the parameters are out there in the world, but the the data that trained it is not, and it's very hard.
1:32:02You people can use the model, but they can't compete with you by creating a a rip-off model cuz they don't have the underlying data. Now, you can generate synthetic data using the prior model. Alex is dead right about that, but I don't think that it's I don't think it's all these companies killing each other. I think it's the whole all the boats rising with the tide. I also think that if you take what Alex said a minute ago, and you know, so many college seniors ask me, "What should I do? What should I do? How many How do I you know, what Just replay 10 times what Alex just said slowly until you fully understand everything he just said. And then ask your favorite AI
1:32:36to generalize on it and find as many documents as you can around the internet to read. At the end of that process, you'll be able to build a distilled focused model that solves some problem better than anyone else on the planet. And that's that's instant business, instant value add, instant success. So, just just really In fact, the other thing you can do is take your Open Claw and have it look for every episode of this podcast where Alex said something related to what he just said. And have it also synthesize that and bring it back and feed it into your machine. I guarantee that's a good move. There's something good good spring project for anyone listening. If Sam Altman were in
1:33:13this discussion, he might point in terms of the moat question that you were asking, Peter, to well, Open AI is building up its own data centers. Although, that's no longer really true. Stargate is being pivoted to now renting servers. So, maybe less of a moat there. He might point to having the best research team in the world generating the best models.
1:33:35But, they've been hemorrhaging researchers, and those are becoming a commodity. Then he might point to being the becoming the core subscription having as as he said, a billion plus users. I'd much rather have a more than a billion users than I would have state-of-the-art model because models walk out the door every day. There's a lot of fungibility in terms of research employees.
1:33:58Only one problem. That the billion user distribution advantage may be a a little bit tenuous at the edges because you see maybe enterprises are more valuable as customers than individuals. So, maybe the billion users a little bit less valuable on margin. And then maybe also you see other labs that are able to use cheap Chinese open-weight models maybe fine-tuned legally or otherwise with clawed outputs are able to put out seemingly miraculous results. So, I I I do think we're seeing the baseline models for the moment become something of a commodity, and the value then migrates up the stack to Open Claw or other higher-level frameworks.
1:34:36This episode is brought to you by Blitzcy, autonomous software development with infinite code context. [music] Blitzcy uses thousands of specialized AI agents that think for hours to understand enterprise-scale code bases with millions of lines of code. Engineers start every development sprint with the Blitzcy platform bringing in their development requirements. [music]
1:35:00The Blitzcy platform provides a plan, then generates and precompiles code for each task. Blitzcy delivers 80% or more of the development work autonomously while providing a guide for the final 20% of human development work required to complete the sprint. Enterprises are achieving a 5x engineering velocity increase when incorporating Blitzcy as their pre-IDE development tool pairing it with their coding copilot of choice to bring an AI native SDLC into their org. Ready to 5x your engineering velocity? Visit blitzcy.com to schedule a demo and start building with Blitzcy today. [music] If we could, I'm going to move on to number seven, machines that build
1:35:44machines. AI designed a CPU in half a day, and now it wants to put data centers in orbit. So, here's the article that you know, prompted me to have this conversation about, you know, machines building machines. We're seeing recursive self-improvement happening at a faster and faster and a more fundamental rate than ever before.
1:36:06So, an AI agent called Design Conductor by Vector autonomously built a 1.5 GHz Linux-capable RISC RISC-V CPU from concept to tapeout in 12 hours compressing a quarterly engineering cycle into a lunch break. pretty extraordinary. And so, you know, here's the here's the actual numbers. Vector AI did this particular design task in 12 hours. You can see in the chart here while the traditional engineering team would have normally taken 90 days. Now, maybe this is a little bit overplayed. I'm sure it's not just 90 days. I'm sure that they were saving time along the way.
1:36:52But, what we're seeing over and over again is AI being able to do, you know, .0 to completion on its own iterating faster than any humans. Alex, this is recursive self-improvement starting to break out of the software loop. this is the innermost at least portion of maybe a rivulet of the the innermost loop where it's You see this recursive self-improvement which would otherwise be software optimizing software starting to eat through its container. It's eating down to the the EDA electronic design automation level of designing RISC-V cores. And then it I think it's going to eat further out and redesign the data centers and the energy supplies and then
1:37:37the entire economy. So, it is in one sense very satisfying to to see this happening. In another sense, maybe a person who's slightly more skeptical that that this is that this represents a broader trend would say, "Well, of course it was able to to automatically design a RISC-V core. RISC-V has all of these unit tests, so it's it's easy to define verifiable rewards, and then you can do RL and all of these other things.
1:38:03You can iterate and put react loops on it because you you have an easy way of knowing whether a given architecture a given floor plan for the chip works or not because it's such a common architecture." But, I I think that cynical perspective completely overlooks how remarkable it is that we're now at the stage of recursive self-improvement where the thing is designing its own chips.
1:38:25And not only here, it's going to be robots building robots. It's going to be everything. so, here's the here's a couple of questions, you know, so if a senior design engineer earns $400,000 and a full tapeout team costs, you know, millions over course of a year, if AI collapses it to a lunch break, what happens to the 50,000 hardware engineers currently working?
1:38:48where do they get applied? Oh, I don't think that I think that vision is flawed in a huge way in that right now because the cost of engineering a new chip is so high, we all use the same GPU and CPU for every single task even though it's nowhere near optimal for that task. What this unlocks is chip designs that are specific to the use case that are probably about a factor of 10 more efficient and maybe more. And if you think, well, we're going to spend two, three trillion dollars on these chips over the next couple of years on these data centers, if you can unlock a 10x performance improvement for a use case, that has hundreds of billions of dollars
1:39:27of implications. So, all 50,000 of those engineers are going to be useful using the AI for all the different use cases, for all the different chip designs. Also, the fab doesn't care a whit. Like, you can change the mask every day for a different design and still get the same throughput through the factory. So, it's it the the fabs don't care a whit if there are tens of thousands of different designs instead of us all using the exact same CPU for everything. So, this is just a huge unlock. I I love the way we're using trillions and trillions now on a regular basis. We're just, you know, a couple years ago it was billions and billions.
1:40:00It's more fun and you can feel it.
1:40:02you can feel the acceleration. We need a new TV series that's called Trillions, Not Billions. for sure. I I want to hit a few other stories quickly before we get to our AMA segment, and these are stories that didn't fit in the other categories. And again, please give us your feedback on the format here today. Are you enjoying it more? we'd love to know. So, in other news, here we go.
1:40:28the DOE announced about 300 million dollars for the Genesis project, inviting teams to leverage AI across 20 national challenges spanning manufacturing, biotech, and energy. Of course, the Genesis project is about the US actually, you know, using its national labs and the data contained within national labs to accelerate and expand the US in its AI and scientific pursuits.
1:40:57Alex. I think it's a generally a good thing for the US to have an industrial policy, and I think Genesis mission to the extent that for the first time, at least from the Department of Energy's vantage point, it is starting to articulate grand challenges that are in collectively part of a broader industrial policy, which the US hasn't had for decades. I think it's very important. So, fusion, obviously, one one of the grand challenges. I I think it's so important for to the extent we have a federal government that has a budget to to fund progress, to put money behind grand challenges in general. So, I'm I'm I'm I'm in the weeds from a bunch of different dimensions with the
1:41:41Genesis mission. Actually, Dario, who's the leading the relevant portions of DOE on this, I worked with him as an undergrad at MIT and as an undergraduate researcher. So, so some fond memories. [laughter]
1:41:57It's a what a tangled web we weave. What can I say? But, I'm I'm I'm generally a big fan of of what Genesis is doing.
1:42:02what concerns me here, Alex, is that this is great, right? These are like X prizes in one sense that the government's going to be running. But, and it's moving us in the direction that China has been doing for a while now. Yes.
1:42:17is deploying hundreds of billions of dollars into state-directed AI investments and saying, you know, we want fully development in in the architectures around robotics, around these AI models, and so forth. This is a relatively [clears throat] small amount of money for the US government.
1:42:34Hopefully, it's just a first toe in. It's true, but on the other hand, I I would argue China distorts its markets so much relative to to what if you compare US industrial policy distortion versus Chinese industrial policy distortion, they're they're not in the same league, and we have much deeper private capital markets that China lacks. I I like our odds on balance much more than China's. Mhm. [clears throat]
1:43:03our next article here is the rural Ohio Ohioans seek a constitutional amendment to ban data centers over 25 megawatts in the state. And, you know, this is the ultimate NIMBY, not my backyard. And it's pretty extreme. I mean, to go after an you know, a constitutional amendment. This is a genuine grassroots revolt at the end of the day.
1:43:30We we need I think, Peter, we need a new acronym. Like, what what was it?
1:43:35N- not not not in my backyard. Like, yes in my orbital plane. Yeah. Okay. Or something. [laughter] This is just going to drive all these data centers to orbit. But, this is this is crazy. I mean, these communities don't realize the amount of wealth these data centers are going to create for them. I think it's about 10 billion dollars per gigawatt of invested power.
1:43:57Or in in invested. They'll miss them when they see them in the night sky. [laughter]
1:44:04Yeah, obviously utterly insane and utterly insane to use a constitutional amendment for this purpose. I mean, to point to point out the obvious, the data centers are are tiny as a footprint on land. They're absolutely tiny. Yeah. And the wealth that they create is astronomical for the neighborhood they're in. So, there's got to be a much better way to make a win-win than to ban something that's obviously going to benefit your state tremendously. But, let put that aside.
1:44:28You're in California. Alex and I are in Massachusetts. The way we make decisions through legislatures is so messed up. Yeah.
1:44:38Like, that something like this could even get proposed is ludicrous. And that's what really needs to change cuz when you talk to the governors, they're like, I don't want this. Like, okay. Well, you we're representative democracy. There's supposed to be very, very smart people thinking about complex issues and then deciding what happens. You don't throw things like this out to a referendum of people who just got laid off.
1:44:59And it's and it's people saying, you know, my access to clean water and energy and my you know, my consumer price index of energy is going through the roof, and there's other ways to deal with this. then instead of banning it by constitutional amendment. That's for me, that's insane. All right.
1:45:17I've never seen a data center that affected the water supply. That's like it's so I hear it all the time. It's utterly ludicrous. The data center needs a fixed amount of water to cool itself. It doesn't drink the water. It just goes around in a circle. It's It's nuts. All right. another story worth mentioning is Nvidia won approval to sell its H200 chips, its most advanced chips, to Beijing.
1:45:43That's a big deal. I'm surprised.
1:45:47the Well, the realization is the ban didn't work. China both was getting access to chip through third parties, and China was developing its own competitive. And this cost Nvidia tens of billions of dollars. And since it's not working, in fact, it's stimulating a homegrown homegrown, you know, equivalent of Nvidia in China, they said let's reverse policy. The question is, is it too late? Yeah. That everything that you just said is correct except the one part where when Jensen complains he lost tens of billions of dollars, every single thing he's manufactured is sold out for years to come. So, that's true.
1:46:24that it didn't go to China, it definitely sold. Even if it was like a you know, a a dysfunctional 8080 design or whatever that did the you know, Chinese design, it still got sold. Everyone everyone in the world wants these things. So, that he didn't actually lose any money. I'm surprised though because I think I think the embargo or the ban didn't work. You're dead right. China is doing its own thing now. But, I also think that if you say, well, let's start selling them again, maybe they'll stop.
1:46:51No, that's not going to you cut them off. They're not going to forget. Like, that that isn't going to happen. So, I was really surprised that you know, that they reversed course on this. I don't know if I want to say anything further.
1:47:04Yeah, if if if I'm Beijing, I mean, on the one hand, I I read the same stories, and of course, variety of Chinese frontier AI labs are all slurping up as many H200s as they can get. And of course, it's borderline obvious that the Blackwills are are now the frontier. So, in some sense, Beijing is being kept a half step or two behind the frontier chips available to US AI labs.
1:47:27I think the story behind the story, not not to be overly speculative, but if I were the Chinese Communist Party, I'd be doing the moral equivalent of having people taste my water at this point in terms of these chips. Nvidia's been very public about how there are a variety of countermeasures that can be put in place to prevent the wrong chips from ending up in the wrong locations.
1:47:50I would and this is based on stories that I've read, stories where the Chinese government is suspicious at the at the circuit level of American chips. I I have to imagine that they're looking now at our chips with renewed scrutiny to see what else is in these chips that we're shipping to them.
1:48:09What what what algorithms are embedded deep inside. We've seen this in the opposite direction. All right. Here's a story, Alex, that you and I have enjoyed talking about. Scientists successfully froze an entire pig brain while locking in the cellular activity with minimal damage. this is cryogenics, and it's happening in a large mammal. Of course, the pig has organs, heart, liver, lung, kidney, and brain on the order of human organs.
1:48:36So, this is this is significant. So, actually, I I played a minor role in the story, and I'm not subject to confidentiality on the story, so I can tell the story. This is from This is from a company I informally advise named Nectome. and I have a another company with that where the the founder of Nectome is also involved.
1:48:59This is Eon, focusing on whole brain uploading and emulation. Nectome, which I'm not formally involved with, is focused on just the preservation side. And I I'd been nudging them like they have these amazing results, publish the results. They published the results and it it is so wonderful to see now for the first time real competition in call it the cryonics space or the preservation space since the way Nectome works isn't quite the same way as say the way 21st century medicine, which we've spoken about previously on on the pod works.
1:49:3221CM is more focused on vitrification. Nectome is more focused on a type of chemical preservation, but nonetheless
1:49:42the cryopreservant is? I mean this is about and just to describe it to the to our listenership, you're basically at or near death, you're replacing the blood supply with something that goes and fundamentally you know, infiltrates the cells and keeps the water in the cells from crystallizing and destroying the structure in the cells.
1:50:04That's right. So you're you're you're searching your latest model to find the answer.
1:50:08I'm I'm I'm double checking to see how much they've made public. -huh. So so maybe let me just talk about it at a high level. So so it's it's a chemical technique. It's it's a little bit less focused on vitrification with the whole point of vitrification on the 21CM side is is basically in ensuring that ice crystals don't form and that there isn't strong osmotic pressure reverse osmotic pressure that that cause cells to explode. On the Nectome side, it's a chemical process. I'm I'll be cautious with what I say because I haven't I need to check to see what what's in the public and and what isn't about the process.
1:50:49But the more important I think results here in addition to Nectome putting out I think their their first bio archive paper in years since the original paper that won the brain preservation Foundation award for demonstrating local preservation of the structure of of neurons is that now they've demonstrated in in full public view scaling this process up to an entire mammalian brain, a large mammalian brain, not even just a mouse brain. So I I think we're we're finding ourselves in in a near future slash present where finally we have enough data to to be confident that entire mammalian brains are being preserved and this immediately raises the question, which is the
1:51:34question I ask almost everyone, why where are all of the cryonics patients? Why don't we have billions of people now that we have a growing body of evidence that brain structure can be preserved by whichever technique, whether it's Nectome on the one side, 21CM on the other, why don't we have a billion people signing up for cryonics? And I would again to the audience, sign up for cryonics. Like just do it.
1:52:00way this is the way you get to see the 23rd century. Yes. It's a I got I heard from from the head of Alcor after my last call to action to to do cryonics. Apparently lots of people flooded into Alcor. It's a nonprofit. I make no money off of saying this, no financial interest. Just get yourself a cryonics plan as part of a portfolio for longevity. Love it. Love it. I want to take a second and just say thank you to Nick Singh and and Dana Khan our producers for supporting us on this new format. I enjoyed it. Did you guys enjoy it?
1:52:37I love it. It just feels organized. Yeah, well it feels organized and fun. It actually feels fun to think through the topics with you guys. It's like an entire episode worth of AMA. Yeah, with ourselves. Yeah, it actually changes the stories we cover too, you know, cuz we we normally go through the most important stories to change your life.
1:52:56But here when you put it into themes, you actually dig up other stories that are related to the primary topic that you otherwise would have missed. So I love that. All right, here we go. Let's pick one each from page one and one from page two. Alex, would you go first? Oh, so many good options.
1:53:15Yes. We'll start with number two. Where should entrepreneurs actually run their AI compute? Local hardware, AWS cloud, or iPhone? And that comes from Frank Gerard Marketing. There isn't a good answer. There are at least no single good answer, lots of decent answers. There are benefits to each. So with local hardware, you have greater control, greater confidentiality and data privacy. You're going to on the other hand end up maintaining said local hardware. You have to worry about your own backups. Can be a pain in the neck from variety of different perspectives.
1:53:53With AWS or one of the many other public clouds, you don't have to worry about that. That's abstracted away. On the other hand, you might have to compete ferociously for access to say GPU resources. You're competing with other tenants for common resources. You might have to worry on margin depending on how familiar you or your organization are with with opsec and and cybersecurity.
1:54:18You might have greater surface for attack. On the other hand, you have more more scalability. With the iPhone, you have it it's sort of the ultimate edge device until all of us and not just some of us are running foundation models on our watches and our smart glasses, which is already happening and is going to be more evenly distributed. You have even greater privacy. So I I don't think this is maybe this goes without saying. I I don't think this should be viewed as a black or white or binary trade-off.
1:54:49There is a spectrum from edge edge compute to data centers at the core. I I think the best answer actually is I want to run my AI compute in the Dyson swarm. And that Dyson swarm will be perfect blend when fully realized in a few years of of data centers. We'll have lots of maybe if if Elon's statistics are to be believed, 100 kilowatt nodes filling the sky.
1:55:16But also it'll be incredibly elastic. If if we're disassembling the moon to fire off new 100 kilowatt nodes in in the the stellar or Dyson swarm fabric, [clears throat] it perfect blend. Okay Dave, which one you going to choose?
1:55:31one thing on this topic only because I spent the whole weekend dorking around on Amazon AWS Bedrock, which is a great choice by the way, even though if my bed was made of rock, it would feel like getting started on Amazon Bedrock. I mean that's that's it's it's a brutal get up and running process on Bedrock. But it does the critical thing that you need, which is it captures all of your prompt history for you and any teammates that you have into easy to manage S3 buckets. So your AI can analyze everything that you've done, which is a critically important function. So that may be available elsewhere, too, but it's it's probably as good a choice
1:56:12as any. But whatever you do, don't just start running on some random hardware and then lose all the prompt history. So this is just an easy way to capture it. Pick a number. I'll take number three. Are humanoid robots over-engineered? Would it be more efficient to isolate basic needs like food, water, and clothes and automate those directly instead from Go Unite FB3GN.
1:56:33Short answer, yes, absolutely. So why are we putting all this energy into humanoid robots? The reason we're putting all the energy into humanoid robots is because AI kind of came into the world almost overnight and we're in a race to capital right now. And so what's critical for all these projects and startups is getting funded.
1:56:54The humanoids are so much more visually appealing that they're easier to fund. They're also easier to recruit into. And that'll unlock the supply chain of all the parts and that'll unlock all of the other robots that you know, that farm and create clothes and whatever, which will probably not look all that humanoid. But you know, when you look at the gigafactory like Peter and I did, vast majority of the automation there is not humanoid robots. It's machines that look like you know, machines doing their jobs. Yep.
1:57:21And then the humanoids just operate those machines. So I think they are over-engineered and over-invested relative to where we'll end up, but for a very good reason you should think about like visual appeal and capital raising are a core core part of this step function we're living in right now.
1:57:35So I'm going to go with number four. How can AI be used to end a war, not as a weapon, but as an impartial negotiator that all parties could trust? This is from at JN kind five. So I find that absolutely fascinating and I do think it's a powerful tool. If you haven't used a large language model for negotiations, one of the things is we don't know how to think other than the way that we we know how to think. So being able to put yourself in the mindset of another individual is extraordinarily powerful.
1:58:11If you haven't said listen, I'm you know, I'm anti-guns, my my neighbor, my friend, my spouse loves guns, you know, could you please help me explain to them my feelings in a way that lands with them and isn't viewed as offensive, you can get some you know, extreme you know, support on your negotiation skills. And at the end of the day, I think this is one of the most exciting unexplored applications for AI because the system can ingest also every peace treaty, every negotiation script, every conflict resolution framework that's ever been had and can you know, model outcomes with no tribal allegiance. One of the biggest challenges we have as humans is we have these cognitive biases
1:58:59and these tribal biases that are driving us. So, you know, can we use this for negotiation? Absolutely. and I think both sides, if you set as an objective function that you want to reach a balanced solution that both sides have and both sides are using AI, I mean, it could be different models, I think the probability of getting to a solution is much, much higher.
1:59:27we are biased when we're dealing with humans. and one of the things that goes on when you're talking, for example, to an AI model for, you know, psychological therapy, when you realize you're, you know, you're telling your innermost thoughts to a human, you feel like you're going to be judged, but you don't feel judged when you're talking to an AI model. And so, I think there's real value to be had here.
1:59:52I don't know if Salim's back online or not, but
1:59:55He's not, but if if if if if I might add just one thing to to your point, Peter. Something that I'm seeing more and more in the past few months, not for for war, but for commercial negotiation, I'm seeing this all the time. Two parties that are at loggerheads in a commercial negotiation, one of the parties will bring in a frontier model and ask the frontier model what the commercially reasonable outcome is, bring it to the other side, the other side will will consult their model, and they'll come to rapid agreement. Yeah.
2:00:24I'm seeing this happen now over and over again. All right, here's the next eight questions from our AMA. Alex, over to you. Sure. again, there are so many fun questions here. rather than choose eight, which would require me to give implicit investment advice, number seven, I'll avoid that one. Number six, less slightly less interesting. I'll tackle number five since since I've been beating the drum a bit for for solve everything including disease.
2:00:54So, the question is, once AI solves most diseases, how soon will treatments be available to everyone? Will access lag? And this is from Katisse 896. So, maybe the sub question first is, when do I think AI has a decent chance of solving most diseases? My timeline, and this is not specific to me, I I think if you ask the the more optimistic elements of CZI, the Chan Zuckerberg Initiative, Biohub, maybe Arc Institute, and some other organizations, maybe Anthropic on a good day, I think they'll say something like five years from now.
2:01:31So, five years is pretty rapid time scale. It's more rapid in in many cases than what historically has been the the clinical trial process end to end through three phases. So, second sub question here is, how soon will treatments be available to everyone? If if say tomorrow one of the frontier labs says, all right, here here are the cures to the top 5,000 unsolved or untreatable diseases, we have vast computational experiments demonstrating to to the satisfaction of all experts that these are the cures or at least the treatments for these diseases, how soon would those treatments be broadly available?
2:02:14Under the present regime, which by the way is not the same as the regime even one year ago, there would still be probably a multi-year process. The FDA has recently announced two major developments that are, I think, relevant here. One, the FDA under this administration has decided to adopt a Bayesian perspective as opposed to a frequentist perspective, meaning that they're willing to incorporate for the first time in history evidence in terms of clinical approvals from outside a particular drug. That's that's a huge sea change. It means that in in principle, drug approvals can operate much more quickly because they can take into account lots of pre-existing
2:02:53information that predated the particular drug. Second big development, a move from, and this was again relatively recently announced by this FDA, from a two clinical trial process to a one clinical trial process for certain cases, expediting the the approval process. So, I I think, fast forwarding to as long-winded answer to how soon will treatments be available to everyone, I think if tomorrow or five years from now a frontier lab or multiple frontier labs said, here are the very well-motivated top 5,000 cures to everything, I think we would see similar developments from the FDA to go to a zero clinical trial model given enough Bayesian evidence and
2:03:34given enough computational evidence, which is to say a zero trial model. I I think there would be so much political pressure that we would probably, barring some exceptional circumstance, see relatively fast availability.
2:03:47All right, so Dave, let's go to you. Okay. well, I'll take number eight since Alex couldn't touch it and we've lost Salim. if you had to choose one public company to bet on in the age of AI, which one and why? From Matthew Johnson 6525. so, we can't give investment advice, obviously, but I will tell you, I've said a bunch of times on the pod, go to 13f.info.
2:04:11Look up the Situational Awareness Fund, which is Leopold Aschenbrenner, and every quarter he has to file his his holdings. He's killing it, and the reason he's killing it is cuz he listens to exactly what Alex is always saying, look for the innermost loop. Find you the tailwinds in equities and assets are like nothing you've ever seen.
2:04:30But you have to be in the AI loop to be relevant. And so, you'll see all Leopold's holdings are somehow in the centerpiece of the innermost loop. And so, those are the things you want to own. So, those include things that are chip fabs, things that have power, things that are related to chip design, things that are algorithmic that are directly deriving use cases. Those are all in that fund. So, that's your road map. So, look at his holdings and then generalize from there, and you'll find lots of great stuff that you should be you should be buying. Also, you know, W-2 income is going to get pummeled in this next three years, but assets, you know, holdings and ownership and things
2:05:06is going to go through the roof. So, so buy stuff, you know, whether it's equities, public public or private, real estate, you know, things that'll appreciate. That's what you need in this next three-year window. Amazing. I am under time pressure myself. I've got to jump on a film recording. I'm going to go to our outro music here, which is brought to us by CJ Trueheart. it was a piece he developed for the Abundance Summit called Moonshot Minds. take a listen.
2:05:39Love the lyrics. here we are. Thank you to CJ for Moonshot Minds. [music] [music] All right, gentlemen.
2:07:48gets better and better all the time.
2:07:50all the time. AWG, DB2, this was fun. Wishing you guys an extraordinary day. Salim, who's airborn to Brazil, safe travels, buddy. soon we'll be putting you on rocket rides to get you there. Anyway, thank you everybody for listening. please tell
2:08:10hyperloop. Yeah, yeah, I guess we can do hyperloop under the under the Gulf. Anyway, long story short, thank you for listening. please join us. If you've not subscribed, please do. We're putting out these podcasts on cadence that you want to get alerts, so our mission, help you see an abundant world. The world is getting better at an extraordinary rate. The technologies to solve the world's biggest problems. If you're an entrepreneur, thank you for being an entrepreneur. Entrepreneurs are individuals who find juicy problems and solve problems. The more entrepreneurs on the planet, the better Earth and humanity is.
2:08:46Gentlemen, till I see you next time. Be well. Peter Diamandis, your host, signing off. See you guys. Take care.
2:08:53Have a good movie, Peter. Thanks, pal. If you made it to the end of this episode, which you obviously did, I consider you a moonshot mate. Every week my moonshot mates and I spend a lot of energy and time to really deliver you the news that [music] matters. If you're a subscriber, thank you. If you're not a subscriber yet, please consider subscribing so you get the news as it comes out. I also want to invite you to join me on my weekly newsletter called Metatrends. I have a research team. You may not know this, but we spend the entire week looking at the Metatrends that are impacting your family, your company, your industry, [music] your nation. And I put this into a
2:09:28two-minute read every week. If you'd like to get access to the Metatrends newsletter every week, go to dmandis.com/metatrends. That's dmandis.com/metatrends. [music] Thank you again for joining us today. It's a blast [music] for us to put this together every week.
资源与出处
- 视频在 YouTube 收看
- 档案silicon-valley.tech 档案页本页视频与文稿数据来自 Elon Musk Archive
- 精读本篇精读 Markdown 版